1;;; -*-  Mode: Lisp; Package: Maxima; Syntax: Common-Lisp; Base: 10 -*- ;;;;
2;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3;;;     The data in this file contains enhancments.                    ;;;;;
4;;;                                                                    ;;;;;
5;;;  Copyright (c) 1984,1987 by William Schelter,University of Texas   ;;;;;
6;;;     All rights reserved                                            ;;;;;
7;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
8;;;     (c) Copyright 1982 Massachusetts Institute of Technology         ;;;
9;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
10
11(in-package :maxima)
12
13(macsyma-module simp)
14
15(declare-top (special rulesw *inv* substp limitp
16		      prods negprods sums negsums
17		      $scalarmatrixp *nounl*
18		      $keepfloat $ratprint
19		      $demoivre $float
20		      bigfloatzero bigfloatone $assumescalar
21		      opers-list *opers-list $dontfactor *n
22		      *out *in varlist genvar $factorflag radcanp
23                      *builtin-numeric-constants*))
24
25;; General purpose simplification and conversion switches.
26
27(defmvar $negdistrib t
28  "Causes negations to be distributed over sums, e.g. -(A+B) is
29	 simplified to -A-B.")
30
31(defmvar $numer nil
32  "Causes SOME mathematical functions (including exponentiation)
33	 with numerical arguments to be evaluated in floating point.
34	 It causes variables in an expression which have been given
35	 NUMERVALs to be replaced by their values.  It also turns
36	 on the FLOAT switch."
37  see-also ($numerval $float))
38
39(defmvar $simp t "Enables simplification.")
40
41(defmvar $sumexpand nil
42  "If TRUE, products of sums and exponentiated sums go into nested
43	 sums.")
44
45(defmvar $numer_pbranch nil)
46
47;; Switches dealing with matrices and non-commutative multiplication.
48
49(defmvar $doscmxplus nil
50  "Causes SCALAR + MATRIX to return a matrix answer.  This switch
51	 is not subsumed under DOALLMXOPS.")
52
53(defmvar $domxexpt t
54  "Causes SCALAR^MATRIX([1,2],[3,4]) to return
55	 MATRIX([SCALAR,SCALAR^2],[SCALAR^3,SCALAR^4]).  In general, this
56         transformation affects exponentiations where the *print-base* is a
57         scalar and the power is a matrix or list.")
58
59(defmvar $domxplus nil)
60
61(defmvar $domxtimes nil)
62
63(defmvar $mx0simp t)
64
65;; Switches dealing with expansion.
66
67(defmvar $expop 0
68  "The largest positive exponent which will be automatically
69	 expanded.  (X+1)^3 will be automatically expanded if
70	 EXPOP is greater than or equal to 3."
71  fixnum
72  see-also ($expon $maxposex $expand))
73
74(defmvar $expon 0
75  "The largest negative exponent which will be automatically
76	 expanded.  (X+1)^(-3) will be automatically expanded if
77	 EXPON is greater than or equal to 3."
78  fixnum
79  see-also ($expop $maxnegex $expand))
80
81(defmvar $maxposex 1000.
82  "The largest positive exponent which will be expanded by
83	 the EXPAND command."
84  fixnum
85  see-also ($maxnegex $expop $expand))
86
87;; Check assignment to be a positive integer
88(putprop '$maxposex 'posintegerset 'assign)
89
90(defmvar $maxnegex 1000.
91  "The largest negative exponent which will be expanded by
92	 the EXPAND command."
93  fixnum
94  see-also ($maxposex $expon $expand))
95
96;; Check assignment to be a positive integer
97(putprop '$maxnegex 'posintegerset 'assign)
98
99;; Lisp level variables
100
101(defmvar dosimp nil
102  "Causes SIMP flags to be ignored.  $EXPAND works by binding
103	 $EXPOP to $MAXPOSEX, $EXPON to $MAXNEGEX, and DOSIMP to T.")
104
105(defmvar errorsw nil
106  "Causes a throw to the tag ERRORSW when certain errors occur
107	 rather than the printing of a message.  Kludgy MAXIMA-SUBSTITUTE for
108	 MAXIMA-ERROR signalling.")
109
110(defmvar derivsimp t "Hack in `simpderiv' for `rwg'")
111
112(defmvar $rootsepsilon #+gcl (float 1/10000000) #-gcl 1e-7)
113(defmvar $grindswitch nil)
114(defmvar $algepsilon 100000000)
115(defmvar $true t)
116(defmvar $false nil)
117(defmvar $on t)
118(defmvar $off nil)
119(defmvar $logabs nil)
120(defmvar $limitdomain '$complex)
121(defmvar $listarith t)
122(defmvar $domain '$real)
123(defmvar $m1pbranch nil)
124(defmvar $%e_to_numlog nil)
125(defmvar $%emode t)
126(defmvar $lognegint nil)
127(defmvar $ratsimpexpons nil)
128(defmvar $logexpand t) ; Possible values are T, $ALL and $SUPER
129(defmvar $radexpand t)
130(defmvar $subnumsimp nil)
131(defmvar $logsimp t)
132(defmvar $distribute_over t) ; If T, functions are distributed over bags.
133
134(defvar rischp nil)
135(defvar rp-polylogp nil)
136(defvar wflag nil)
137(defvar expandp nil)
138(defvar timesinp nil)
139(defvar %e-val (mget '$%e '$numer))
140(defvar %pi-val (mget '$%pi '$numer))
141(defvar derivflag nil)
142(defvar exptrlsw nil)
143(defvar expandflag nil)
144(defvar *zexptsimp? nil)
145(defvar *const* 0)
146
147(defprop mnctimes t associative)
148(defprop lambda t lisp-no-simp)
149
150;; Local functions should not be simplified. Various lisps
151;; use various names for the list structure defining these:
152(eval-when
153    #+gcl (load)
154    #-gcl (:load-toplevel)
155  (eval '(let* ((x 1)
156		(z #'(lambda () 3)))
157	  (dolist (y (list x z))
158	    (and (consp y)
159		 (symbolp (car y))
160		 (setf (get (car y) 'lisp-no-simp) t))))))
161
162(dolist (x '(mplus mtimes mnctimes mexpt mncexpt %sum))
163  (setf (get x 'msimpind) (cons x '(simp))))
164
165;; operators properties
166
167(mapc #'(lambda (x) (setf (get (first x) 'operators) (second x)))
168      '((mplus simplus) (mtimes simptimes) (mncexpt simpncexpt)
169	(mminus simpmin) (%gamma simpgamma) (mfactorial simpfact)
170	(mnctimes simpnct) (mquotient simpquot) (mexpt simpexpt)
171	(%log simpln)
172        (%derivative simpderiv)
173        (%signum simpsignum)
174	(%integrate simpinteg) (%limit simp-limit)
175	(bigfloat simpbigfloat) (lambda simplambda) (mdefine simpmdef)
176	(mqapply simpmqapply) (%gamma simpgamma)
177	($beta simpbeta) (%sum simpsum) (%binomial simpbinocoef)
178	(%plog simpplog) (%product simpprod) (%genfact simpgfact)
179	($atan2 simpatan2) ($matrix simpmatrix) (%matrix simpmatrix)
180	($bern simpbern) ($euler simpeuler)))
181
182(defprop $li lisimp specsimp)
183(defprop $psi psisimp specsimp)
184
185(defprop $equal t binary)
186(defprop $notequal t binary)
187
188(defmfun $bfloatp (x)
189  (and (consp x)
190       (consp (car x))
191       (eq (caar x) 'bigfloat)))
192
193(defun zerop1 (x)
194  (or (and (integerp x) (= 0 x))
195      (and (floatp x) (= 0.0 x))
196      (and ($bfloatp x) (= 0 (second x)))))
197
198(defun onep1 (x)
199  (or (and (integerp x) (= 1 x))
200      (and (floatp x) (= 1.0 x))
201      (and ($bfloatp x) (zerop1 (sub x 1)))))
202
203(defun mnump (x)
204  (or (numberp x)
205      (and (not (atom x)) (not (atom (car x)))
206	   (member (caar x) '(rat bigfloat)))))
207
208;; Does X or a subexpression match PREDICATE?
209;;
210;; If X is a tree then we recurse depth-first down its arguments. The specrep
211;; check is because rat forms are built rather differently from normal Maxima
212;; expressions so we need to unpack them for the recursion to work properly.
213(defun subexpression-matches-p (predicate x)
214  (or (funcall predicate x)
215      (and (consp x)
216           (if (specrepp x)
217               (subexpression-matches-p predicate (specdisrep x))
218               (some (lambda (arg) (subexpression-matches-p predicate arg))
219                     (cdr x))))))
220
221;; Is there a bfloat anywhere in X?
222(defun some-bfloatp (x) (subexpression-matches-p '$bfloatp x))
223
224;; Is there a float anywhere in X?
225(defun some-floatp (x) (subexpression-matches-p 'floatp x))
226
227;; EVEN works for any arbitrary lisp object since it does an integer
228;; check first.  In other cases, you may want the Lisp EVENP function
229;; which only works for integers.
230
231(defun even (a) (and (integerp a) (not (oddp a))))
232
233(defun ratnump (x) (and (not (atom x)) (eq (caar x) 'rat)))
234
235(defun mplusp (x) (and (not (atom x)) (eq (caar x) 'mplus)))
236
237(defun mtimesp (x) (and (not (atom x)) (eq (caar x) 'mtimes)))
238
239(defun mexptp (x) (and (not (atom x)) (eq (caar x) 'mexpt)))
240
241(defun mnctimesp (x) (and (not (atom x)) (eq (caar x) 'mnctimes)))
242
243(defun mncexptp (x) (and (not (atom x)) (eq (caar x) 'mncexpt)))
244
245(defun mlogp (x) (and (not (atom x)) (eq (caar x) '%log)))
246
247(defun mmminusp (x) (and (not (atom x)) (eq (caar x) 'mminus)))
248
249(defun mnegp (x)
250  (cond ((realp x) (minusp x))
251        ((or (ratnump x) ($bfloatp x)) (minusp (cadr x)))))
252
253(defun mqapplyp (e) (and (not (atom e)) (eq (caar e) 'mqapply)))
254
255(defun ratdisrep (e) (simplifya ($ratdisrep e) nil))
256
257(defun sratsimp (e) (simplifya ($ratsimp e) nil))
258
259(defun simpcheck (e flag)
260  (cond ((specrepp e) (specdisrep e))
261        (flag e)
262        (t (let (($%enumer $numer))
263             ;; Switch $%enumer on, when $numer is TRUE to allow
264             ;; simplification of $%e to its numerical value.
265             (simplifya e nil)))))
266
267(defun mratcheck (e) (if ($ratp e) (ratdisrep e) e))
268
269(defmfun $numberp (e) (or ($ratnump e) ($floatnump e) ($bfloatp e)))
270
271(defmfun $integerp (x)
272  (or (integerp x)
273      (and ($ratp x)
274	   (not (member 'trunc (car x)))
275	   (integerp (cadr x))
276	   (equal (cddr x) 1))))
277
278;; The call to $INTEGERP in the following two functions checks for a CRE
279;; rational number with an integral numerator and a unity denominator.
280
281(defmfun $oddp (x)
282  (cond ((integerp x) (oddp x))
283	(($integerp x) (oddp (cadr x)))))
284
285(defmfun $evenp (x)
286  (cond ((integerp x) (evenp x))
287	(($integerp x) (not (oddp (cadr x))))))
288
289(defmfun $floatnump (x)
290  (or (floatp x)
291      (and ($ratp x) (floatp (cadr x)) (onep1 (cddr x)))))
292
293(defmfun $ratnump (x)
294  (or (integerp x)
295      (ratnump x)
296      (and ($ratp x)
297	   (not (member 'trunc (car x)))
298	   (integerp (cadr x))
299	   (integerp (cddr x)))))
300
301(defmfun $ratp (x)
302  (and (not (atom x))
303       (consp (car x))
304       (eq (caar x) 'mrat)))
305
306(defmfun $taylorp (x)
307  (and (not (atom x))
308       (eq (caar x) 'mrat)
309       (member 'trunc (cdar x) :test #'eq) t))
310
311(defun specrepcheck (e) (if (specrepp e) (specdisrep e) e))
312
313;; Note that the following two functions are carefully coupled.
314
315(defun specrepp (e)
316  (and (not (atom e))
317       (member (caar e) '(mrat mpois) :test #'eq)))
318
319(defun specdisrep (e)
320  (cond ((eq (caar e) 'mrat) (ratdisrep e))
321	(t ($outofpois e))))
322
323(defmfun $polysign (x)
324  (setq x (cadr (ratf x)))
325  (cond ((equal x 0) 0) ((pminusp x) -1) (t 1)))
326
327;; These check for the correct number of operands within Macsyma expressions,
328;; not arguments in a procedure call as the name may imply.
329
330(defun arg-count-check (required-arg-count expr)
331  (unless (= required-arg-count (length (rest expr)))
332    (wna-err expr required-arg-count)))
333
334(defun oneargcheck (expr)
335  (arg-count-check 1 expr))
336
337(defun twoargcheck (expr)
338  (arg-count-check 2 expr))
339
340;; WNA-ERR: Wrong Number of Arguments error
341;;
342;; If REQUIRED-ARG-COUNT is non-NIL, then we check that EXPR has the
343;; correct number of arguments. A informative error message is shown
344;; if the number of arguments is not given.
345;;
346;; Otherwise, EXPR must be a symbol and a generic message is printed.
347;; (This is for backward compatibility for existing uses of WNA-ERR.)
348(defun wna-err (exprs &optional required-arg-count)
349  (if required-arg-count
350      (let ((op (caar exprs))
351	    (actual-count (length (rest exprs))))
352	(merror (intl:gettext "~M: expected exactly ~M arguments but got ~M: ~M")
353		op required-arg-count actual-count (list* '(mlist) (rest exprs))))
354      (merror (intl:gettext "~:@M: wrong number of arguments.")
355	      exprs)))
356
357(defun improper-arg-err (exp fn)
358  (merror (intl:gettext "~:M: improper argument: ~M") fn exp))
359
360(defun subargcheck (form subsharp argsharp fun)
361  (if (or (not (= (length (subfunsubs form)) subsharp))
362	  (not (= (length (subfunargs form)) argsharp)))
363      (merror (intl:gettext "~:@M: wrong number of arguments or subscripts.") fun)))
364
365;; Constructor and extractor primitives for subscripted functions, e.g.
366;; F[1,2](X,Y).  SUBL is (1 2) and ARGL is (X Y).
367
368;; These will be flushed when NOPERS is finished.  They will be macros in
369;; NOPERS instead of functions, so we have to be careful that they aren't
370;; mapped or applied anyplace.  What we really want is open-codable routines.
371
372(defun subfunmakes (fun subl argl)
373  `((mqapply simp) ((,fun simp array) . ,subl) . ,argl))
374
375(defun subfunmake (fun subl argl)
376  `((mqapply) ((,fun simp array) . ,subl) . ,argl))
377
378(defun subfunname (exp) (caaadr exp))
379
380(defun subfunsubs (exp) (cdadr exp))
381
382(defun subfunargs (exp) (cddr exp))
383
384(defmfun $numfactor (x)
385  (setq x (specrepcheck x))
386  (cond ((mnump x) x)
387	((atom x) 1)
388	((not (eq (caar x) 'mtimes)) 1)
389	((mnump (cadr x)) (cadr x))
390	(t 1)))
391
392(defun scalar-or-constant-p (x flag)
393  (if flag (not ($nonscalarp x)) ($scalarp x)))
394
395(defmfun $constantp (x)
396  (cond ((atom x) (or ($numberp x) (kindp x '$constant)))
397	((member (caar x) '(rat bigfloat) :test #'eq) t)
398	((specrepp x) ($constantp (specdisrep x)))
399	((or (mopp (caar x)) (kindp (caar x) '$constant))
400	 (do ((x (cdr x) (cdr x))) ((null x) t)
401	   (if (not ($constantp (car x))) (return nil))))))
402
403(defun constant (x)
404  (cond ((symbolp x) (kindp x '$constant))
405	(($subvarp x)
406	 (and (kindp (caar x) '$constant)
407	      (do ((x (cdr x) (cdr x))) ((null x) t)
408		(if (not ($constantp (car x))) (return nil)))))))
409
410(defun maxima-constantp (x)
411  (or (numberp x)
412      (and (symbolp x) (kindp x '$constant))))
413
414(defun consttermp (x) (and ($constantp x) (not ($nonscalarp x))))
415
416(defmfun $scalarp (x) (or (consttermp x) (eq (scalarclass x) '$scalar)))
417
418(defmfun $nonscalarp (x) (eq (scalarclass x) '$nonscalar))
419
420(defun scalarclass (exp) ;  Returns $SCALAR, $NONSCALAR, or NIL (unknown).
421  (cond ((mnump exp)
422         ;; Maxima numbers are scalar.
423         '$scalar)
424        ((atom exp)
425	 (cond ((or (mget exp '$nonscalar)
426	            (and (not (mget exp '$scalar))
427	                 ;; Arrays are nonscalar, but not if declared scalar.
428	                 (or (arrayp exp)
429	                     ($member exp $arrays))))
430	        '$nonscalar)
431	       ((or (mget exp '$scalar)
432	            ;; Include constant atoms which are not declared nonscalar.
433	            ($constantp exp))
434	        '$scalar)))
435        ((member 'array (car exp))
436         (cond ((mget (caar exp) '$scalar) '$scalar)
437               ((mget (caar exp) '$nonscalar) '$nonscalar)
438               (t nil)))
439	((specrepp exp) (scalarclass (specdisrep exp)))
440	;; If the function is declared scalar or nonscalar, then return. If it
441        ;; isn't explicitly declared, then try to be intelligent by looking at
442        ;; the arguments to the function.
443	((scalarclass (caar exp)))
444	;; <number> + <scalar> is SCALARP because that seems to be useful.
445        ;; This should probably only be true if <number> is a member of the
446        ;; field of scalars. <number> * <scalar> is SCALARP since
447        ;; <scalar> + <scalar> is SCALARP. Also, this has to be done to make
448        ;; <scalar> - <scalar> SCALARP.
449	((member (caar exp) '(mplus mtimes) :test #'eq)
450	 (do ((l (cdr exp) (cdr l))) ((null l) '$scalar)
451	   (if (not (consttermp (car l)))
452	       (return (scalarclass-list l)))))
453	((and (eq (caar exp) 'mqapply) (scalarclass (cadr exp))))
454	((mxorlistp exp) '$nonscalar)
455	;; If we can't find out anything about the operator, then look at the
456        ;; arguments to the operator.  I think NIL should be returned at this
457        ;; point.  -cwh
458	(t
459	 (do ((exp (cdr exp) (cdr exp)) (l '(1)))
460	      ((null exp) (scalarclass-list l))
461	    (if (not (consttermp (car exp)))
462	        (setq l (cons (car exp) l)))))))
463
464;;  Could also do <scalar> +|-|*|/ |^ <declared constant>, but this is not
465;;  always correct and could screw somebody.
466
467;;  SCALARCLASS-LIST takes a list of expressions as its argument.  If their
468;;  scalarclasses all agree, then that scalarclass is returned.
469
470(defun scalarclass-list (llist)
471  (cond ((null llist) nil)
472	((null (cdr llist)) (scalarclass (car llist)))
473	(t (let ((sc-car (scalarclass (car llist)))
474		 (sc-cdr (scalarclass-list (cdr llist))))
475	     (cond ((or (eq sc-car '$nonscalar)
476			(eq sc-cdr '$nonscalar))
477		    '$nonscalar)
478		   ((and (eq sc-car '$scalar) (eq sc-cdr '$scalar))
479		    '$scalar))))))
480
481(defun mbagp (x)
482  (and (not (atom x))
483       (member (caar x) '(mequal mlist $matrix) :test #'eq)))
484
485(defun mequalp (x) (and (not (atom x)) (eq (caar x) 'mequal)))
486
487(defun mxorlistp (x)
488  (and (not (atom x))
489       (member (caar x) '(mlist $matrix) :test #'eq)))
490
491(defun mxorlistp1 (x)
492  (and (not (atom x))
493       (or (eq (caar x) '$matrix)
494	   (and (eq (caar x) 'mlist) $listarith))))
495
496(defun constfun (ign)
497  (declare (ignore ign)) ; Arg ignored.  Function used for mapping down lists.
498  *const*)
499
500(defun constmx (*const* x)
501  (simplifya (fmapl1 'constfun x) t))
502
503;;; ISINOP returns the complete subexpression with the operator OP, when the
504;;; operator OP is found in EXPR.
505
506(defun isinop (expr op)    ; OP is assumed to be an atom
507  (cond ((atom expr) nil)
508        ((and (eq (caar expr) op)
509              (not (member 'array (cdar expr) :test #'eq)))
510         expr)
511        (t
512         (do ((expr (cdr expr) (cdr expr))
513              (res nil))
514             ((null expr))
515           (when (setq res (isinop (car expr) op))
516             (return res))))))
517
518(defun free (exp var)
519  (cond ((alike1 exp var) nil)
520	((atom exp) t)
521	(t
522	 (and (listp (car exp))
523	      (free (caar exp) var)
524	      (freel (cdr exp) var)))))
525
526(defun freel (l var)
527  (do ((l l (cdr l))) ((null l) t)
528    (cond
529     ((atom l) (return (free l var)))	;; second element of a pair
530     ((not (free (car l) var)) (return nil)))))
531
532
533(defun freeargs (exp var)
534  (cond ((alike1 exp var) nil)
535	((atom exp) t)
536	(t (do ((l (margs exp) (cdr l))) ((null l) t)
537	     (cond ((not (freeargs (car l) var)) (return nil)))))))
538
539(defun simplifya (x y)
540  (cond ((not $simp) x)
541        ((atom x)
542         (cond ((and $%enumer $numer (eq x '$%e))
543                ;; Replace $%e with its numerical value,
544                ;; when %enumer and $numer TRUE
545                (setq x %e-val))
546               (t x)))
547	((atom (car x))
548	 (cond ((and (cdr x) (atom (cdr x)))
549		(merror (intl:gettext "simplifya: malformed expression (atomic cdr).")))
550	       ((get (car x) 'lisp-no-simp)
551		;; this feature is to be used with care. it is meant to be
552		;; used to implement data objects with minimum of consing.
553		;; forms must not bash the DISPLA package. Only new forms
554		;; with carefully chosen names should use this feature.
555		x)
556	       (t (cons (car x)
557			(mapcar #'(lambda (x) (simplifya x y)) (cdr x))))))
558	((eq (caar x) 'rat) (*red1 x))
559	;; Enforced resimplification: Reset dosimp and strip 'simp tags from x.
560	(dosimp (let ((dosimp nil)) (simplifya (unsimplify x) y)))
561	((member 'simp (cdar x) :test #'eq) x)
562	((eq (caar x) 'mrat) x)
563	((not (atom (caar x)))
564	 (cond ((or (eq (caaar x) 'lambda)
565		    (and (not (atom (caaar x))) (eq (caaaar x) 'lambda)))
566		(mapply1 (caar x) (cdr x) (caar x) x))
567	       (t (merror (intl:gettext "simplifya: operator is neither an atom nor a lambda expression: ~S") x))))
568        ((and $distribute_over
569              (get (caar x) 'distribute_over)
570              ;; A function with the property 'distribute_over.
571              ;; Look, if we have a bag as argument to the function.
572              (distribute-over x)))
573	((get (caar x) 'opers)
574	 (let ((opers-list *opers-list)) (oper-apply x y)))
575	((and (eq (caar x) 'mqapply)
576	      (or (atom (cadr x))
577		  (and (eq substp 'mqapply)
578		       (or (eq (car (cadr x)) 'lambda)
579			   (eq (caar (cadr x)) 'lambda)))))
580	 (cond ((or (symbolp (cadr x)) (not (atom (cadr x))))
581		(simplifya (cons (cons (cadr x) (cdar x)) (cddr x)) y))
582	       ((or (not (member 'array (cdar x) :test #'eq)) (not $subnumsimp))
583		(merror (intl:gettext "simplifya: I don't know how to simplify this operator: ~M") x))
584	       (t (cadr x))))
585	(t (let ((w (get (caar x) 'operators)))
586	     (cond ((and w
587	                 (or (not (member 'array (cdar x) :test #'eq))
588	                     (rulechk (caar x))))
589		    (funcall w x 1 y))
590		   (t (simpargs x y)))))))
591
592;; EQTEST returns an expression which is the same as X
593;; except that it is marked with SIMP and maybe other flags from CHECK.
594;;
595;; Following description is inferred from the code. Dunno why the function is named "EQTEST".
596;;
597;; (1) if X is already marked with SIMP flag or doesn't need it: return X.
598;; (2) if X is pretty much the same as CHECK (same operator and same arguments),
599;; then return CHECK after marking it with SIMP flag.
600;; (3) if operator of X has the MSIMPIND property, replace it
601;; with value of MSIMPIND (something like '(MPLUS SIMP)) and return X.
602;; (4) if X or CHECK is an array expression, return X after marking it with SIMP and ARRAY flags.
603;; (5) otherwise, return X after marking it with SIMP flag.
604
605(defun eqtest (x check)
606  (let ((y nil))
607    (cond ((or (atom x)
608	       (eq (caar x) 'rat)
609	       (eq (caar x) 'mrat)
610	       (member 'simp (cdar x) :test #'eq))
611	   x)
612	  ((and (eq (caar x) (caar check))
613		(equal (cdr x) (cdr check)))
614	   (cond ((and (null (cdar check))
615		       (setq y (get (caar check) 'msimpind)))
616		  (cons y (cdr check)))
617		 ((member 'simp (cdar check) :test #'eq)
618		  check)
619		 (t
620		  (cons (cons (caar check)
621			      (if (cdar check)
622				  (cons 'simp (cdar check))
623				  '(simp)))
624			(cdr check)))))
625	  ((setq y (get (caar x) 'msimpind))
626	   (rplaca x y))
627	  ((or (member 'array (cdar x) :test #'eq)
628	       (and (eq (caar x) (caar check))
629		    (member 'array (cdar check) :test #'eq)))
630	   (rplaca x (cons (caar x) '(simp array))))
631	  (t
632	   (rplaca x (cons (caar x) '(simp)))))))
633
634;; A function, which distributes of bags like a list, matrix, or equation.
635;; Check, if we have to distribute of one of the bags or any other operator.
636(defun distribute-over (expr)
637  (cond ((= 1 (length (cdr expr)))
638         ;; Distribute over for a function with one argument.
639         (cond ((and (not (atom (cadr expr)))
640                     (member (caaadr expr) (get (caar expr) 'distribute_over)))
641                (simplify
642                  (cons (caadr expr)
643                        (mapcar #'(lambda (u) (simplify (list (car expr) u)))
644                                (cdadr expr)))))
645                (t nil)))
646        (t
647         ;; A function with more than one argument.
648         (do ((args (cdr expr) (cdr args))
649              (first-args nil))
650             ((null args) nil)
651           (when (and (not (atom (car args)))
652                      (member (caar (car args))
653                              (get (caar expr) 'distribute_over)))
654             ;; Distribute the function over the arguments and simplify again.
655             (return
656               (simplify
657                 (cons (ncons (caar (car args)))
658                       (mapcar #'(lambda (u)
659                                   (simplify
660                                     (append
661                                       (append
662                                         (cons (ncons (caar expr))
663                                               (reverse first-args))
664                                         (ncons u))
665                                       (rest args))))
666                               (cdr (car args)))))))
667           (setq first-args (cons (car args) first-args))))))
668
669(defun rulechk (x) (or (mget x 'oldrules) (get x 'rules)))
670
671(defun resimplify (x) (let ((dosimp t)) (simplifya x nil)))
672
673(defun unsimplify (x)
674  (if (or (atom x) (specrepp x))
675      x
676      (cons (remove 'simp (car x) :count 1) (mapcar #'unsimplify (cdr x)))))
677
678(defun simpargs (x y)
679  (if (or (eq (get (caar x) 'dimension) 'dimension-infix)
680	  (get (caar x) 'binary))
681      (twoargcheck x))
682  (if (and (member 'array (cdar x) :test #'eq) (null (margs x)))
683      (merror (intl:gettext "SIMPARGS: subscripted variable found with no subscripts.")))
684  (eqtest (if y x (let ((flag (member (caar x) '(mlist mequal) :test #'eq)))
685		    (cons (ncons (caar x))
686			  (mapcar #'(lambda (u)
687				      (if flag (simplifya u nil)
688					  (simpcheck u nil)))
689				  (cdr x)))))
690	  x))
691
692;;;-----------------------------------------------------------------------------
693;;; ADDK (X Y)                                                   27.09.2010/DK
694;;;
695;;; Arguments and values:
696;;;   X      - a Maxima number
697;;;   Y      - a Maxima number
698;;;   result - a simplified Maxima number
699;;;
700;;; Description:
701;;;   ADDK adds two Maxima numbers and returns a simplified Maxima number.
702;;;   ADDK can be called in Lisp code, whenever the arguments are valid
703;;;   Maxima numbers, these are integer, float, Maxima rational, or
704;;;   Maxima bigfloat numbers. The arguments must not be simplified. The
705;;;   precision of a bigfloat result depends on the setting of the
706;;;   global variable $FPPREC. If the option variable $FLOAT is T, a
707;;;   Maxima rational number as a result is converted to a float number.
708;;;
709;;; Examples:
710;;;   (addk 2 3) -> 5
711;;;   (addk 2.0 3) -> 5.0
712;;;   (addk ($bfloat 2) 3)-> ((BIGFLOAT SIMP 56) 45035996273704960 3)
713;;;   (addk 2 '((rat) 1 2)) -> ((RAT SIMP) 5 2)
714;;;   (let (($float t)) (addk 2 '((rat) 1 2))) -> 2.5
715;;;
716;;; Affected by:
717;;;   The option variables $FLOAT and $FPPREC.
718;;;
719;;; See also:
720;;;   TIMESK to multiply and EXPTRL to exponentiate two Maxima numbers.
721;;;
722;;; Notes:
723;;;   The routine works for Lisp rational and Lisp complex numbers too.
724;;;   This feature is not used in Maxima code. If Lisp complex and
725;;;   rational numbers are mixed with Maxima rational or bigfloat
726;;;   numbers the result is wrong or a Lisp error is generated.
727;;;-----------------------------------------------------------------------------
728
729(defun addk (x y)
730  (cond ((eql x 0) y)
731	((eql y 0) x)
732	((and (numberp x) (numberp y)) (+ x y))
733	((or ($bfloatp x) ($bfloatp y)) ($bfloat (list '(mplus) x y)))
734	(t (prog (g a b)
735	      (cond ((numberp x)
736		     (cond ((floatp x) (return (+ x (fpcofrat y))))
737			   (t (setq x (list '(rat) x 1)))))
738		    ((numberp y)
739		     (cond ((floatp y) (return (+ y (fpcofrat x))))
740			   (t (setq y (list '(rat) y 1))))))
741	      (setq g (gcd (caddr x) (caddr y)))
742	      (setq a (truncate (caddr x) g)
743	            b (truncate (caddr y) g))
744	      (return (timeskl (list '(rat) 1 g)
745			       (list '(rat)
746				     (+ (* (cadr x) b)
747					   (* (cadr y) a))
748				     (* a b))))))))
749
750;;;-----------------------------------------------------------------------------
751;;; *RED1 (X)                                                      27.09.2010/DK
752;;; *RED (N D)
753;;;
754;;; Arguments and values:
755;;;   X      - a Maxima rational number (for *RED1)
756;;;   N      - an integer number representing the numerator of a rational
757;;;   D      - an integer number representing the denominator of a rational
758;;;   result - a simplified Maxima rational number
759;;;
760;;; Description:
761;;;   *RED1 is called from SIMPLIFYA to reduce and simplify a Maxima rational
762;;;   number. *RED1 checks if the rational number is already simplified. If
763;;;   the option variable $FLOAT is T, the rational number is converted to a
764;;;   float number. If the number is not simplified, *RED is called.
765;;;
766;;;   *RED reduces the numerator N and the demoniator D and returns a
767;;;   simplified Maxima rational number. The result is converted to a float
768;;;   number, if the option variable $FLOAT is T.
769;;;
770;;; Affected by:
771;;;   The option variable $FLOAT.
772;;;-----------------------------------------------------------------------------
773
774(defun *red1 (x)
775  (cond ((member 'simp (cdar x) :test #'eq)
776	 (if $float (fpcofrat x) x))
777	(t (*red (cadr x) (caddr x)))))
778
779(defun *red (n d)
780  (cond ((zerop n) 0)
781	((equal d 1) n)
782	(t (let ((u (gcd n d)))
783	     (setq n (truncate n u)
784	           d (truncate d u))
785	     (if (minusp d) (setq n (- n) d (- d)))
786	     (cond ((equal d 1) n)
787		   ($float (fpcofrat1 n d))
788		   (t (list '(rat simp) n d)))))))
789
790;;;-----------------------------------------------------------------------------
791;;; TIMESK (X Y)                                                   27.09.2010/DK
792;;;
793;;; Arguments and values:
794;;;   X      - a Maxima number
795;;;   Y      - a Maxima number
796;;;   result - a simplified Maxima number
797;;;
798;;; Description:
799;;;   TIMESK Multiplies two Maxima numbers and returns a simplified Maxima
800;;;   number. TIMESK can be called in Lisp code, whenever the arguments are
801;;;   valid Maxima numbers, these are integer, float, Maxima rational, or
802;;;   Maxima bigfloat numbers. The arguments must not be simplified. The
803;;;   precision of a bigfloat result depends on the setting of the
804;;;   global variable $FPPREC. If the option variable $FLOAT is T, a
805;;;   Maxima rational number as a result is converted to a float number.
806;;;
807;;;   TIMESKL is called from TIMESK to multiply two Maxima rational numbers or
808;;;   a rational number with an integer number.
809;;;
810;;; Examples:
811;;;   (timesk 2 3) -> 6
812;;;   (timesk 2.0 3) -> 6.0
813;;;   (timesk ($bfloat 2) 3)-> ((BIGFLOAT SIMP 56) 54043195528445952 3)
814;;;   (timesk 3 '((rat) 1 2)) -> ((RAT SIMP) 3 2)
815;;;   (let (($float t)) (timesk 3 '((rat) 1 2))) -> 1.5
816;;;
817;;; Affected by:
818;;;   The option variables $FLOAT and $FPPREC.
819;;;
820;;; See also:
821;;;   ADDK to add and EXPTRL to exponentiate two Maxima numbers.
822;;;
823;;; Notes:
824;;;   The routine works for Lisp rational and Lisp complex numbers too.
825;;;   This feature is not used in Maxima code. If Lisp complex and
826;;;   rational numbers are mixed with Maxima rational or bigfloat
827;;;   numbers the result is wrong or a Lisp error is generated.
828;;;-----------------------------------------------------------------------------
829
830;; NUM1 and DENOM1 are helper functions for TIMESKL to get the numerator and the
831;; denominator of an integer or Maxima rational number. For an integer the
832;; denominator is 1. Both functions are used at other places in Maxima code too.
833
834(defun num1 (a)
835  (if (numberp a) a (cadr a)))
836
837(defun denom1 (a)
838  (if (numberp a) 1 (caddr a)))
839
840(defun timesk (x y)     ; X and Y are assumed to be already reduced
841  (cond ((equal x 1) y)
842	((equal y 1) x)
843	((and (numberp x) (numberp y)) (* x y))
844	((or ($bfloatp x) ($bfloatp y)) ($bfloat (list '(mtimes) x y)))
845	((floatp x) (* x (fpcofrat y)))
846	((floatp y) (* y (fpcofrat x)))
847	(t (timeskl x y))))
848
849;; TIMESKL takes one or two Maxima rational numbers, one argument can be an
850;; integer number. The result is a Maxima rational or an integer number.
851;; If the option variable $FLOAT is T, a Maxima rational number in converted
852;; to a float value.
853
854(defun timeskl (x y)
855  (prog (u v g)
856     (setq u (*red (num1 x) (denom1 y)))
857     (setq v (*red (num1 y) (denom1 x)))
858     (setq g (cond ((or (equal u 0) (equal v 0)) 0)
859		   ((equal v 1) u)
860		   ((and (numberp u) (numberp v)) (* u v))
861		   (t (list '(rat simp)
862			    (* (num1 u) (num1 v))
863			    (* (denom1 u) (denom1 v))))))
864     (return (cond ((numberp g) g)
865		   ((equal (caddr g) 1) (cadr g))
866		   ($float (fpcofrat g))
867		   (t g)))))
868
869;;;-----------------------------------------------------------------------------
870;;; FPCOFRAT (RATNO)                                               27.09.2010/DK
871;;; FPCOFRT1 (NU D)
872;;;
873;;; Arguments and values:
874;;;   RATNO  - a Maxima rational number (for FPCOFRAT)
875;;;   NU     - an integer number which represents the numerator of a rational
876;;;   D      - an integer number which represents the denominator of a rational
877;;;   result - floating point approximation of a rational number
878;;;
879;;; Description:
880;;;   Floating Point Conversion OF RATional number routine.
881;;;   Finds floating point approximation to rational number.
882;;;
883;;;   FPCOFRAT1 computes the quotient of NU/D.
884;;;
885;;; Exceptional situations:
886;;;   A Lisp error is generated, if the rational number does not fit into a
887;;;   float number.
888;;;-----------------------------------------------------------------------------
889
890;; This constant is only needed in the file float.lisp.
891(eval-when
892    #+gcl (compile load eval)
893    #-gcl (:compile-toplevel :load-toplevel :execute)
894    (defconstant machine-mantissa-precision (float-digits 1.0)))
895
896(defun fpcofrat (ratno)
897  (fpcofrat1 (cadr ratno) (caddr ratno)))
898
899(defun fpcofrat1 (nu d)
900  (float (/ nu d)))
901
902;;;-----------------------------------------------------------------------------
903;;; EXPTA (X Y)                                                    27.09.2010/DK
904;;;
905;;; Arguments and values:
906;;;   X      - a Maxima number
907;;;   Y      - an integer number
908;;;   result - a simplified Maxima number
909;;;
910;;; Description:
911;;;   Computes X^Y, where X is Maxima number and Y an integer. The result is
912;;;   a simplified Maxima number. Y can be a rational Maxima number. For this
913;;;   case the numerator is taken as the power.
914;;;
915;;; Affected by:
916;;;   The option variables $FLOAT and $FPPREC.
917;;;
918;;; Notes:
919;;;   This routine is not used within the simplifier. There is only one
920;;;   call from the file hayat.lisp. This call can be replaced with a
921;;;   call of the function power.
922;;;-----------------------------------------------------------------------------
923
924(defun expta (x y)
925  (cond ((equal y 1)
926	 x)
927	((numberp x)
928	 (exptb x (num1 y)))
929	(($bfloatp x)
930	 ($bfloat (list '(mexpt) x y)))
931	((minusp (num1 y))
932	 (*red (exptb (caddr x) (- (num1 y)))
933	       (exptb (cadr x) (- (num1 y)))))
934	(t
935	 (*red (exptb (cadr x) (num1 y))
936	       (exptb (caddr x) (num1 y))))))
937
938;;;-----------------------------------------------------------------------------
939;;; EXPTB (A B)                                                    27.09.2010/DK
940;;;
941;;; Arguments and values:
942;;;   A      - a float or integer number
943;;;   B      - an integer number
944;;;   result - a simplified Maxima number
945;;;
946;;; Description:
947;;;   Computes A^B, where A is a float or an integer number and B is an
948;;;   integer number. The result is an integer, float, or Maxima
949;;;   rational number.
950;;;
951;;; Examples:
952;;;   (exptb 3 2)   -> 9
953;;;   (exptb 3.0 2) -> 9.0
954;;;   (exptb 3 -2)  -> ((RAT SiMP) 1 9)
955;;;   (let (($float t)) (exptb 3 -2)) -> 0.1111111111111111
956;;;
957;;; Affected by:
958;;;   The option variable $FLOAT.
959;;;
960;;; Notes:
961;;;   EXPTB calls the Lisp functions EXP or EXPT to compute the result.
962;;;-----------------------------------------------------------------------------
963
964(defun exptb (a b)
965  (let ((result
966	 (cond ((equal a %e-val)
967		;; Make B a float so we'll get double-precision result.
968		(exp (float b)))
969	       ((or (floatp a) (not (minusp b)))
970		(expt a b))
971	       (t
972		(setq b (expt a (- b)))
973		(*red 1 b)))))
974    (if (float-inf-p result)	;; needed for gcl - no trap of overflow
975	(signal 'floating-point-overflow)
976      result)))
977
978
979;;;-----------------------------------------------------------------------------
980;;; SIMPLUS (X W Z)                                                27.09.2010/DK
981;;;
982;;; Arguments and values:
983;;;   X      - a Maxima expression of the form ((mplus) term1 term2 ...)
984;;;   W      - an arbitrary value, the value is ignored
985;;;   Z      - T or NIL, if T the arguments are assumed to be simplified
986;;;   result - a simplified mplus-expression or an atom
987;;;
988;;; Description:
989;;;  Implementation of the simplifier for the "+" operator.
990;;;  A general description of SIMPLUS can be found in the paper:
991;;;    http://www.cs.berkeley.edu/~fateman/papers/simplifier.txt
992;;;
993;;; Affected by:
994;;;   The addition of matrices and lists is affected by the following option
995;;;   variables:
996;;;   $DOALLMXOPS, $DOMXMXOPS, $DOMXPLUS, $DOSCMXOPPS, $DOSCMXPLUS, $LISTARITH
997;;;
998;;; Notes:
999;;;   This routine should not be called directely. It is called by SIMPLIFYA.
1000;;;   A save access is to call the function ADD.
1001;;;-----------------------------------------------------------------------------
1002
1003(defun simplus (x w z)
1004  (prog (res check eqnflag matrixflag sumflag)
1005     (if (null (cdr x)) (return 0))
1006     (setq check x)
1007  start
1008     (setq x (cdr x))
1009     (if (null x) (go end))
1010     (setq w (if z (car x) (simplifya (car x) nil)))
1011  st1
1012     (cond ((atom w) nil)
1013           ((eq (caar w) 'mrat)
1014            (cond ((or eqnflag
1015                       matrixflag
1016                       (and sumflag
1017                            (not (member 'trunc (cdar w) :test #'eq)))
1018                       (spsimpcases (cdr x) w))
1019                   (setq w (ratdisrep w))
1020                   (go st1))
1021                  (t
1022                   (return
1023                     (ratf (cons '(mplus)
1024                                 (nconc (mapcar #'simplify (cons w (cdr x)))
1025                                        (cdr res))))))))
1026           ((eq (caar w) 'mequal)
1027            (setq eqnflag
1028                  (if (not eqnflag)
1029                      w
1030                      (list (car eqnflag)
1031                            (add2 (cadr eqnflag) (cadr w))
1032                            (add2 (caddr eqnflag) (caddr w)))))
1033            (go start))
1034           ((member (caar w) '(mlist $matrix) :test #'eq)
1035            (setq matrixflag
1036                  (cond ((not matrixflag) w)
1037                        ((and (or $doallmxops $domxmxops $domxplus
1038                                  (and (eq (caar w) 'mlist)
1039                                       ($listp matrixflag)))
1040                              (or (not (eq (caar w) 'mlist)) $listarith))
1041                         (addmx matrixflag w))
1042                        (t (setq res (pls w res)) matrixflag)))
1043            (go start))
1044           ((eq (caar w) '%sum)
1045            (setq sumflag t res (sumpls w res))
1046            (setq w (car res) res (cdr res))))
1047     (setq res (pls w res))
1048     (go start)
1049  end
1050     (setq res (testp res))
1051     (if matrixflag
1052         (setq res
1053               (cond ((and (or ($listp matrixflag)
1054                               $doallmxops $doscmxplus $doscmxops)
1055                           (or (not ($listp matrixflag)) $listarith))
1056                      (mxplusc res matrixflag))
1057                     (t (testp (pls matrixflag (pls res nil)))))))
1058     (setq res (eqtest res check))
1059     (return (if eqnflag
1060                 (list (car eqnflag)
1061                       (add2 (cadr eqnflag) res)
1062                       (add2 (caddr eqnflag) res))
1063                 res))))
1064
1065;;;-----------------------------------------------------------------------------
1066;;; PLS (X OUT)                                                    27.09.2010/DK
1067;;;
1068;;; Arguments and values:
1069;;;   X      - a Maxima expression or an atom
1070;;;   OUT    - a form ((mplus) <number> term1 term2 ...) or NIL
1071;;;   result - a form ((mplus) <number> term1 ...), where x is added in.
1072;;;
1073;;; Description:
1074;;;   Adds the argument X into the form OUT. If OUT is NIL a form
1075;;;   ((mplus) 0 X) is initialized, if X is an expression or a symbol,
1076;;;   or ((mplus) X), if X is a number. Numbers are added to the first
1077;;;   term <number> of the form. Any other symbol or expression is added
1078;;;   into the canonical ordered list of arguments. The result is in a
1079;;;   canonical order, but it is not a valid Maxima expression. To get a
1080;;;   valid Maxima expression the result has to be checked with the
1081;;;   function TESTP. This is done by the calling routine SIMPLUS.
1082;;;
1083;;;   PLS checks the global flag *PLUSFLAG*, which is set in PLUSIN to T,
1084;;;   if a mplus-expression is part of the result.
1085;;;
1086;;; Examples:
1087;;;   (pls 2 nil) -> ((MPLUS) 2)
1088;;;   (pls '$A nil) -> ((MPLUS) 0 $A)
1089;;;   (pls '$B '((mplus) 0 $A)) -> ((MPLUS) 0 $A $B)
1090;;;   (pls '$A '((mplus) 0 $A)) -> ((MPLUS) 0 ((MTIMES SIMP) 2 $A))
1091;;;
1092;;; Examples with the option variables $NUMER and $NEGDISTRIB:
1093;;;   (let (($numer t)) (pls '$%e nil)) -> ((MPLUS) 2.718281828459045)
1094;;;   (let (($negdistrib t)) (pls '((mtimes) -1 ((mplus) $A $B)) nil))
1095;;;           -> ((MPLUS) 0 ((MTIMES SIMP) -1 $A) ((MTIMES SIMP) -1 $B))
1096;;;   (let (($negdistrib nil)) (pls '((mtimes) -1 ((mplus) $A $B)) nil))
1097;;;           -> ((MPLUS) 0 ((MTIMES) -1 ((MPLUS) $A $B)))
1098;;;
1099;;; Affected by:
1100;;;   The option variables $NUMER and $NEGDISTRIB and the global flag
1101;;;   *PLUSFLAG*, which is set in the routine PLUSIN.
1102;;;
1103;;; See also:
1104;;;   PLUSIN and ADDK which are called from PLS and SIMPLUS.
1105;;;
1106;;; Notes:
1107;;;   To add an expression into the list (CDR OUT), the list is passed
1108;;;   to the routine PLUSIN as an argument. PLUSIN adds the argument to
1109;;;   the list of terms by modifying the list (CDR OUT) destructively.
1110;;;   The new value of OUT is returned as a result by PLS.
1111;;;-----------------------------------------------------------------------------
1112
1113;; Set in PLUSIN to T to indicate a nested mplus expression.
1114(defvar *plusflag* nil)
1115
1116;; TESTP checks the result of PLS to get a valid Maxima mplus-expression.
1117
1118(defun testp (x)
1119  (cond ((atom x) 0)
1120        ((null (cddr x)) (cadr x))
1121        ((zerop1 (cadr x))
1122         (cond ((null (cdddr x)) (caddr x)) (t (rplacd x (cddr x)))))
1123        (t x)))
1124
1125(defun pls (x out)
1126  (prog (fm *plusflag*)
1127     (if (mtimesp x) (setq x (testtneg x)))
1128     (when (and $numer (atom x) (eq x '$%e))
1129       ;; Replace $%e with its numerical value, when $numer ist TRUE
1130       (setq x %e-val))
1131     (cond ((null out)
1132            ;; Initialize a form like ((mplus) <number> expr)
1133            (return
1134              (cons '(mplus)
1135                    (cond ((mnump x) (ncons x))
1136                          ((not (mplusp x))
1137                           (list 0 (cond ((atom x) x) (t (copy-list x)))))
1138                          ((mnump (cadr x)) (copy-list (cdr x) ))
1139                          (t (cons 0 (copy-list (cdr x) )))))))
1140           ((mnump x)
1141            ;; Add a number into the first term of the list out.
1142            (return (cons '(mplus)
1143                          (if (mnump (cadr out))
1144                              (cons (addk (cadr out) x) (cddr out))
1145                              (cons x (cdr out))))))
1146           ((not (mplusp x)) (plusin x (cdr out)) (go end)))
1147     ;; At this point we have a mplus expression as argument x. The following
1148     ;; code assumes that the argument x is already simplified and the terms
1149     ;; are in a canonical order.
1150     ;; First we add the number to the first term of the list out.
1151     (rplaca (cdr out)
1152             (addk (if (mnump (cadr out)) (cadr out) 0)
1153                   (cond ((mnump (cadr x)) (setq x (cdr x)) (car x)) (t 0))))
1154     ;; Initialize fm with the list of terms and start the loop to add the
1155     ;; terms of an mplus expression into the list out.
1156     (setq fm (cdr out))
1157  start
1158     (if (null (setq x (cdr x))) (go end))
1159     ;; The return value of PLUSIN is a list, where the first element is the
1160     ;; added argument and the rest are the terms which follow the added
1161     ;; argument.
1162     (setq fm (plusin (car x) fm))
1163     (go start)
1164  end
1165     (if (not *plusflag*) (return out))
1166     (setq *plusflag* nil)   ; *PLUSFLAG* T handles e.g. a+b+3*(a+b)-2*(a+b)
1167  a
1168     ;; *PLUSFLAG* is set by PLUSIN to indicate that a mplus expression is
1169     ;; part of the result. For this case go again through the terms of the
1170     ;; result and add any term of the mplus expression into the list out.
1171     (setq fm (cdr out))
1172  loop
1173     (when (mplusp (cadr fm))
1174       (setq x (cadr fm))
1175       (rplacd fm (cddr fm))
1176       (pls x out)
1177       (go a))
1178     (setq fm (cdr fm))
1179     (if (null (cdr fm)) (return out))
1180     (go loop)))
1181
1182;;;-----------------------------------------------------------------------------
1183;;; PLUSIN (X FM)                                                  27.09.2010/DK
1184;;;
1185;;; Arguments and values:
1186;;;   X      - a Maxima expression or atom
1187;;;   FM     - a list with the terms of an addition
1188;;;   result - part of the list fm, which starts at the inserted expression
1189;;;
1190;;; Description:
1191;;;   Adds X into running list of additive terms FM. The routine modifies
1192;;;   the argument FM destructively, but does not return the modified list as
1193;;;   a result. The return value is a part of the list FM, which starts at the
1194;;;   inserted term. PLUSIN can not handle Maxima numbers. PLUSIN is called
1195;;;   only from the routine PLS.
1196;;;
1197;;; Examples:
1198;;;   (setq fm '(0))
1199;;;   (plusin '$a fm) -> ($A)
1200;;;   fm -> (0 $A)
1201;;;   (plusin '$b fm) -> ($B)
1202;;;   fm -> (0 $A $B)
1203;;;   (plusin '$a fm) -> (((MTIMES SIMP) 2 $A) $B)
1204;;;   fm -> (0 ((MTIMES SIMP) 2 $A) $B)
1205;;;
1206;;; Side effects:
1207;;;   Modifies destructively the argument FM, which contains the result of the
1208;;;   addition of the argument X into the list FM.
1209;;;
1210;;; Affected by;
1211;;;   The option variables $doallmxops and $listarith.
1212;;;
1213;;; Notes:
1214;;;   The return value is used in PLS to go in parallel through the list of
1215;;;   terms, when adding a complete mplus-expression into the list of terms.
1216;;;   This is triggered by the flag *PLUSFLAG*, which is set in PLUSIN, if
1217;;;   a mplus-expression is added to the result list.
1218;;;-----------------------------------------------------------------------------
1219
1220(defun plusin (x fm)
1221  (prog (x1 x2 flag check v w xnew a n m c)
1222     (setq w 1)
1223     (setq v 1)
1224     (cond ((mtimesp x)
1225            (setq check x)
1226            (if (mnump (cadr x)) (setq w (cadr x) x (cddr x))
1227                (setq x (cdr x))))
1228           (t (setq x (ncons x))))
1229     (setq x1 (if (null (cdr x)) (car x) (cons '(mtimes) x))
1230           xnew (list* '(mtimes) w x))
1231  start
1232     (cond ((null (cdr fm)))
1233           ((and (alike1 x1 (cadr fm)) (null (cdr x)))
1234            (go equ))
1235           ;; Implement the simplification of
1236           ;;   v*a^(c+n)+w*a^(c+m) -> (v*a^n+w*a^m)*a^c
1237           ;; where a, v, w, and (n-m) are integers.
1238           ((and (or (and (mexptp (setq x2 (cadr fm)))
1239                          (setq v 1))
1240                     (and (mtimesp x2)
1241                          (not (alike1 x1 x2))
1242                          (null (cadddr x2))
1243                          (integerp (setq v (cadr x2)))
1244                          (mexptp (setq x2 (caddr x2)))))
1245                 (integerp (setq a (cadr x2)))
1246                 (mexptp x1)
1247                 (equal a (cadr x1))
1248                 (integerp (sub (caddr x2) (caddr x1))))
1249            (setq n (if (and (mplusp (caddr x2))
1250                             (mnump (cadr (caddr x2))))
1251                        (cadr (caddr x2))
1252                        (if (mnump (caddr x2))
1253                            (caddr x2)
1254                            0)))
1255            (setq m (if (and (mplusp (caddr x1))
1256                             (mnump (cadr (caddr x1))))
1257                        (cadr (caddr x1))
1258                        (if (mnump (caddr x1))
1259                            (caddr x1)
1260                            0)))
1261            (setq c (sub (caddr x2) n))
1262            (cond ((integerp n)
1263                   ;; The simple case:
1264                   ;; n and m are integers and the result is (v*a^n+w*a^m)*a^c.
1265                   (setq x1 (mul (addk (timesk v (exptb a n))
1266                                       (timesk w (exptb a m)))
1267                                 (power a c)))
1268                   (go equt2))
1269                  (t
1270                   ;; n and m are rational numbers: The difference n-m is an
1271                   ;; integer. The rational numbers might be improper fractions.
1272                   ;; The mixed numbers are: n = n1 + d1/r and m = n2 + d2/r,
1273                   ;; where r is the common denominator. We have two cases:
1274                   ;; I)  d1 = d2: e.g. 2^(1/3+c)+2^(4/3+c)
1275                   ;;     The result is (v*a^n1+w*a^n2)*a^(c+d1/r)
1276                   ;; II) d1 # d2: e.g. 2^(1/2+c)+2^(-1/2+c)
1277                   ;;     In this case one of the exponents d1 or d2 must
1278                   ;;     be negative. The negative exponent is factored out.
1279                   ;;     This guarantees that the factor (v*a^n1+w*a^n2)
1280                   ;;     is an integer. But the positive exponent has to be
1281                   ;;     adjusted accordingly. E.g. when we factor out
1282                   ;;     a^(d2/r) because d2 is negative, then we have to
1283                   ;;     adjust the positive exponent to n1 -> n1+(d1-d2)/r.
1284                   ;; Remark:
1285                   ;; Part of the simplification is done in simptimes. E.g.
1286                   ;; this algorithm simplifies the sum sqrt(2)+3*sqrt(2)
1287                   ;; to 4*sqrt(2). In simptimes this is further simplified
1288                   ;; to 2^(5/2).
1289                   (multiple-value-bind (n1 d1)
1290                       (truncate (num1 n) (denom1 n))
1291                     (multiple-value-bind (n2 d2)
1292                         (truncate (num1 m) (denom1 m))
1293                       (cond ((equal d1 d2)
1294                              ;; Case I: -> (v*a^n1+w*a^n2)*a^(c+d1/r)
1295                              (setq x1
1296                                    (mul (addk (timesk v (exptb a n1))
1297                                               (timesk w (exptb a n2)))
1298                                         (power a
1299                                                (add c
1300                                                     (div d1 (denom1 n))))))
1301                              (go equt2))
1302                             ((minusp d2)
1303                              ;; Case II:: d2 is negative, adjust n1.
1304                              (setq n1 (add n1 (div (sub d1 d2) (denom1 n))))
1305                              (setq x1
1306                                    (mul (addk (timesk v (exptb a n1))
1307                                               (timesk w (exptb a n2)))
1308                                         (power a
1309                                                (add c
1310                                                     (div d2 (denom1 n))))))
1311                              (go equt2))
1312                             ((minusp d1)
1313                              ;; Case II: d1 is negative, adjust n2.
1314                              (setq n2 (add n2 (div (sub d2 d1) (denom1 n))))
1315                              (setq x1
1316                                    (mul (addk (timesk v (exptb a n1))
1317                                               (timesk w (exptb a n2)))
1318                                         (power a
1319                                                (add c
1320                                                     (div d1 (denom1 n))))))
1321                              (go equt2))
1322                             ;; This clause should never be reached.
1323                             (t (merror "Internal error in simplus."))))))))
1324           ((mtimesp (cadr fm))
1325            (cond ((alike1 x1 (cadr fm))
1326                   (go equt))
1327                  ((and (mnump (cadadr fm)) (alike x (cddadr fm)))
1328                   (setq flag t) ; found common factor
1329                   (go equt))
1330                  ((great xnew (cadr fm)) (go gr))))
1331           ((great x1 (cadr fm)) (go gr)))
1332     (setq xnew (eqtest (testt xnew) (or check '((foo)))))
1333     (return (cdr (rplacd fm (cons xnew (cdr fm)))))
1334  gr
1335     (setq fm (cdr fm))
1336     (go start)
1337  equ
1338     (rplaca (cdr fm)
1339             (if (equal w -1)
1340                 (list* '(mtimes simp) 0 x)
1341                 ;; Call muln to get a simplified product.
1342                 (if (mtimesp (setq x1 (muln (cons (addk 1 w) x) t)))
1343                     (testtneg x1)
1344                     x1)))
1345  del
1346     (cond ((not (mtimesp (cadr fm)))
1347            (go check))
1348           ((onep (cadadr fm))
1349            ;; Do this simplification for an integer 1, not for 1.0 and 1.0b0
1350            (rplacd (cadr fm) (cddadr fm))
1351            (return (cdr fm)))
1352           ((not (zerop1 (cadadr fm)))
1353            (return (cdr fm)))
1354           ;; Handle the multiplication with a zero.
1355           ((and (or (not $listarith) (not $doallmxops))
1356                 (mxorlistp (caddr (cadr fm))))
1357            (return (rplacd fm
1358                            (cons (constmx 0 (caddr (cadr fm))) (cddr fm))))))
1359     ;; (cadadr fm) is zero. If the first term of fm is a number,
1360     ;;  add it to preserve the type.
1361     (when (mnump (car fm))
1362       (rplaca fm (addk (car fm) (cadadr fm))))
1363     (return (rplacd fm (cddr fm)))
1364  equt
1365     ;; Call muln to get a simplified product.
1366     (setq x1 (muln (cons (addk w (if flag (cadadr fm) 1)) x) t))
1367     ;; Make a mplus expression to guarantee that x1 is added again into the sum
1368     (setq x1 (list '(mplus) x1))
1369  equt2
1370     (rplaca (cdr fm)
1371             (if (zerop1 x1)
1372                 (list* '(mtimes) x1 x)
1373                 (if (mtimesp x1) (testtneg x1) x1)))
1374     (if (not (mtimesp (cadr fm))) (go check))
1375     (when (and (onep (cadadr fm)) flag (null (cdddr (cadr fm))))
1376       ;; Do this simplification for an integer 1, not for 1.0 and 1.0b0
1377       (rplaca (cdr fm) (caddr (cadr fm))) (go check))
1378     (go del)
1379  check
1380     (if (mplusp (cadr fm)) (setq *plusflag* t)) ; A nested mplus expression
1381     (return (cdr fm))))
1382
1383;;;-----------------------------------------------------------------------------
1384
1385;; Routines to add matrices
1386
1387(defun mxplusc (sc mx)
1388  (cond ((mplusp sc)
1389	 (setq sc (partition-ns (cdr sc)))
1390	 (cond ((null (car sc)) (cons '(mplus) (cons mx (cadr sc))))
1391	       ((not (null (cadr sc)))
1392		(cons '(mplus)
1393		      (cons (simplify
1394			     (outermap1 'mplus (cons '(mplus) (car sc)) mx))
1395			    (cadr sc))))
1396	       (t (simplify (outermap1 'mplus (cons '(mplus) (car sc)) mx)))))
1397	((not (scalar-or-constant-p sc $assumescalar))
1398	 (testp (pls mx (pls sc nil))))
1399	(t (simplify (outermap1 'mplus sc mx)))))
1400
1401(defun partition-ns (x)
1402  (let (sp nsp)		      ; SP = scalar part, NSP = nonscalar part
1403    (mapc #'(lambda (z) (if (scalar-or-constant-p z $assumescalar)
1404			    (setq sp (cons z sp))
1405			    (setq nsp (cons z nsp))))
1406	  x)
1407    (list (nreverse sp) (nreverse nsp))))
1408
1409(defun addmx (x1 x2)
1410  (let (($doscmxops t) ($domxmxops t) ($listarith t))
1411    (simplify (fmapl1 'mplus x1 x2))))
1412
1413;;; ----------------------------------------------------------------------------
1414
1415;;; Simplification of the Log function
1416
1417;; The log function distributes over lists, matrices, and equations
1418(defprop %log (mlist $matrix mequal) distribute_over)
1419
1420(defun simpln (x y z)
1421  (oneargcheck x)
1422  (setq y (simpcheck (cadr x) z))
1423  (cond ((onep1 y) (addk -1 y))
1424        ((zerop1 y)
1425         (cond (radcanp (list '(%log simp) 0))
1426               ((not errorsw)
1427                (merror (intl:gettext "log: encountered log(0).")))
1428               (t (throw 'errorsw t))))
1429        ;; Check evaluation in floating point precision.
1430        ((flonum-eval (mop x) y))
1431        ;; Check evaluation in bigfloag precision.
1432        ((and (not (member 'simp (car x) :test #'eq))
1433              (big-float-eval (mop x) y)))
1434        ((eq y '$%e) 1)
1435        ((mexptp y)
1436         (cond ((or (and $logexpand (eq $domain '$real))
1437                    (member $logexpand '($all $super))
1438                    (and (eq ($csign (cadr y)) '$pos)
1439                         (not (member ($csign (caddr y))
1440                                      '($complex $imaginary)))))
1441                ;; Simplify log(x^a) -> a*log(x), where x > 0 and a is real
1442                (mul (caddr y) (take '(%log) (cadr y))))
1443               ((or (and (ratnump (caddr y))
1444                         (or (eql 1 (cadr (caddr y)))
1445                             (eql -1 (cadr (caddr y)))))
1446                    (maxima-integerp (inv (caddr y))))
1447                ;; Simplify log(z^(1/n)) -> log(z)/n, where n is an integer
1448                (mul (caddr y)
1449                     (take '(%log) (cadr y))))
1450               ((and (eq (cadr y) '$%e)
1451                     (or (not (member ($csign (caddr y))
1452                                      '($complex $imaginary)))
1453                         (not (member ($csign (mul '$%i (caddr y)))
1454                                      '($complex $imaginary)))))
1455                ;; Simplify log(exp(x)) and log(exp(%i*x)), where x is a real
1456                (caddr y))
1457               (t (eqtest (list '(%log) y) x))))
1458        ((ratnump y)
1459         ;; Simplify log(n/d)
1460         (cond ((eql (cadr y) 1)
1461                (mul -1 (take '(%log) (caddr y))))
1462               ((eq $logexpand '$super)
1463                (sub (take '(%log) (cadr y)) (take '(%log) (caddr y))))
1464               (t (eqtest (list '(%log) y) x))))
1465        ((and (member $logexpand '($all $super) :test #'eq)
1466              (mtimesp y))
1467         (do ((y (cdr y) (cdr y))
1468              (b nil))
1469             ((null y) (return (addn b t)))
1470           (setq b (cons (take '(%log) (car y)) b))))
1471        ((and (member $logexpand '($all $super))
1472              (consp y)
1473              (member (caar y) '(%product $product)))
1474         (let ((new-op (if (char= (get-first-char (caar y)) #\%) '%sum '$sum)))
1475           (simplifya `((,new-op) ((%log) ,(cadr y)) ,@(cddr y)) t)))
1476        ((and $lognegint
1477              (maxima-integerp y)
1478              (eq ($sign y) '$neg))
1479         (add (mul '$%i '$%pi) (take '(%log) (neg y))))
1480        ((taylorize (mop x) (second x)))
1481        (t (eqtest (list '(%log) y) x))))
1482
1483(defun simpln1 (w)
1484  (simplifya (list '(mtimes) (caddr w)
1485		   (simplifya (list '(%log) (cadr w)) t)) t))
1486
1487;;; ----------------------------------------------------------------------------
1488
1489;;; Implementation of the Square root function
1490
1491(defprop $sqrt %sqrt verb)
1492(defprop $sqrt %sqrt alias)
1493
1494(defprop %sqrt $sqrt noun)
1495(defprop %sqrt $sqrt reversealias)
1496
1497(defprop %sqrt simp-sqrt operators)
1498
1499(defmfun $sqrt (z)
1500  (simplify (list '(%sqrt) z)))
1501
1502(defun simp-sqrt (x ignored z)
1503  (declare (ignore ignored))
1504  (oneargcheck x)
1505  (simplifya (list '(mexpt) (cadr x) '((rat simp) 1 2)) z))
1506
1507;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1508
1509;;; Simplification of the "/" operator.
1510
1511(defun simpquot (x y z)
1512  (twoargcheck x)
1513  (cond ((and (integerp (cadr x)) (integerp (caddr x)) (not (zerop (caddr x))))
1514	 (*red (cadr x) (caddr x)))
1515	((and (numberp (cadr x)) (numberp (caddr x)) (not (zerop (caddr x))))
1516	 (/ (cadr x) (caddr x)))
1517	((and (floatp (cadr x)) (floatp (caddr x)) #-ieee-floating-point (not (zerop (caddr x))))
1518	 (/ (cadr x) (caddr x)))
1519	((and ($bfloatp (cadr x)) ($bfloatp (caddr x)) (not (equal bigfloatzero (caddr x))))
1520	 ;; Call BIGFLOATP to ensure that arguments have same precision.
1521	 ;; Otherwise FPQUOTIENT could return a spurious value.
1522	 (bcons (fpquotient (cdr (bigfloatp (cadr x))) (cdr (bigfloatp (caddr x))))))
1523	(t (setq y (simplifya (cadr x) z))
1524	   (setq x (simplifya (list '(mexpt) (caddr x) -1) z))
1525	   (if (equal y 1) x (simplifya (list '(mtimes) y x) t)))))
1526
1527;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1528
1529;;; Implementation of the abs function.
1530
1531;; Put the properties alias, reversealiases, noun and verb on the property list.
1532(defprop $abs mabs alias)
1533(defprop $abs mabs verb)
1534(defprop mabs $abs reversealias)
1535(defprop mabs $abs noun)
1536
1537;; The abs function distributes over bags.
1538(defprop mabs (mlist $matrix mequal) distribute_over)
1539
1540;; Define a verb function $abs
1541(defmfun $abs (x)
1542  (simplify (list '(mabs) x)))
1543
1544;; The abs function is a simplifying function.
1545(defprop mabs simpabs operators)
1546
1547(defun simpabs (e y z)
1548  (declare (ignore y))
1549  (oneargcheck e)
1550  (let ((sgn)
1551	(x (simpcheck (second e) z)))
1552
1553    (cond ((complex-number-p x #'(lambda (s) (or (floatp s) ($bfloatp s))))
1554	   (maxima::to (bigfloat::abs (bigfloat:to x))))
1555
1556	  ((complex-number-p x #'mnump)
1557	   ($cabs x))
1558
1559	  ;; nounform for arrays...
1560	  ((or (arrayp x) ($member x $arrays)) `((mabs simp) ,x))
1561
1562	  ;; taylor polynomials
1563	  ((taylorize 'mabs x))
1564
1565	  ;; values for extended real arguments:
1566	  ((member x '($inf $infinity $minf) :test #'eq) '$inf)
1567	  ((member x '($ind $und) :test #'eq) x)
1568
1569	  ;; abs(abs(expr)) --> abs(expr). Since x is simplified, it's OK to return x.
1570	  ((and (consp x) (consp (car x)) (eq (caar x) 'mabs))
1571	   x)
1572
1573	  ;; abs(conjugate(expr)) = abs(expr).
1574	  ((and (consp x) (consp (car x)) (eq (caar x) '$conjugate))
1575	   (take '(mabs) (cadr x)))
1576
1577	  (t
1578	   (setq sgn ($csign x))
1579	   (cond ((member sgn '($neg $nz) :test #'eq) (mul -1 x))
1580		 ((eq '$zero sgn) (mul 0 x))
1581		 ((member sgn '($pos $pz) :test #'eq) x)
1582
1583		 ;; for complex constant expressions, use $cabs
1584		 ((and (eq sgn '$complex) ($constantp x))
1585		  ($cabs x))
1586
1587		 ;; abs(pos^complex) --> pos^(realpart(complex)).
1588		 ((and (eq sgn '$complex) (mexptp x) (eq '$pos ($csign (second x))))
1589		  (power (second x) ($realpart (third x))))
1590
1591		 ;; for abs(neg^z), use cabs.
1592		 ((and (mexptp x) (eq '$neg ($csign (second x))))
1593		  ($cabs x))
1594
1595		 ;; When x # 0, we have abs(signum(x)) = 1.
1596		 ((and (eq '$pn sgn) (consp x) (consp (car x)) (eq (caar x) '%signum)) 1)
1597
1598		 ;; multiplicative property: abs(x*y) = abs(x) * abs(y). We would like
1599		 ;; assume(a*b > 0), abs(a*b) --> a*b. Thus the multiplicative property
1600		 ;; is applied after the sign test.
1601		 ((mtimesp x)
1602		  (muln (mapcar #'(lambda (u) (take '(mabs) u)) (margs x)) t))
1603
1604		 ;; abs(x^n) = abs(x)^n for integer n. Is the featurep check worthwhile?
1605		 ;; Again the sign check is done first because we'd like abs(x^2) --> x^2.
1606		 ((and (mexptp x) ($featurep (caddr x) '$integer))
1607		  (power (take '(mabs) (cadr x)) (caddr x)))
1608
1609		 ;; Reflection rule: abs(-x) --> abs(x)
1610		 ((great (neg x) x) (take '(mabs) (neg x)))
1611
1612		 ;; nounform return
1613		 (t (eqtest (list '(mabs) x) e)))))))
1614
1615(defun abs-integral (x)
1616  (mul (div 1 2) x (take '(mabs) x)))
1617
1618(putprop 'mabs `((x) ,#'abs-integral) 'integral)
1619
1620;; I (rtoy) think this does some simple optimizations of x * y.
1621(defun testt (x)
1622  (cond ((mnump x)
1623	 x)
1624	((null (cddr x))
1625	 ;; We have something like ((mtimes) foo).  This is the same as foo.
1626	 (cadr x))
1627	((eql 1 (cadr x))
1628	 ;; We have 1*foo.  Which is the same as foo.  This should not
1629	 ;; be applied to 1.0 or 1b0!
1630	 (cond ((null (cdddr x))
1631		(caddr x))
1632	       (t (rplacd x (cddr x)))))
1633	(t
1634	 (testtneg x))))
1635
1636;; This basically converts -(a+b) to -a-b.
1637(defun testtneg (x)
1638  (cond ((and (equal (cadr x) -1)
1639	      (null (cdddr x))
1640	      (mplusp (caddr x))
1641	      $negdistrib)
1642	 ;; If x is exactly of the form -1*(sum), and $negdistrib is
1643	 ;; true, we distribute the -1 across the sum.
1644	 (addn (mapcar #'(lambda (z)
1645			   (mul2 -1 z))
1646		       (cdaddr x))
1647	       t))
1648	(t x)))
1649
1650;; Simplification of the "-" operator
1651(defun simpmin (x vestigial z)
1652  (declare (ignore vestigial))
1653  (cond ((null (cdr x)) 0)
1654        ((null (cddr x))
1655         (mul -1 (simplifya (cadr x) z)))
1656        (t
1657         ;; ((mminus) a b ...) -> ((mplus) a ((mtimes) -1 b) ...)
1658         (sub (simplifya (cadr x) z) (addn (cddr x) z)))))
1659
1660(defun simptimes (x w z)		; W must be 1
1661  (prog (res check eqnflag matrixflag sumflag)
1662     (if (null (cdr x)) (return 1))
1663     (setq check x)
1664  start
1665     (setq x (cdr x))
1666     (cond ((zerop1 res)
1667	    (cond ($mx0simp
1668		   (cond ((and matrixflag (mxorlistp1 matrixflag))
1669			  (return (constmx res matrixflag)))
1670			 (eqnflag (return (list '(mequal simp)
1671						(mul2 res (cadr eqnflag))
1672						(mul2 res (caddr eqnflag)))))
1673		         (t
1674		          (dolist (u x)
1675			    (cond ((mxorlistp u)
1676				   (return (setq res (constmx res u))))
1677				  ((and (mexptp u)
1678					(mxorlistp1 (cadr u))
1679					($numberp (caddr u)))
1680				   (return (setq res (constmx res (cadr u)))))
1681				  ((mequalp u)
1682				   (return
1683				     (setq res
1684				           (list '(mequal simp)
1685						 (mul2 res (cadr u))
1686						 (mul2 res (caddr u))))))))))))
1687	    (return res))
1688	   ((null x) (go end)))
1689     (setq w (if z (car x) (simplifya (car x) nil)))
1690  st1
1691     (cond ((atom w) nil)
1692	   ((eq (caar w) 'mrat)
1693	    (cond ((or eqnflag matrixflag
1694	               (and sumflag
1695	                    (not (member 'trunc (cdar w) :test #'eq)))
1696		       (spsimpcases (cdr x) w))
1697	           (setq w (ratdisrep w))
1698	           (go st1))
1699	          (t
1700	           (return
1701	             (ratf (cons '(mtimes)
1702			         (nconc (mapcar #'simplify (cons w (cdr x)))
1703					(cdr res))))))))
1704	   ((eq (caar w) 'mequal)
1705	    (setq eqnflag
1706		  (if (not eqnflag)
1707		      w
1708		      (list (car eqnflag)
1709			    (mul2 (cadr eqnflag) (cadr w))
1710			    (mul2 (caddr eqnflag) (caddr w)))))
1711	    (go start))
1712	   ((member (caar w) '(mlist $matrix) :test #'eq)
1713	    (setq matrixflag
1714		  (cond ((not matrixflag) w)
1715			((and (or $doallmxops $domxmxops $domxtimes)
1716			      (or (not (eq (caar w) 'mlist)) $listarith)
1717			      (not (eq *inv* '$detout)))
1718			 (stimex matrixflag w))
1719			(t (setq res (tms (copy-tree w) 1 (copy-tree res))) matrixflag)))
1720	    (go start))
1721	   ((and (eq (caar w) '%sum) $sumexpand)
1722	    (setq sumflag (sumtimes sumflag w))
1723	    (go start)))
1724     (setq res (tms (copy-tree w) 1 (copy-tree res)))
1725     (go start)
1726  end
1727     (cond ((mtimesp res) (setq res (testt res))))
1728     (cond (sumflag (setq res (cond ((or (null res) (equal res 1)) sumflag)
1729				    ((not (mtimesp res))
1730				     (list '(mtimes) res sumflag))
1731				    (t (nconc res (list sumflag)))))))
1732     (cond ((or (atom res)
1733		(not (member (caar res) '(mexpt mtimes) :test #'eq))
1734		(and (zerop $expop) (zerop $expon))
1735		expandflag))
1736	   ((eq (caar res) 'mtimes) (setq res (expandtimes res)))
1737	   ((and (mplusp (cadr res))
1738		 (fixnump (caddr res))
1739		 (not (or (> (caddr res) $expop)
1740			  (> (- (caddr res)) $expon))))
1741	    (setq res (expandexpt (cadr res) (caddr res)))))
1742     (cond (matrixflag
1743            (setq res
1744                  (cond ((null res) matrixflag)
1745                        ((and (or ($listp matrixflag)
1746                                  $doallmxops
1747			          (and $doscmxops
1748			               (not (member res '(-1 -1.0) :test #'equal)))
1749			          ;; RES should only be -1 here (not = 1)
1750                                  (and $domxmxops
1751                                       (member res '(-1 -1.0) :test #'equal)))
1752                              (or (not ($listp matrixflag)) $listarith))
1753                         (mxtimesc res matrixflag))
1754			(t (testt (tms matrixflag 1 (tms res 1 nil))))))))
1755     (if res (setq res (eqtest res check)))
1756     (return (cond (eqnflag
1757		    (if (null res) (setq res 1))
1758		    (list (car eqnflag)
1759			  (mul2 (cadr eqnflag) res)
1760			  (mul2 (caddr eqnflag) res)))
1761		   (t res)))))
1762
1763(defun spsimpcases (l e)
1764  (dolist (u l)
1765    (if (or (mbagp u) (and (not (atom u))
1766			   (eq (caar u) '%sum)
1767			   (not (member 'trunc (cdar e) :test #'eq))))
1768	(return t))))
1769
1770(defun mxtimesc (sc mx)
1771  (let (sign out)
1772    (and (mtimesp sc) (member (cadr sc) '(-1 -1.0) :test #'equal)
1773	 $doscmxops (not (or $doallmxops $domxmxops $domxtimes))
1774	 (setq sign (cadr sc)) (rplaca (cdr sc) nil))
1775    (setq out (let ((scp* (cond ((mtimesp sc) (partition-ns (cdr sc)))
1776                                ((not (scalar-or-constant-p sc $assumescalar))
1777                                 nil)
1778				(t sc))))
1779		(cond  ((null scp*) (list '(mtimes simp) sc mx))
1780		       ((and (not (atom scp*)) (null (car scp*)))
1781			(append '((mtimes)) (cadr scp*) (list mx)))
1782		       ((or (atom scp*) (and (null (cdr scp*))
1783					     (not (null (cdr sc)))
1784					     (setq scp* (cons '(mtimes) (car scp*))))
1785			    (not (mtimesp sc)))
1786			(simplifya (outermap1 'mtimes scp* mx) nil))
1787		       (t (append '((mtimes))
1788				  (list (simplifya
1789					 (outermap1 'mtimes
1790						    (cons '(mtimes) (car scp*)) mx)
1791					 t))
1792				  (cadr scp*))))))
1793    (cond (sign (if (mtimesp out)
1794		    (rplacd out (cons sign (cdr out)))
1795		    (list '(mtimes) sign out)))
1796	  ((mtimesp out) (testt out))
1797	  (t out))))
1798
1799(defun stimex (x y)
1800  (let (($doscmxops t) ($domxmxops t) ($listarith t))
1801    (simplify (fmapl1 'mtimes x y))))
1802
1803;;  TMS takes a simplified expression FACTOR and a cumulative
1804;;  PRODUCT as arguments and modifies the cumulative product so
1805;;  that the expression is now one of its factors.  The
1806;;  exception to this occurs when a tellsimp rule is triggered.
1807;;  The second argument is the POWER to which the expression is
1808;;  to be raised within the product.
1809
1810(defun tms (factor power product &aux tem)
1811  (let ((rulesw nil)
1812	(z nil))
1813    (when (mplusp product) (setq product (list '(mtimes simp) product)))
1814    (cond ((zerop1 factor)
1815	   (cond ((mnegp power)
1816		  (if errorsw
1817		      (throw 'errorsw t)
1818		      (merror (intl:gettext "Division by 0"))))
1819		 (t factor)))
1820	  ((and (null product)
1821		(or (and (mtimesp factor) (equal power 1))
1822		    (and (setq product (list '(mtimes) 1)) nil)))
1823	   (setq tem (append '((mtimes)) (if (mnump (cadr factor)) nil '(1))
1824			     (cdr factor) nil))
1825	   (if (= (length tem) 1)
1826	       (setq tem (copy-list tem))
1827	       tem))
1828	  ((mtimesp factor)
1829	   (do ((factor-list (cdr factor) (cdr factor-list)))
1830	       ((or (null factor-list) (zerop1 product))  product)
1831	     (setq z (timesin (car factor-list) (cdr product) power))
1832	     (when rulesw
1833	       (setq rulesw nil)
1834	       (setq product (tms-format-product z)))))
1835	  (t
1836	   (setq z (timesin factor (cdr product) power))
1837	   (if rulesw
1838	       (tms-format-product z)
1839	       product)))))
1840
1841(defun tms-format-product (x)
1842  (cond ((zerop1 x) x)
1843	((mnump x) (list '(mtimes) x))
1844	((not (mtimesp x)) (list '(mtimes) 1 x))
1845	((not (mnump (cadr x))) (cons '(mtimes) (cons 1 (cdr x))))
1846	(t x)))
1847
1848(defun plsk (x y)
1849  (cond ($ratsimpexpons (sratsimp (list '(mplus) x y)))
1850	((and (mnump x) (mnump y)) (addk x y))
1851	(t (add2 x y))))
1852
1853(defun mult (x y)
1854  (if (and (mnump x) (mnump y))
1855      (timesk x y)
1856      (mul2 x y)))
1857
1858(defun simp-limit (x vestigial z)
1859  (declare (ignore vestigial))
1860  (let ((l1 (length x))
1861	y)
1862    (unless (or (= l1 2) (= l1 4) (= l1 5))
1863      (merror (intl:gettext "limit: wrong number of arguments.")))
1864    (setq y (simpmap (cdr x) z))
1865    (cond ((and (= l1 5) (not (member (cadddr y) '($plus $minus) :test #'eq)))
1866           (merror (intl:gettext "limit: direction must be either 'plus' or 'minus': ~M") (cadddr y)))
1867	  ((mnump (cadr y))
1868	   (merror (intl:gettext "limit: variable must not be a number; found: ~M") (cadr y)))
1869	  ((equal (car y) 1)
1870	   1)
1871	  (t
1872	   (eqtest (cons '(%limit) y) x)))))
1873
1874(defun simpinteg (x vestigial z)
1875  (declare (ignore vestigial))
1876  (let ((l1 (length x))
1877	y)
1878    (unless (or (= l1 3) (= l1 5))
1879      (merror (intl:gettext "integrate: wrong number of arguments.")))
1880    (setq y (simpmap (cdr x) z))
1881    (cond ((mnump (cadr y))
1882	   (merror (intl:gettext "integrate: variable must not be a number; found: ~M") (cadr y)))
1883	  ((and (= l1 5) (alike1 (caddr y) (cadddr y)))
1884	   0)
1885          ((and (= l1 5)
1886                (free (setq z (sub (cadddr y) (caddr y))) '$%i)
1887                (eq ($sign z) '$neg))
1888	   (neg (simplifya (list '(%integrate) (car y) (cadr y) (cadddr y) (caddr y)) t)))
1889	  ((equal (car y) 1)
1890	   (if (= l1 3)
1891	       (cadr y)
1892	       (if (or (among '$inf z) (among '$minf z))
1893		   (infsimp z)
1894		   z)))
1895	  (t
1896	   (eqtest (cons '(%integrate) y) x)))))
1897
1898(defun simpbigfloat (x vestigial simp-flag)
1899  (declare (ignore vestigial simp-flag))
1900  (bigfloatm* x))
1901
1902;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1903
1904;;; Implementation of the Exp function.
1905
1906(defprop $exp %exp verb)
1907(defprop $exp %exp alias)
1908
1909(defprop %exp $exp noun)
1910(defprop %exp $exp reversealias)
1911
1912(defprop %exp simp-exp operators)
1913
1914(defmfun $exp (z)
1915  (simplify (list '(%exp) z)))
1916
1917;; Support a function for code,
1918;; which depends on an unsimplified noun form.
1919(defmfun $exp-form (z)
1920  (list '(mexpt) '$%e z))
1921
1922(defun simp-exp (x ignored z)
1923  (declare (ignore ignored))
1924  (oneargcheck x)
1925  (simplifya (list '(mexpt) '$%e (cadr x)) z))
1926
1927;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1928
1929(defun simplambda (x vestigial simp-flag)
1930  (declare (ignore vestigial simp-flag))
1931  ; Check for malformed lambda expressions.
1932  ; We verify that we have a valid list of parameters and a non-empty body.
1933  (let ((params (cadr x)))
1934    (unless ($listp params)
1935      (merror (intl:gettext "lambda: first argument must be a list; found: ~M") params))
1936    (do ((params (cdr params) (cdr params))
1937         (seen-params nil))
1938        ((null params))
1939      (when (mdeflistp params)
1940        (setq params (cdar params)))
1941      (let ((p (car params)))
1942        (unless (or (mdefparam p)
1943                    (and (op-equalp p 'mquote)
1944                         (mdefparam (cadr p))))
1945          (merror (intl:gettext "lambda: parameter must be a symbol and must not be a system constant; found: ~M") p))
1946        (setq p (mparam p))
1947        (when (member p seen-params :test #'eq)
1948          (merror (intl:gettext "lambda: ~M occurs more than once in the parameter list") p))
1949        (push p seen-params))))
1950  (when (null (cddr x))
1951    (merror (intl:gettext "lambda: no body present")))
1952  (cons '(lambda simp) (cdr x)))
1953
1954(defun simpmdef (x vestigial simp-flag)
1955  (declare (ignore vestigial simp-flag))
1956  (twoargcheck x)
1957  (cons '(mdefine simp) (cdr x)))
1958
1959(defun simpmap (e z)
1960  (mapcar #'(lambda (u) (simpcheck u z)) e))
1961
1962(defun infsimp (e)
1963  (let ((x ($expand e 1 1)))
1964    (cond ((or (not (free x '$ind)) (not (free x '$und))
1965	       (not (free x '$zeroa)) (not (free x '$zerob))
1966	       (not (free x '$infinity))
1967	       (mbagp x))
1968	   (infsimp2 x e))
1969	  ((and (free x '$inf) (free x '$minf)) x)
1970	  (t (infsimp1 x e)))))
1971
1972(defun infsimp1 (x e)
1973  (let ((minf-coef (coeff x '$minf 1))
1974	(inf-coef (coeff x '$inf 1)))
1975    (cond ((or (and (equal minf-coef 0)
1976		    (equal inf-coef 0))
1977	       (and (not (free minf-coef '$inf))
1978		    (not (free inf-coef '$minf)))
1979	       (let ((new-exp (sub (add2 (mul2 minf-coef '$minf)
1980					 (mul2 inf-coef '$inf))
1981				   x)))
1982		 (and (not (free new-exp '$inf))
1983		      (not (free new-exp '$minf)))))
1984	   (infsimp2 x e))
1985	  (t (let ((sign-minf-coef ($asksign minf-coef))
1986		   (sign-inf-coef ($asksign inf-coef)))
1987	       (cond ((or (and (eq sign-inf-coef '$zero)
1988			       (eq sign-minf-coef '$neg))
1989			  (and (eq sign-inf-coef '$pos)
1990			       (eq sign-minf-coef '$zero))
1991			  (and (eq sign-inf-coef '$pos)
1992			       (eq sign-minf-coef '$neg)))  '$inf)
1993		     ((or (and (eq sign-inf-coef '$zero)
1994			       (eq sign-minf-coef '$pos))
1995			  (and (eq sign-inf-coef '$neg)
1996			       (eq sign-minf-coef '$zero))
1997			  (and (eq sign-inf-coef '$neg)
1998			       (eq sign-minf-coef '$pos)))  '$minf)
1999		     ((or (and (eq sign-inf-coef '$pos)
2000			       (eq sign-minf-coef '$pos))
2001			  (and (eq sign-inf-coef '$neg)
2002			       (eq sign-minf-coef '$neg)))  '$und)))))))
2003
2004(defun infsimp2 (x e)
2005  (setq x ($limit x))
2006  (if (isinop x '%limit) e x))
2007
2008(defun simpderiv (x y z)
2009  (prog (flag w u)
2010     (cond ((not (even (length x)))
2011	    (cond ((and (cdr x) (null (cdddr x))) (nconc x '(1)))
2012		  (t (wna-err '%derivative)))))
2013     (setq w (cons '(%derivative) (simpmap (cdr x) z)))
2014     (setq y (cadr w))
2015     (do ((u (cddr w) (cddr u))) ((null u))
2016       (cond ((mnump (car u))
2017	      (merror (intl:gettext "diff: variable must not be a number; found: ~M") (car u)))))
2018     (cond ((or (zerop1 y)
2019		(and (or (mnump y) (and (atom y) (constant y)))
2020		     (or (null (cddr w))
2021			 (and (not (alike1 y (caddr w)))
2022			      (do ((u (cddr w) (cddr u))) ((null u))
2023			        (cond ((and (numberp (cadr u))
2024			                    (not (zerop (cadr u))))
2025				       (return t))))))))
2026	    (return 0))
2027	   ((and (not (atom y)) (eq (caar y) '%derivative) derivsimp)
2028	    (rplacd w (append (cdr y) (cddr w)))))
2029     (if (null (cddr w))
2030	 (return (if (null derivflag) (list '(%del simp) y) (deriv (cdr w)))))
2031     (setq u (cdr w))
2032  ztest
2033     (cond ((null u) (go next))
2034           ((zerop1 (caddr u)) (rplacd u (cdddr u)))
2035           (t (setq u (cddr u))))
2036     (go ztest)
2037  next
2038     (cond ((null (cddr w)) (return y))
2039           ((and (null (cddddr w))
2040                 (onep (cadddr w))
2041                 (alike1 (cadr w) (caddr w)))
2042            (return 1)))
2043  again
2044     (setq z (cddr w))
2045  sort
2046     (cond ((null (cddr z)) (go loop))
2047           ((alike1 (car z) (caddr z))
2048            (rplaca (cdddr z) (add2 (cadr z) (cadddr z)))
2049            (rplacd z (cdddr z)))
2050           ((great (car z) (caddr z))
2051            (let ((u1 (car z)) (u2 (cadr z)) (v1 (caddr z)) (v2 (cadddr z)))
2052              (setq flag t)
2053              (rplaca z v1)
2054              (rplacd z (cons v2 (cons u1 (cons u2 (cddddr z))))))))
2055     (cond ((setq z (cddr z)) (go sort)))
2056  loop
2057     (cond ((null flag) (return (cond ((null derivflag) (eqtest w x))
2058                                      (t (deriv (cdr w)))))))
2059     (setq flag nil)
2060     (go again)))
2061
2062(defun signum1 (x)
2063  (cond ((mnump x)
2064	 (setq x (num1 x)) (cond ((plusp x) 1) ((minusp x) -1) (t 0)))
2065	((atom x) 1)
2066	((mplusp x) (if expandp 1 (signum1 (car (last x)))))
2067	((mtimesp x) (if (mplusp (cadr x)) 1 (signum1 (cadr x))))
2068	(t 1)))
2069
2070(defprop %signum (mlist $matrix mequal) distribute_over)
2071
2072(defun simpsignum (e y z)
2073  (declare (ignore y))
2074  (oneargcheck e)
2075  (let ((x (simpcheck (second e) z)) (sgn))
2076
2077    (cond ((complex-number-p x #'mnump)
2078		    (if (complex-number-p x #'$ratnump) ;; nonfloat complex
2079		        (if (zerop1 x) 0 ($rectform (div x ($cabs x))))
2080		      (maxima::to (bigfloat::signum (bigfloat::to x)))))
2081
2082	  ;; idempotent: signum(signum(z)) = signum(z).
2083	  ((and (consp x) (consp (car x)) (eq '%signum (mop x))) x)
2084
2085	  (t
2086	   (setq sgn ($csign x))
2087	   (cond ((eq sgn '$neg) -1)
2088		 ((eq sgn '$zero) 0)
2089		 ((eq sgn '$pos) 1)
2090
2091		 ;; multiplicative: signum(ab) = signum(a) * signum(b).
2092		 ((mtimesp x)
2093		  (muln (mapcar #'(lambda (s) (take '(%signum) s)) (margs x)) t))
2094
2095		 ;; Reflection rule: signum(-x) --> -signum(x).
2096		 ((great (neg x) x) (neg (take '(%signum) (neg x))))
2097
2098		 ;; nounform return
2099		 (t (eqtest (list '(%signum) x) e)))))))
2100
2101(defun exptrl (r1 r2)
2102  (cond ((equal r2 1) r1)
2103        ((equal r2 1.0)
2104         (cond ((mnump r1) (addk 0.0 r1))
2105               ;; Do not simplify the type of the number away.
2106               (t (list '(mexpt simp) r1 1.0))))
2107        ((equal r2 bigfloatone)
2108         (cond ((mnump r1) ($bfloat r1))
2109               ;; Do not simplify the type of the number away.
2110               (t (list '(mexpt simp) r1 bigfloatone))))
2111	((zerop1 r1)
2112	 (cond ((or (zerop1 r2) (mnegp r2))
2113		(if (not errorsw)
2114		    (merror (intl:gettext "expt: undefined: ~M") (list '(mexpt) r1 r2))
2115		    (throw 'errorsw t)))
2116	       (t (zerores r1 r2))))
2117	((or (zerop1 r2) (onep1 r1))
2118	 (cond ((or ($bfloatp r1) ($bfloatp r2)) bigfloatone)
2119	       ((or (floatp r1) (floatp r2)) 1.0)
2120	       (t 1)))
2121	((or ($bfloatp r1) ($bfloatp r2)) ($bfloat (list '(mexpt) r1 r2)))
2122	((and (numberp r1) (integerp r2)) (exptb r1 r2))
2123	((and (numberp r1) (floatp r2) (equal r2 (float (floor r2))))
2124	 (exptb (float r1) (floor r2)))
2125	((or $numer (and (floatp r2) (or (plusp (num1 r1)) $numer_pbranch)))
2126	 (let (y  #+kcl(r1 r1) #+kcl(r2 r2))
2127	   (cond ((minusp (setq r1 (addk 0.0 r1)))
2128		  (cond ((or $numer_pbranch (eq $domain '$complex))
2129		         ;; for R1<0:
2130		         ;; R1^R2 = (-R1)^R2*cos(pi*R2) + i*(-R1)^R2*sin(pi*R2)
2131			 (setq r2 (addk 0.0 r2))
2132			 (setq y (exptrl (- r1) r2) r2 (* %pi-val r2))
2133			 (add2 (* y (cos r2))
2134			       (list '(mtimes simp) (* y (sin r2)) '$%i)))
2135			(t (setq y (let ($numer $float $keepfloat $ratprint)
2136				     (power -1 r2)))
2137			   (mul2 y (exptrl (- r1) r2)))))
2138	         ((equal (setq r2 (addk 0.0 r2)) (float (floor r2)))
2139	          (exptb r1 (floor r2)))
2140	         ((and (equal (setq y (* 2.0 r2)) (float (floor y)))
2141	               (not (equal r1 %e-val)))
2142		  (exptb (sqrt r1) (floor y)))
2143		 (t (exp (* r2 (log r1)))))))
2144	((floatp r2) (list '(mexpt simp) r1 r2))
2145	((integerp r2)
2146	 (cond ((minusp r2)
2147	        (exptrl (cond ((equal (abs (cadr r1)) 1)
2148	                       (* (cadr r1) (caddr r1)))
2149	                       ;; We set the simp flag at this place. This
2150	                       ;; changes nothing for an exponent r2 # -1.
2151	                       ;; exptrl is called again and does not look at
2152	                       ;; the simp flag. For the case r2 = -1 exptrl
2153	                       ;; is called with an exponent 1. For this case
2154	                       ;; the base is immediately returned. Now the
2155	                       ;; base has the correct simp flag. (DK 02/2010)
2156			      ((minusp (cadr r1))
2157			       (list '(rat simp) (- (caddr r1)) (- (cadr r1))))
2158			      (t (list '(rat simp) (caddr r1) (cadr r1))))
2159			(- r2)))
2160	       (t (list '(rat simp) (exptb (cadr r1) r2) (exptb (caddr r1) r2)))))
2161	((and (floatp r1) (alike1 r2 '((rat) 1 2)))
2162	 (if (minusp r1)
2163	     (list '(mtimes simp) (sqrt (- r1)) '$%i)
2164	     (sqrt r1)))
2165	((and (floatp r1) (alike1 r2 '((rat) -1 2)))
2166	 (if (minusp r1)
2167	     (list '(mtimes simp) (/ -1.0 (sqrt (- r1))) '$%i)
2168	     (/ (sqrt r1))))
2169	((floatp r1)
2170	 (if (plusp r1)
2171	     (exptrl r1 (fpcofrat r2))
2172	     (mul2 (exptrl -1 r2) ;; (-4.5)^(1/4) -> (4.5)^(1/4) * (-1)^(1/4)
2173		   (exptrl (- r1) r2))))
2174	(exptrlsw (list '(mexpt simp) r1 r2))
2175	(t
2176	 (let ((exptrlsw t))
2177	   (simptimes (list '(mtimes)
2178			    (exptrl r1 (truncate (cadr r2) (caddr r2)))
2179			    (let ((y (let ($keepfloat $ratprint)
2180				       (simpnrt r1 (caddr r2))))
2181				  (z (rem (cadr r2) (caddr r2))))
2182			      (if (mexptp y)
2183				  (list (car y) (cadr y) (mul2 (caddr y) z))
2184				  (power y z))))
2185		      1 t)))))
2186
2187(defun simpexpt (x y z)
2188  (prog (gr pot check res rulesw w mlpgr mlppot)
2189     (setq check x)
2190     (cond (z (setq gr (cadr x) pot (caddr x)) (go cont)))
2191     (twoargcheck x)
2192     (setq gr (simplifya (cadr x) nil))
2193     (setq pot
2194           (let (($%enumer $numer))
2195             ;; Switch $%enumer on, when $numer is TRUE to allow
2196             ;; simplification of $%e to its numerical value.
2197             (simplifya (if $ratsimpexpons ($ratsimp (caddr x)) (caddr x))
2198                        nil)))
2199  cont
2200     (cond (($ratp pot)
2201            (setq pot (ratdisrep pot))
2202            (go cont))
2203           (($ratp gr)
2204            (cond ((member 'trunc (car gr) :test #'eq)
2205                   (return (srf (list '(mexpt) gr pot))))
2206                  ((integerp pot)
2207                   (let ((varlist (caddar gr)) (genvar (cadddr (car gr))))
2208                     (return (ratrep* (list '(mexpt) gr pot)))))
2209                  (t
2210                   (setq gr (ratdisrep gr))
2211                   (go cont))))
2212           ((or (setq mlpgr (mxorlistp gr))
2213                (setq mlppot (mxorlistp pot)))
2214            (go matrix))
2215           ((onep1 pot) (go atgr))
2216           ((or (zerop1 pot) (onep1 gr)) (go retno))
2217
2218           ;; This code tries to handle 0^a more complete.
2219           ;; If the sign of realpart(a) is not known return an unsimplified
2220           ;; expression. The handling of the flag *zexptsimp? is not changed.
2221           ;; Reverting the return of an unsimplified 0^a, because timesin
2222           ;; can not handle such expressions. (DK 02/2010)
2223           ((zerop1 gr)
2224            (cond ((or (member (setq z ($csign pot)) '($neg $nz))
2225                       (and *zexptsimp? (eq ($asksign pot) '$neg)))
2226                   ;; A negative exponent. Maxima error.
2227                   (cond ((not errorsw) (merror (intl:gettext "expt: undefined: 0 to a negative exponent.")))
2228                         (t (throw 'errorsw t))))
2229                  ((and (member z '($complex $imaginary))
2230                        ;; A complex exponent. Look at the sign of the realpart.
2231                        (member (setq z ($sign ($realpart pot)))
2232                                '($neg $nz $zero)))
2233                   (cond ((not errorsw)
2234                          (merror (intl:gettext "expt: undefined: 0 to a complex exponent.")))
2235                         (t (throw 'errorsw t))))
2236                  ((and *zexptsimp? (eq ($asksign pot) '$zero))
2237                   (cond ((not errorsw)
2238                          (merror (intl:gettext "expt: undefined: 0^0")))
2239                         (t (throw 'errorsw t))))
2240                  ((not (member z '($pos $pz)))
2241                   ;; The sign of realpart(pot) is not known. We can not return
2242                   ;; an unsimplified 0^a expression, because timesin can not
2243                   ;; handle it. We return ZERO. That is the old behavior.
2244                   ;; Look for the imaginary symbol to be consistent with
2245                   ;; old code.
2246                   (cond ((not (free pot '$%i))
2247                          (cond ((not errorsw)
2248                                 (merror (intl:gettext "expt: undefined: 0 to a complex exponent.")))
2249                                (t (throw 'errorsw t))))
2250                         (t
2251                          ;; Return ZERO and not an unsimplified expression.
2252                          (return (zerores gr pot)))))
2253                  (t (return (zerores gr pot)))))
2254
2255           ((and (mnump gr)
2256                 (mnump pot)
2257                 (or (not (ratnump gr)) (not (ratnump pot))))
2258            (return (eqtest (exptrl gr pot) check)))
2259           ;; Check for numerical evaluation of the sqrt.
2260           ((and (alike1 pot '((rat) 1 2))
2261                 (or (setq res (flonum-eval '%sqrt gr))
2262                     (and (not (member 'simp (car x) :test #'eq))
2263                          (setq res (big-float-eval '%sqrt gr)))))
2264            (return res))
2265           ((eq gr '$%i)
2266            (return (%itopot pot)))
2267           ((and (realp gr) (minusp gr) (mevenp pot))
2268            (setq gr (- gr))
2269            (go cont))
2270           ((and (realp gr) (minusp gr) (moddp pot))
2271            (return (mul2 -1 (power (- gr) pot))))
2272           ((and (equal gr -1) (maxima-integerp pot) (mminusp pot))
2273            (setq pot (neg pot))
2274            (go cont))
2275           ((and (equal gr -1)
2276                 (maxima-integerp pot)
2277                 (mtimesp pot)
2278                 (= (length pot) 3)
2279                 (fixnump (cadr pot))
2280                 (oddp (cadr pot))
2281                 (maxima-integerp (caddr pot)))
2282            (setq pot (caddr pot))
2283            (go cont))
2284           ((atom gr) (go atgr))
2285           ((and (eq (caar gr) 'mabs)
2286                 (or (evnump pot)
2287                     (mevenp pot))
2288                 (or (and (eq $domain '$real) (not (apparently-complex-to-judge-by-$csign-p (cadr gr))))
2289                     (and (eq $domain '$complex) (apparently-real-to-judge-by-$csign-p (cadr gr)))))
2290            (return (power (cadr gr) pot)))
2291           ((and (eq (caar gr) 'mabs)
2292                 (integerp pot)
2293                 (oddp pot)
2294                 (not (equal pot -1))
2295                 (or (and (eq $domain '$real) (not (apparently-complex-to-judge-by-$csign-p (cadr gr))))
2296                     (and (eq $domain '$complex) (apparently-real-to-judge-by-$csign-p (cadr gr)))))
2297            ;; abs(x)^(2*n+1) -> abs(x)*x^(2*n), n an integer number
2298            (if (plusp pot)
2299                (return (mul (power (cadr gr) (add pot -1))
2300                             gr))
2301                (return (mul (power (cadr gr) (add pot 1))
2302                             (inv gr)))))
2303           ((eq (caar gr) 'mequal)
2304            (return (eqtest (list (ncons (caar gr))
2305                                  (power (cadr gr) pot)
2306                                  (power (caddr gr) pot))
2307                            gr)))
2308           ((symbolp pot) (go opp))
2309           ((eq (caar gr) 'mexpt) (go e1))
2310           ((and (eq (caar gr) '%sum)
2311                 $sumexpand
2312                 (integerp pot)
2313                 (signp g pot)
2314                 (< pot $maxposex))
2315            (return (do ((i (1- pot) (1- i))
2316                         (an gr (simptimes (list '(mtimes) an gr) 1 t)))
2317                        ((signp e i) an))))
2318           ((equal pot -1)
2319            (return (eqtest (testt (tms gr pot nil)) check)))
2320           ((fixnump pot)
2321            (return (eqtest (cond ((and (mplusp gr)
2322                                        (not (or (> pot $expop)
2323                                                 (> (- pot) $expon))))
2324                                   (expandexpt gr pot))
2325                                  (t (simplifya (tms gr pot nil) t)))
2326                            check))))
2327
2328  opp
2329     (cond ((eq (caar gr) 'mexpt) (go e1))
2330           ((eq (caar gr) 'rat)
2331            (return (mul2 (power (cadr gr) pot)
2332                          (power (caddr gr) (mul2 -1 pot)))))
2333           ((not (eq (caar gr) 'mtimes)) (go up))
2334           ((or (eq $radexpand '$all) (and $radexpand (simplexpon pot)))
2335            (setq res (list 1))
2336            (go start))
2337           ((and (or (not (numberp (cadr gr)))
2338                     (equal (cadr gr) -1))
2339                 (equal -1 ($num gr)) ; only for -1
2340                 ;; Do not simplify for a complex base.
2341                 (not (member ($csign gr) '($complex $imaginary)))
2342                 (and (eq $domain '$real) $radexpand))
2343            ;; (-1/x)^a -> 1/(-x)^a for x negative
2344            ;; For all other cases (-1)^a/x^a
2345            (if (eq ($csign (setq w ($denom gr))) '$neg)
2346                (return (inv (power (neg w) pot)))
2347                (return (div (power -1 pot)
2348                             (power w pot)))))
2349           ((or (eq $domain '$complex) (not $radexpand)) (go up)))
2350     (return (do ((l (cdr gr) (cdr l)) (res (ncons 1)) (rad))
2351                 ((null l)
2352                  (cond ((equal res '(1))
2353                         (eqtest (list '(mexpt) gr pot) check))
2354                        ((null rad)
2355                         (testt (cons '(mtimes simp) res)))
2356                        (t
2357                         (setq rad (power* ; RADEXPAND=()?
2358                                     (cons '(mtimes) (nreverse rad)) pot))
2359                         (cond ((not (onep1 rad))
2360                                (setq rad
2361                                      (testt (tms rad 1 (cons '(mtimes) res))))
2362                                (cond (rulesw
2363                                       (setq rulesw nil res (cdr rad))))))
2364                         (eqtest (testt (cons '(mtimes) res)) check))))
2365               ;; Check with $csign to be more complete. This prevents wrong
2366               ;; simplifications like sqrt(-z^2)->%i*sqrt(z^2) for z complex.
2367               (setq z ($csign (car l)))
2368               (if (member z '($complex $imaginary))
2369                   (setq z '$pnz)) ; if appears complex, unknown sign
2370               (setq w (cond ((member z '($neg $nz) :test #'eq)
2371                              (setq rad (cons -1 rad))
2372                              (mult -1 (car l)))
2373                             (t (car l))))
2374               (cond ((onep1 w))
2375                     ((alike1 w gr) (return (list '(mexpt simp) gr pot)))
2376                     ((member z '($pn $pnz) :test #'eq)
2377                      (setq rad (cons w rad)))
2378                     (t
2379                      (setq w (testt (tms (simplifya (list '(mexpt) w pot) t)
2380                                          1 (cons '(mtimes) res))))))
2381               (cond (rulesw (setq rulesw nil res (cdr w))))))
2382
2383  start
2384     (cond ((and (cdr res) (onep1 (car res)) (ratnump (cadr res)))
2385            (setq res (cdr res))))
2386     (cond ((null (setq gr (cdr gr)))
2387            (return (eqtest (testt (cons '(mtimes) res)) check)))
2388           ((mexptp (car gr))
2389            (setq y (power (cadar gr) (mult (caddar gr) pot))))
2390           ((eq (car gr) '$%i)
2391            (setq y (%itopot pot)))
2392           ((mnump (car gr))
2393            (setq y (list '(mexpt) (car gr) pot)))
2394           (t (setq y (list '(mexpt simp) (car gr) pot))))
2395     (setq w (testt (tms (simplifya y t) 1 (cons '(mtimes) res))))
2396     (cond (rulesw (setq rulesw nil res (cdr w))))
2397     (go start)
2398
2399  retno
2400     (return (exptrl gr pot))
2401
2402  atgr
2403     (cond ((zerop1 pot) (go retno))
2404           ((onep1 pot)
2405            (let ((y (mget gr '$numer)))
2406              (if (and y (floatp y) (or $numer (not (equal pot 1))))
2407                  ;; A numeric constant like %e, %pi, ... and
2408                  ;; exponent is a float or bigfloat value.
2409                  (return (if (and (member gr *builtin-numeric-constants*)
2410                                   (equal pot bigfloatone))
2411                              ;; Return a bigfloat value.
2412                              ($bfloat gr)
2413                              ;; Return a float value.
2414                              y))
2415                  ;; In all other cases exptrl simplifies accordingly.
2416                  (return (exptrl gr pot)))))
2417           ((eq gr '$%e)
2418            ;; Numerically evaluate if the power is a flonum.
2419            (when $%emode
2420              (let ((val (flonum-eval '%exp pot)))
2421		(if (float-inf-p val)
2422		    ;; needed for gcl - no trap of overflow
2423		    (signal 'floating-point-overflow))
2424                (when val
2425                  (return val)))
2426              ;; Numerically evaluate if the power is a (complex)
2427              ;; big-float.  (This is basically the guts of
2428              ;; big-float-eval, but we can't use big-float-eval.)
2429              (when (and (not (member 'simp (car x) :test #'eq))
2430                         (complex-number-p pot 'bigfloat-or-number-p))
2431                (let ((x ($realpart pot))
2432                      (y ($imagpart pot)))
2433                  (cond ((and ($bfloatp x) (like 0 y))
2434                         (return ($bfloat `((mexpt simp) $%e ,pot))))
2435                        ((or ($bfloatp x) ($bfloatp y))
2436                         (let ((z (add ($bfloat x) (mul '$%i ($bfloat y)))))
2437                           (setq z ($rectform `((mexpt simp) $%e ,z)))
2438                           (return ($bfloat z))))))))
2439            (cond ((and $logsimp (among '%log pot)) (return (%etolog pot)))
2440                  ((and $demoivre (setq z (demoivre pot))) (return z))
2441                  ((and $%emode
2442                        (among '$%i pot)
2443                        (among '$%pi pot)
2444                        ;; Exponent contains %i and %pi and %emode is TRUE:
2445                        ;; Check simplification of exp(%i*%pi*p/q*x)
2446                        (setq z (%especial pot)))
2447                   (return z))
2448                  (($taylorp (third x))
2449                   ;; taylorize %e^taylor(...)
2450                   (return ($taylor x)))))
2451           (t
2452            (let ((y (mget gr '$numer)))
2453              ;; Check for a numeric constant.
2454              (and y
2455                   (floatp y)
2456                   (or (floatp pot)
2457                       ;; The exponent is a bigfloat. Convert base to bigfloat.
2458                       (and ($bfloatp pot)
2459                            (member gr *builtin-numeric-constants*)
2460                            (setq y ($bfloat gr)))
2461                       (and $numer (integerp pot)))
2462                   (return (exptrl y pot))))))
2463
2464  up
2465     (return (eqtest (list '(mexpt) gr pot) check))
2466
2467  matrix
2468     (cond ((zerop1 pot)
2469            (cond ((mxorlistp1 gr) (return (constmx (addk 1 pot) gr)))
2470                  (t (go retno))))
2471           ((onep1 pot) (return gr))
2472           ((or $doallmxops $doscmxops $domxexpt)
2473            (cond ((or (and mlpgr
2474                            (or (not ($listp gr)) $listarith)
2475                            (scalar-or-constant-p pot $assumescalar))
2476                       (and $domxexpt
2477                            mlppot
2478                            (or (not ($listp pot)) $listarith)
2479                            (scalar-or-constant-p gr $assumescalar)))
2480                   (return (simplifya (outermap1 'mexpt gr pot) t)))
2481                  (t (go up))))
2482           ((and $domxmxops (member pot '(-1 -1.0) :test #'equal))
2483            (return (simplifya (outermap1 'mexpt gr pot) t)))
2484           (t (go up)))
2485  e1
2486     ;; At this point we have an expression: (z^a)^b with gr = z^a and pot = b
2487     (cond ((or (eq $radexpand '$all)
2488                ;; b is an integer or an odd rational
2489                (simplexpon pot)
2490                (and (eq $domain '$complex)
2491                     (not (member ($csign (caddr gr)) '($complex $imaginary)))
2492                         ;; z >= 0 and a not a complex
2493                     (or (member ($csign (cadr gr)) '($pos $pz $zero))
2494                         ;; -1 < a <= 1
2495                         (and (mnump (caddr gr))
2496                              (eq ($sign (sub 1 (take '(mabs) (caddr gr))))
2497                                  '$pos))))
2498                (and (eq $domain '$real)
2499                     (member ($csign (cadr gr)) '($pos $pz $zero)))
2500                ;; (1/z)^a -> 1/z^a when z a constant complex
2501                (and (eql (caddr gr) -1)
2502                     (or (and $radexpand
2503                              (eq $domain '$real))
2504                         (and (eq ($csign (cadr gr)) '$complex)
2505                              ($constantp (cadr gr)))))
2506                ;; This does (1/z)^a -> 1/z^a. This is in general wrong.
2507                ;; We switch this type of simplification on, when
2508                ;; $ratsimpexpons is T. E.g. radcan sets this flag to T.
2509                ;; radcan hangs for expressions like sqrt(1/(1+x)) without
2510                ;; this simplification.
2511                (and $ratsimpexpons
2512                     (equal (caddr gr) -1))
2513                (and $radexpand
2514                     (eq $domain '$real)
2515                     (odnump (caddr gr))))
2516            ;; Simplify (z^a)^b -> z^(a*b)
2517            (setq pot (mul pot (caddr gr))
2518                  gr (cadr gr)))
2519           ((and (eq $domain '$real)
2520                 (free gr '$%i)
2521                 $radexpand
2522                 (not (apparently-complex-to-judge-by-$csign-p (cadr gr)))
2523                 (evnump (caddr gr)))
2524            ;; Simplify (x^a)^b -> abs(x)^(a*b)
2525            (setq pot (mul pot (caddr gr))
2526                  gr (radmabs (cadr gr))))
2527           ((and $radexpand
2528                 (eq $domain '$real)
2529                 (mminusp (caddr gr)))
2530            ;; Simplify (1/z^a)^b -> 1/(z^a)^b
2531            (setq pot (neg pot)
2532                  gr (power (cadr gr) (neg (caddr gr)))))
2533           (t (go up)))
2534     (go cont)))
2535
2536(defun apparently-complex-to-judge-by-$csign-p (e)
2537  (let ((s ($csign e)))
2538    (member s '($complex $imaginary))))
2539
2540(defun apparently-real-to-judge-by-$csign-p (e)
2541  (let ((s ($csign e)))
2542    (member s '($pos $neg $zero $pn $pnz $pz $nz))))
2543
2544;; Basically computes log of m base b.  Except if m is not a power
2545;; of b, we return nil.  m is a positive integer and base an integer
2546;; not equal to +/-1.
2547(defun exponent-of (m base)
2548  ;; Just compute base^k until base^k >= m.  Then check if they're equal.
2549  ;; If so, we have the exponent.  Otherwise, give up.
2550  (let ((expo 0))
2551    (loop
2552      (multiple-value-bind (q r)
2553          (floor m base)
2554        (cond ((zerop r)
2555               (setf m q)
2556               (incf expo))
2557              (t (return nil)))))
2558    (if (zerop expo) nil expo)))
2559
2560(defun timesin (x y w)                  ; Multiply X^W into Y
2561  (prog (fm temp z check u expo)
2562     (if (mexptp x) (setq check x))
2563  top
2564     ;; Prepare the factor x^w and initialize the work of timesin
2565     (cond ((equal w 1)
2566            (setq temp x))
2567           (t
2568            (setq temp (cons '(mexpt) (if check
2569                                          (list (cadr x) (mult (caddr x) w))
2570                                          (list x w))))
2571            (if (and (not timesinp) (not (eq x '$%i)))
2572                (let ((timesinp t))
2573                  (setq temp (simplifya temp t))))))
2574     (setq x (if (mexptp temp)
2575                 (cdr temp)
2576                 (list temp 1)))
2577     (setq w (cadr x)
2578           fm y)
2579  start
2580     ;; Go through the list of terms in fm and look what is to do.
2581     (cond ((null (cdr fm))
2582            ;; The list of terms is empty. The loop is finshed.
2583            (go less))
2584           ((or (and (mnump temp)
2585                     (not (or (integerp temp)
2586                              (ratnump temp))))
2587                (and (integerp temp)
2588                     (equal temp -1)))
2589            ;; Stop the loop for a float or bigfloat number, or number -1.
2590            (go less))
2591           ((mexptp (cadr fm))
2592            (cond ((alike1 (car x) (cadadr fm))
2593                   (cond ((zerop1 (setq w (plsk (caddr (cadr fm)) w)))
2594                          (go del))
2595                         ((and (mnump w)
2596                               (or (mnump (car x))
2597                                   (eq (car x) '$%i)))
2598                          (rplacd fm (cddr fm))
2599                          (cond ((mnump (setq x (if (mnump (car x))
2600                                                    (exptrl (car x) w)
2601                                                    (power (car x) w))))
2602                                 (return (rplaca y (timesk (car y) x))))
2603                                ((mtimesp x)
2604                                 (go times))
2605                                (t
2606                                 (setq temp x
2607                                       x (if (mexptp x) (cdr x) (list x 1)))
2608                                 (setq w (cadr x)
2609                                       fm y)
2610                                 (go start))))
2611                         ((maxima-constantp (car x))
2612                          (go const))
2613                         ((onep1 w)
2614                          (cond ((mtimesp (car x))
2615                                 ;; A base which is a mtimes expression. Remove
2616                                 ;; the factor from the lists of products.
2617                                 (rplacd fm (cddr fm))
2618                                 ;; Multiply the factors of the base with
2619                                 ;; the list of all remaining products.
2620                                 (setq rulesw t)
2621                                 (return (muln (nconc y (cdar x)) t)))
2622                                (t (return (rplaca (cdr fm) (car x))))))
2623                         (t
2624                          (go spcheck))))
2625                  ;; At this place we have to add code for a rational number
2626                  ;; as a factor to the list of products.
2627                  ((and (onep1 w)
2628                        (or (ratnump (car x))
2629                            (and (integerp (car x))
2630                                 (not (onep (car x))))))
2631                   ;; Multiplying bas^k * num/den
2632                   (let ((num (num1 (car x)))
2633                         (den (denom1 (car x)))
2634                         (bas (second (cadr fm))))
2635                     (cond ((and (integerp bas)
2636                                 (not (eql 1 (abs bas)))
2637                                 (setq expo (exponent-of (abs num) bas)))
2638                            ;; We have bas^m*bas^k = bas^(k+m).
2639                            (setq temp (power bas
2640                                              (add (third (cadr fm)) expo)))
2641                            ;; Set fm to have 1/denom term.
2642                            (setq x (mul (car y)
2643                                         (div (div num
2644                                                   (exptrl bas expo))
2645                                              den))))
2646                           ((and (integerp bas)
2647                                 (not (eql 1 (abs bas)))
2648                                 (setq expo (exponent-of den bas)))
2649                            (setq expo (- expo))
2650                            ;; We have bas^(-m)*bas^k = bas^(k-m).
2651                            (setq temp (power bas
2652                                              (add (third (cadr fm)) expo)))
2653                            ;; Set fm to have the numerator term.
2654                            (setq x (mul (car y)
2655                                         (div num
2656                                              (div den
2657                                                   (exptrl bas (- expo)))))))
2658                           (t
2659                            ;; Next term in list of products.
2660                            (setq fm (cdr fm))
2661                            (go start)))
2662                     ;; Add in the bas^(k+m) term or bas^(k-m)
2663                     (setf y (rplaca y 1))
2664                     (rplacd fm (cddr fm))
2665                     (rplacd fm (cons temp (cdr fm)))
2666                     (setq temp x
2667                           x (list x 1)
2668                           w 1
2669                           fm y)
2670                     (go start)))
2671                  ((and (not (atom (car x)))
2672                        (eq (caar (car x)) 'mabs)
2673                        (equal (cadr x) 1)
2674                        (integerp (caddr (cadr fm)))
2675                        (< (caddr (cadr fm)) -1)
2676                        (alike1 (cadr (car x)) (cadr (cadr fm)))
2677                        (not (member ($csign (cadr (car x)))
2678                                     '($complex imaginary))))
2679                   ;; 1/x^n*abs(x) -> 1/(x^(n-2)*abs(x)), where n an integer
2680                   ;; Replace 1/x^n -> 1/x^(n-2)
2681                   (setq temp (power (cadr (cadr fm))
2682                                     (add (caddr (cadr fm)) 2)))
2683                   (rplacd fm (cddr fm))
2684                   (if (not (equal temp 1))
2685                       (rplacd fm (cons temp (cdr fm))))
2686                   ;; Multiply factor 1/abs(x) into list of products.
2687                   (setq x (list (car x) -1))
2688                   (setq temp (power (car x) (cadr x)))
2689                   (setq w (cadr x))
2690                   (go start))
2691
2692                  ((and (not (atom (car x)))
2693                        (eq (caar (car x)) 'mabs)
2694                        (equal (cadr x) -1)
2695                        (integerp (caddr (cadr fm)))
2696                        (> (caddr (cadr fm)) 1)
2697                        (alike1 (cadr (car x)) (cadr (cadr fm)))
2698                        (not (member ($csign (cadr (car x)))
2699                                     '($complex imaginary))))
2700                   ;; x^n/abs(x) -> x^(n-2)*abs(x), where n an integer.
2701                   ;; Replace x^n -> x^(n-2)
2702                   (setq temp (power (cadr (cadr fm))
2703                                     (add (caddr (cadr fm)) -2)))
2704                   (rplacd fm (cddr fm))
2705                   (if (not (equal temp 1))
2706                       (rplacd fm (cons temp (cdr fm))))
2707                   ;; Multiply factor abs(x) into list of products.
2708                   (setq x (list (car x) 1))
2709                   (setq temp (power (car x) (cadr x)))
2710                   (setq w (cadr x))
2711                   (go start))
2712
2713                  ((and (not (atom (cadr fm)))
2714                        (not (atom (cadr (cadr fm))))
2715                        (eq (caaadr (cadr fm)) 'mabs)
2716                        (equal (caddr (cadr fm)) -1)
2717                        (integerp (cadr x))
2718                        (> (cadr x) 1)
2719                        (alike1 (cadadr (cadr fm)) (car x))
2720                        (not (member ($csign (cadadr (cadr fm)))
2721                                     '($complex imaginary))))
2722                   ;; 1/abs(x)*x^n -> x^(n-2)*abs(x), where n an integer.
2723                   ;; Replace 1/abs(x) -> abs(x)
2724                   (setq temp (cadr (cadr fm)))
2725                   (rplacd fm (cddr fm))
2726                   (rplacd fm (cons temp (cdr fm)))
2727                   ;; Multiply factor x^(n-2) into list of products.
2728                   (setq x (list (car x) (add (cadr x) -2)))
2729                   (setq temp (power (car x) (cadr x)))
2730                   (setq w (cadr x))
2731                   (go start))
2732
2733                  ((or (maxima-constantp (car x))
2734                       (maxima-constantp (cadadr fm)))
2735                   (if (great temp (cadr fm))
2736                       (go gr)))
2737                  ((great (car x) (cadadr fm))
2738                   (go gr)))
2739            (go less))
2740           ((alike1 (car x) (cadr fm))
2741            (go equ))
2742          ((mnump temp)
2743           ;; When a number goto start and look in the next term.
2744           (setq fm (cdr fm))
2745           (go start))
2746
2747           ((and (not (atom (cadr fm)))
2748                 (eq (caar (cadr fm)) 'mabs)
2749                 (integerp (cadr x))
2750                 (< (cadr x) -1)
2751                 (alike1 (cadr (cadr fm)) (car x))
2752                 (not (member ($csign (cadr (cadr fm)))
2753                                     '($complex imaginary))))
2754            ;; abs(x)/x^n -> 1/(x^(n-2)*abs(x)), where n an integer.
2755            ;; Replace abs(x) -> 1/abs(x).
2756            (setq temp (power (cadr fm) -1))
2757            (rplacd fm (cddr fm))
2758            (rplacd fm (cons temp (cdr fm)))
2759            ;; Multiply factor x^(-n+2) into list of products.
2760            (setq x (list (car x) (add (cadr x) 2)))
2761            (setq temp (power (car x) (cadr x)))
2762            (setq w (cadr x))
2763            (go start))
2764
2765           ((maxima-constantp (car x))
2766            (when (great temp (cadr fm))
2767              (go gr)))
2768           ((great (car x) (cadr fm))
2769            (go gr)))
2770  less
2771     (cond ((mnump temp)
2772           ;; Multiply a number into the list of products.
2773           (return (rplaca y (timesk (car y) temp))))
2774           ((and (eq (car x) '$%i)
2775                 (fixnump w))
2776            (go %i))
2777           ((and (eq (car x) '$%e)
2778                 $numer
2779                 (integerp w))
2780            (return (rplaca y (timesk (car y) (exp (float w))))))
2781           ((and (onep1 w)
2782                 (not (constant (car x))))
2783            (go less1))
2784           ;; At this point we will insert a mexpt expression,
2785           ;; but first we look at the car of the list of products and
2786           ;; modify the expression if we found a rational number.
2787           ((and (mexptp temp)
2788                 (not (onep1 (car y)))
2789                 (or (integerp (car y))
2790                     (ratnump (car y))))
2791            ;; Multiplying bas^k * num/den.
2792            (let ((num (num1 (car y)))
2793                  (den (denom1 (car y)))
2794                  (bas (car x)))
2795              (cond ((and (integerp bas)
2796                          (not (eql 1 (abs bas)))
2797                          (setq expo (exponent-of (abs num) bas)))
2798                     ;; We have bas^m*bas^k.
2799                     (setq temp (power bas (add (cadr x) expo)))
2800                     ;; Set fm to have 1/denom term.
2801                     (setq x (div (div num (exptrl bas expo)) den)))
2802                    ((and (integerp bas)
2803                          (not (eql 1 (abs bas)))
2804                          (setq expo (exponent-of den bas)))
2805                     (setq expo (- expo))
2806                     ;; We have bas^(-m)*bas^k.
2807                     (setq temp (power bas (add (cadr x) expo)))
2808                     ;; Set fm to have the numerator term.
2809                     (setq x (div num (div den (exptrl bas (- expo))))))
2810                    (t
2811                     ;; The rational doesn't contain any (simple) powers of
2812                     ;; the exponential term.  We're done.
2813                     (return (cdr (rplacd fm (cons temp (cdr fm)))))))
2814              ;; Add in the a^(m+k) or a^(k-m) term.
2815              (setf y (rplaca y 1))
2816              (rplacd fm (cons temp (cdr fm)))
2817              (setq temp x
2818                    x (list x 1)
2819                    w 1
2820                    fm y)
2821              (go start)))
2822           ((and (maxima-constantp (car x))
2823                 (do ((l (cdr fm) (cdr l)))
2824                     ((null (cdr l)))
2825                   (when (and (mexptp (cadr l))
2826                              (alike1 (car x) (cadadr l)))
2827                     (setq fm l)
2828                     (return t))))
2829            (go start))
2830           ((or (and (mnump (car x))
2831                     (mnump w))
2832                (and (eq (car x) '$%e)
2833                     $%emode
2834                     (among '$%i w)
2835                     (among '$%pi w)
2836                     (setq u (%especial w))))
2837            (setq x (cond (u)
2838                          ((alike (cdr check) x)
2839                           check)
2840                          (t
2841                           (exptrl (car x) w))))
2842            (cond ((mnump x)
2843                   (return (rplaca y (timesk (car y) x))))
2844                  ((mtimesp x)
2845                   (go times))
2846                  ((mexptp x)
2847                   (return (cdr (rplacd fm (cons x (cdr fm))))))
2848                  (t
2849                   (setq temp x
2850                         x (list x 1)
2851                         w 1
2852                         fm y)
2853                   (go start))))
2854           ((onep1 w)
2855            (go less1))
2856           (t
2857            (setq temp (list '(mexpt) (car x) w))
2858            (setq temp (eqtest temp (or check '((foo)))))
2859            (return (cdr (rplacd fm (cons temp (cdr fm)))))))
2860  less1
2861     (return (cdr (rplacd fm (cons (car x) (cdr fm)))))
2862  gr
2863     (setq fm (cdr fm))
2864     (go start)
2865  equ
2866     (cond ((and (eq (car x) '$%i) (equal w 1))
2867            (rplacd fm (cddr fm))
2868            (return (rplaca y (timesk -1 (car y)))))
2869           ((zerop1 (setq w (plsk 1 w)))
2870            (go del))
2871           ((and (mnump (car x)) (mnump w))
2872            (return (rplaca (cdr fm) (exptrl (car x) w))))
2873           ((maxima-constantp (car x))
2874            (go const)))
2875  spcheck
2876     (setq z (list '(mexpt) (car x) w))
2877     (cond ((alike1 (setq x (simplifya z t)) z)
2878            (return (rplaca (cdr fm) x)))
2879           (t
2880            (rplacd fm (cddr fm))
2881            (setq rulesw t)
2882            (return (muln (cons x y) t))))
2883  const
2884     (rplacd fm (cddr fm))
2885     (setq x (car x) check nil)
2886     (go top)
2887  times
2888     (setq z (tms x 1 (setq temp (cons '(mtimes) y))))
2889     (return (cond ((eq z temp)
2890                    (cdr z))
2891                   (t
2892                    (setq rulesw t) z)))
2893  del
2894     (return (rplacd fm (cddr fm)))
2895  %i
2896     (if (minusp (setq w (rem w 4)))
2897         (incf w 4))
2898     (return (cond ((zerop w)
2899                    fm)
2900                   ((= w 2)
2901                    (rplaca y (timesk -1 (car y))))
2902                   ((= w 3)
2903                    (rplaca y (timesk -1 (car y)))
2904                    (rplacd fm (cons '$%i (cdr fm))))
2905                   (t
2906                    (rplacd fm (cons '$%i (cdr fm))))))))
2907
2908(defun simpmatrix (x vestigial z)
2909  (declare (ignore vestigial))
2910  (if (and (null (cddr x))
2911	   $scalarmatrixp
2912	   (or (eq $scalarmatrixp '$all) (member 'mult (cdar x) :test #'eq))
2913	   ($listp (cadr x)) (cdadr x) (null (cddadr x)))
2914      (simplifya (cadadr x) z)
2915      (let ((badp (dolist (row (cdr x)) (if (not ($listp row)) (return t))))
2916	    (args (simpmap (cdr x) z)))
2917	(if (and args (not badp)) (matcheck args))
2918	(cons (if badp '(%matrix simp) '($matrix simp)) args))))
2919
2920(defun %itopot (pot)
2921  (if (fixnump pot)
2922      (let ((i (boole  boole-and pot 3)))
2923	(cond ((= i 0) 1)
2924	      ((= i 1) '$%i)
2925	      ((= i 2) -1)
2926	      (t (list '(mtimes simp) -1 '$%i))))
2927    (power -1 (mul2 pot '((rat simp) 1 2)))))
2928
2929(defun mnlogp (pot)
2930  (cond ((eq (caar pot) '%log) (simplifya (cadr pot) nil))
2931	((and (eq (caar pot) 'mtimes)
2932	      (or (maxima-integerp (cadr pot))
2933	          (and $%e_to_numlog ($numberp (cadr pot))))
2934	      (not (atom (caddr pot))) (eq (caar (caddr pot)) '%log)
2935	      (null (cdddr pot)))
2936	 (power (cadr (caddr pot)) (cadr pot)))))
2937
2938(defun mnlog (pot)
2939  (prog (a b c)
2940   loop (cond ((null pot)
2941	       (cond (a (setq a (cons '(mtimes) a))))
2942	       (cond (c (setq c (list '(mexpt simp) '$%e (addn c nil)))))
2943	       (return (cond ((null c) (simptimes a 1 nil))
2944			     ((null a) c)
2945			     (t (simptimes (append a (list c)) 1 nil)))))
2946	      ((and (among '%log (car pot)) (setq b (mnlogp (car pot))))
2947	       (setq a (cons b a)))
2948	      (t (setq c (cons (car pot) c))))
2949   (setq pot (cdr pot))
2950   (go loop)))
2951
2952(defun %etolog (pot) (cond ((mnlogp pot))
2953			   ((eq (caar pot) 'mplus) (mnlog (cdr pot)))
2954			   (t (list '(mexpt simp) '$%e pot))))
2955
2956(defun zerores (r1 r2)
2957  (cond ((or ($bfloatp r1) ($bfloatp r2)) bigfloatzero)
2958	((or (floatp r1) (floatp r2)) 0.0)
2959	(t 0)))
2960
2961(defmfun $orderlessp (a b)
2962  (setq a ($totaldisrep (specrepcheck a))
2963        b ($totaldisrep (specrepcheck b)))
2964  (and (not (alike1 a b)) (great b a) t))
2965
2966(defmfun $ordergreatp (a b)
2967  (setq a ($totaldisrep (specrepcheck a))
2968        b ($totaldisrep (specrepcheck b)))
2969  (and (not (alike1 a b)) (great a b) t))
2970
2971;; Test function to order a and b by magnitude. If it is not possible to
2972;; order a and b by magnitude they are ordered by great. This function
2973;; can be used by sort, e.g. sort([3,1,7,x,sin(1),minf],ordermagnitudep)
2974(defmfun $ordermagnitudep (a b)
2975  (let (sgn)
2976    (setq a ($totaldisrep (specrepcheck a))
2977          b ($totaldisrep (specrepcheck b)))
2978    (cond ((and (or (constp a) (member a '($inf $minf)))
2979                (or (constp b) (member b '($inf $minf)))
2980                (member (setq sgn ($csign (sub b a))) '($pos $neg $zero)))
2981           (cond ((eq sgn '$pos) t)
2982                 ((eq sgn '$zero) (and (not (alike1 a b)) (great b a)))
2983                 (t nil)))
2984          ((or (constp a) (member a '($inf $minf))) t)
2985          ((or (constp b) (member b '($inf $minf))) nil)
2986          (t (and (not (alike1 a b)) (great b a))))))
2987
2988(defun evnump (n) (or (even n) (and (ratnump n) (even (cadr n)))))
2989(defun odnump (n) (or (and (integerp n) (oddp n))
2990		      (and (ratnump n) (oddp (cadr n)))))
2991
2992(defun simplexpon (e)
2993  (or (maxima-integerp e)
2994      (and (eq $domain '$real) (ratnump e) (oddp (caddr e)))))
2995
2996;; This function is not called in Maxima core or share code
2997;; and can be cut out.
2998(defun noneg (p)
2999  (and (free p '$%i) (member ($sign p) '($pos $pz $zero) :test #'eq)))
3000
3001(defun radmabs (e)
3002  (if (and limitp (free e '$%i)) (asksign-p-or-n e))
3003  (simplifya (list '(mabs) e) t))
3004
3005(defun simpmqapply (exp y z)
3006  (let ((simpfun (and (not (atom (cadr exp))) (safe-get (caaadr exp) 'specsimp))) u)
3007    (if simpfun
3008	(funcall simpfun exp y z)
3009	(progn (setq u (simpargs exp z))
3010	       (if (symbolp (cadr u))
3011		   (simplifya (cons (cons (cadr u) (cdar u)) (cddr u)) z)
3012		   u)))))
3013
3014;; TRUE, if the symbol e is declared to be $complex or $imaginary.
3015(defun decl-complexp (e)
3016  (and (symbolp e)
3017       (kindp e '$complex)))
3018
3019;; TRUE, if the symbol e is declared to be $real, $rational, $irrational
3020;; or $integer
3021(defun decl-realp (e)
3022  (and (symbolp e)
3023       (or (kindp e '$real)
3024           (kindp e '$rational)
3025           (kindp e '$irrational)
3026           (kindp e '$integer))))
3027
3028;; WARNING:  Exercise extreme caution when modifying this function!
3029;;
3030;; Richard Fateman and Stavros Macrakis both say that changing the
3031;; actual ordering relations (as opposed to making them faster to
3032;; determine) could have very subtle and wide-ranging effects.  Also,
3033;; the simplifier spends the vast majority of its time here, so be
3034;; very careful about changes that may drastically slow down the
3035;; simplifier.
3036(defun great (x y)
3037  (cond ((atom x)
3038	 (cond ((atom y)
3039		(cond ((numberp x)
3040		       (cond ((numberp y)
3041			      (setq y (- x y))
3042			      (cond ((zerop y) (floatp x)) (t (plusp y))))))
3043		      ((constant x)
3044		       (cond ((constant y) (alphalessp y x)) (t (numberp y))))
3045		      ((mget x '$scalar)
3046		       (cond ((mget y '$scalar) (alphalessp y x))
3047		             (t (maxima-constantp y))))
3048		      ((mget x '$mainvar)
3049		       (cond ((mget y '$mainvar) (alphalessp y x)) (t t)))
3050		      (t (or (maxima-constantp y) (mget y '$scalar)
3051			     (and (not (mget y '$mainvar)) (not (null (alphalessp y x))))))))
3052	       (t (not (ordfna y x)))))
3053	((atom y) (ordfna x y))
3054	((eq (caar x) 'rat)
3055	 (cond ((eq (caar y) 'rat)
3056		(> (* (caddr y) (cadr x)) (* (caddr x) (cadr y))))))
3057	((eq (caar y) 'rat))
3058	((or (member (caar x) '(mtimes mplus mexpt %del) :test #'eq)
3059	     (member (caar y) '(mtimes mplus mexpt %del) :test #'eq))
3060	 (ordfn x y))
3061	((and (eq (caar x) 'bigfloat) (eq (caar y) 'bigfloat)) (mgrp x y))
3062	((or (eq (caar x) 'mrat) (eq (caar y) 'mrat))
3063	 (error "GREAT: internal error: unexpected MRAT argument"))
3064	(t (do ((x1 (margs x) (cdr x1)) (y1 (margs y) (cdr y1))) (())
3065	     (cond ((null x1)
3066		    (return (cond (y1 nil)
3067				  ((not (alike1 (mop x) (mop y)))
3068				   (great (mop x) (mop y)))
3069				  ((member 'array (cdar x) :test #'eq) t))))
3070		   ((null y1) (return t))
3071		   ((not (alike1 (car x1) (car y1)))
3072		    (return (great (car x1) (car y1)))))))))
3073
3074;; Trivial function used only in ALIKE1.
3075;; Should be defined as an open-codable subr.
3076
3077(defmacro memqarr (l)
3078  `(if (member 'array ,l :test #'eq) t))
3079
3080;; Compares two Macsyma expressions ignoring SIMP flags and all other
3081;; items in the header except for the ARRAY flag.
3082
3083(defun alike1 (x y)
3084  (cond ((eq x y))
3085	((atom x)
3086     (cond
3087       ((arrayp x)
3088	(and (arrayp y) (lisp-array-alike1 x y)))
3089
3090    ;; NOT SURE IF WE WANT TO ENABLE COMPARISON OF MAXIMA ARRAYS
3091    ;; ASIDE FROM THAT, ADD2LNC CALLS ALIKE1 (VIA MEMALIKE) AND THAT CAUSES TROUBLE
3092    ;; ((maxima-declared-arrayp x)
3093    ;;  (and (maxima-declared-arrayp y) (maxima-declared-array-alike1 x y)))
3094    ;; ((maxima-undeclared-arrayp x)
3095    ;;  (and (maxima-undeclared-arrayp y) (maxima-undeclared-array-alike1 x y)))
3096
3097       (t (equal x y))))
3098	((atom y) nil)
3099	((and
3100	  (not (atom (car x)))
3101	  (not (atom (car y)))
3102	  (eq (caar x) (caar y)))
3103         (cond
3104	  ((eq (caar x) 'mrat)
3105	   ;; Punt back to LIKE, which handles CREs.
3106	   (like x y))
3107	  (t (and
3108	      (eq (memqarr (cdar x)) (memqarr (cdar y)))
3109	      (alike (cdr x) (cdr y))))))))
3110
3111(defun lisp-array-alike1 (x y)
3112  (and
3113    (equal (array-dimensions x) (array-dimensions y))
3114    (progn
3115      (dotimes (i (array-total-size x))
3116	(if (not (alike1 (row-major-aref x i) (row-major-aref y i)))
3117	  (return-from lisp-array-alike1 nil)))
3118      t)))
3119
3120(defun maxima-declared-array-alike1 (x y)
3121  (lisp-array-alike1 (get (mget x 'array) 'array) (get (mget y 'array) 'array)))
3122
3123(defun maxima-undeclared-array-alike1 (x y)
3124  (and
3125    (alike1 (mfuncall '$arrayinfo x) (mfuncall '$arrayinfo y))
3126    (alike1 ($listarray x) ($listarray y))))
3127
3128;; Maps ALIKE1 down two lists.
3129
3130(defun alike (x y)
3131  (do ((x x (cdr x)) (y y (cdr y))) ((atom x) (equal x y))
3132    (cond ((or (atom y) (not (alike1 (car x) (car y))))
3133	   (return nil)))))
3134
3135(defun ordfna (e a)			; A is an atom
3136  (cond ((numberp a)
3137	 (or (not (eq (caar e) 'rat))
3138	     (> (cadr e) (* (caddr e) a))))
3139        ((and (constant a)
3140              (not (member (caar e) '(mplus mtimes mexpt) :test #'eq)))
3141	 (not (member (caar e) '(rat bigfloat) :test #'eq)))
3142	((eq (caar e) 'mrat)) ;; all MRATs succeed all atoms
3143	((null (margs e)) nil)
3144	((eq (caar e) 'mexpt)
3145	 (cond ((and (maxima-constantp (cadr e))
3146		     (or (not (constant a)) (not (maxima-constantp (caddr e)))))
3147		(or (not (free (caddr e) a)) (great (caddr e) a)))
3148	       ((eq (cadr e) a) (great (caddr e) 1))
3149	       (t (great (cadr e) a))))
3150	((member (caar e) '(mplus mtimes) :test #'eq)
3151	 (let ((u (car (last e))))
3152	   (cond ((eq u a) (not (ordhack e))) (t (great u a)))))
3153	((eq (caar e) '%del))
3154	((prog2 (setq e (car (margs e)))	; use first arg of e
3155	     (and (not (atom e)) (member (caar e) '(mplus mtimes) :test #'eq)))
3156	 (let ((u (car (last e))))		; and compare using
3157	   (cond ((eq u a) (not (ordhack e)))	; same procedure as above
3158		 (t (great u a)))))
3159	((eq e a))
3160	(t (great e a))))
3161
3162;; compare lists a and b elementwise from back to front
3163(defun ordlist (a b cx cy)
3164  (prog (l1 l2 c d)
3165     (setq l1 (length a) l2 (length b))
3166     loop (cond ((= l1 0)
3167		 (return (cond ((= l2 0) (eq cx 'mplus))
3168			       ((and (eq cx cy) (= l2 1))
3169				(great (cond ((eq cx 'mplus) 0) (t 1)) (car b))))))
3170		((= l2 0) (return (not (ordlist b a cy cx)))))
3171     (setq c (nthelem l1 a) d (nthelem l2 b))
3172     (cond ((not (alike1 c d)) (return (great c d))))
3173     (setq l1 (1- l1) l2 (1- l2))
3174     (go loop)))
3175
3176(defun term-list (x)
3177  (if (mplusp x)
3178      (cdr x)
3179    (list x)))
3180
3181(defun factor-list (x)
3182  (if (mtimesp x)
3183      (cdr x)
3184    (list x)))
3185
3186;; one of the exprs x or y should be one of:
3187;; %del, mexpt, mplus, mtimes
3188(defun ordfn (x y)
3189  (let ((cx (caar x)) (cy (caar y)))
3190    (cond ((eq cx '%del) (if (eq cy '%del) (great (cadr x) (cadr y)) t))
3191	  ((eq cy '%del) nil)
3192	  ((or (eq cx 'mtimes) (eq cy 'mtimes))
3193	   (ordlist (factor-list x) (factor-list y) 'mtimes 'mtimes))
3194	  ((or (eq cx 'mplus) (eq cy 'mplus))
3195	   (ordlist (term-list x) (term-list y) 'mplus 'mplus))
3196	  ((eq cx 'mexpt) (ordmexpt x y))
3197	  ((eq cy 'mexpt) (not (ordmexpt y x))))))
3198
3199(defun ordhack (x)
3200  (if (and (cddr x) (null (cdddr x)))
3201      (great (if (eq (caar x) 'mplus) 0 1) (cadr x))))
3202
3203(defun ordmexpt (x y)
3204  (cond ((eq (caar y) 'mexpt)
3205	 (cond ((alike1 (cadr x) (cadr y)) (great (caddr x) (caddr y)))
3206	       ((maxima-constantp (cadr x))
3207		(if (maxima-constantp (cadr y))
3208		    (if (or (alike1 (caddr x) (caddr y))
3209			    (and (mnump (caddr x)) (mnump (caddr y))))
3210			(great (cadr x) (cadr y))
3211			(great (caddr x) (caddr y)))
3212		    (great x (cadr y))))
3213	       ((maxima-constantp (cadr y)) (great (cadr x) y))
3214	       ((mnump (caddr x))
3215		(great (cadr x) (if (mnump (caddr y)) (cadr y) y)))
3216	       ((mnump (caddr y)) (great x (cadr y)))
3217	       (t (let ((x1 (simpln1 x)) (y1 (simpln1 y)))
3218		    (if (alike1 x1 y1) (great (cadr x) (cadr y))
3219			(great x1 y1))))))
3220	((alike1 (cadr x) y) (great (caddr x) 1))
3221	((mnump (caddr x)) (great (cadr x) y))
3222	(t (great (simpln1 x) (simpln (list '(%log) y) 1 t)))))
3223
3224(defmfun $multthru (e1 &optional e2)
3225  (let (arg1 arg2)
3226    (cond (e2				;called with two args
3227	   (setq arg1 (specrepcheck e1)
3228		 arg2 (specrepcheck e2))
3229           (cond ((or (atom arg2)
3230                      (not (member (caar arg2) '(mplus mequal) :test #'eq)))
3231		  (mul2 arg1 arg2))
3232		 ((eq (caar arg2) 'mequal)
3233		  (list (car arg2) ($multthru arg1 (cadr arg2))
3234			($multthru arg1 (caddr arg2))))
3235		 (t (expandterms arg1 (cdr arg2)))))
3236	  (t 				;called with only one arg
3237	   (prog (l1)
3238	      (setq arg1 (setq arg2 (specrepcheck e1)))
3239	      (cond ((atom arg1) (return arg1))
3240		    ((eq (caar arg1) 'mnctimes)
3241		     (setq arg1 (cdr arg1)) (go nct))
3242		    ((not (eq (caar arg1) 'mtimes)) (return arg1)))
3243	      (setq arg1 (reverse (cdr arg1)))
3244	      times (when (mplusp (car arg1))
3245		      (setq l1 (nconc l1 (cdr arg1)))
3246		      (return (expandterms (muln l1 t) (cdar arg1))))
3247	      (setq l1 (cons (car arg1) l1))
3248	      (setq arg1 (cdr arg1))
3249	      (if (null arg1) (return arg2))
3250	      (go times)
3251	      nct  (when (mplusp (car arg1))
3252		     (setq l1 (nreverse l1))
3253		     (return (addn (mapcar
3254				    #'(lambda (u)
3255					(simplifya
3256					 (cons '(mnctimes)
3257					       (append l1 (ncons u) (cdr arg1)))
3258					 t))
3259				    (cdar arg1))
3260				   t)))
3261	      (setq l1 (cons (car arg1) l1))
3262	      (setq arg1 (cdr arg1))
3263	      (if (null arg1) (return arg2))
3264	      (go nct))))))
3265
3266;;  EXPANDEXPT computes the expansion of (x1 + x2 + ... + xm)^n
3267;;  taking a sum and integer power as arguments.
3268;;  Its theory is to recurse down the binomial expansion of
3269;;  (x1 + (x2 + x3 + ... + xm))^n using the Binomial Expansion
3270;;  Thus it does a sigma:
3271;;
3272;;                n
3273;;             -------
3274;;              \         / n \    k                     (n - k)
3275;;               >        |   |  x1  (x2 + x3 + ... + xm)
3276;;              /         \ k /
3277;;             -------
3278;;               k=0
3279;;
3280;;   The function EXPONENTIATE-SUM does this and recurses through the second
3281;;   sum raised to a power.  It takes a list of terms and a positive integer
3282;;   power as arguments.
3283
3284
3285(defun expandexpt (sum power)
3286  (declare (fixnum power))
3287  (let ((expansion (exponentiate-sum (cdr sum) (abs power))))
3288    (cond ((plusp power) expansion)
3289	  (t `((mexpt simp) ,expansion -1)))))
3290
3291(defun exponentiate-sum (terms rpower)
3292  (declare (fixnum rpower))
3293  (cond ((= rpower 0) 1)
3294	((null (cdr terms)) (power (car terms) rpower))
3295	((= rpower 1) (cons '(mplus simp) terms))
3296	(t (do ((i 0 (1+ i))
3297		(result 0 (add2 result
3298				(muln (list (combination rpower i)
3299					    (exponentiate-sum (cdr terms)
3300							      (- rpower i))
3301					    (power (car terms) i)) t))))
3302	       ((> i rpower) result)
3303	     (declare (fixnum i))))))
3304
3305;;  Computes the combination of n elements taken m at a time by the formula
3306;;
3307;;     (n * (n-1) * ... * (n - m + 1)) / m! =
3308;;	(n / 1) * ((n - 1) / 2) * ... * ((n - m + 1) / m)
3309;;
3310;;  Checks for the case when m is greater than n/2 and translates
3311;;  to an equivalent expression.
3312
3313(defun combination (n m)
3314  (declare (fixnum n m))
3315  (cond ((> m (truncate n 2))
3316	 (combination n (- n m)))
3317	(t
3318	 (do ((result 1 (truncate (* result n1) m1))
3319	      (n1 n (1- n1))
3320	      (m1 1 (1+ m1)))
3321	     ((> m1 m) result)
3322	   (declare (fixnum  n1 m1))))))
3323
3324(defun expandsums (a b)
3325  (addn (prog (c)
3326	   (setq a (fixexpand a) b (cdr b))
3327	   loop
3328	   (when (null a) (return c))
3329	   (setq c (cons (expandterms (car a) b) c))
3330	   (setq a (cdr a))
3331	   (go loop))
3332	t))
3333
3334(defun expandterms (a b)
3335  (addn (prog (c)
3336	 loop
3337	 (when (null b) (return c))
3338	 (setq c (cons (mul2 a (car b)) c))
3339	 (setq b (cdr b))
3340	 (go loop))
3341	t))
3342
3343(defun genexpands (l)
3344  (prog ()
3345   loop
3346   (setq l (cdr l))
3347   (cond ((null l)
3348	  (setq prods (nreverse prods)
3349		negprods (nreverse negprods)
3350		sums (nreverse sums)
3351		negsums (nreverse negsums))
3352	  (return nil))
3353	 ((atom (car l))
3354	  (push (car l) prods))
3355	 ((eq (caaar l) 'rat)
3356	  (unless (equal (cadar l) 1)
3357	    (push (cadar l) prods))
3358	  (push (caddar l) negprods))
3359	 ((eq (caaar l) 'mplus)
3360	  (push (car l) sums))
3361	 ((and (eq (caaar l) 'mexpt)
3362	       (equal (caddar l) -1) (mplusp (cadar l)))
3363	  (push (cadar l) negsums))
3364	 ((and (eq (caaar l) 'mexpt)
3365	       (let ((expandp t))
3366		 (mminusp (caddar l))))
3367	  (push (if (equal (caddar l) -1)
3368		    (cadar l)
3369		    (list (caar l) (cadar l) (neg (caddar l))))
3370		negprods))
3371	 (t
3372	  (push (car l) prods)))
3373   (go loop)))
3374
3375(defun expandtimes (a)
3376  (prog (prods negprods sums negsums expsums expnegsums)
3377     (genexpands a)
3378     (setq prods (cond ((null prods) 1)
3379		       ((null (cdr prods)) (car prods))
3380		       (t (cons '(mtimes simp) prods))))
3381     (setq negprods (cond ((null negprods) 1)
3382			  ((null (cdr negprods)) (car negprods))
3383			  (t (cons '(mtimes simp) negprods))))
3384     (cond ((null sums) (go down))
3385	   (t (setq expsums (car sums))
3386	      (mapc #'(lambda (c)
3387			(setq expsums (expandsums expsums c)))
3388		    (cdr sums))))
3389     (setq prods (cond ((equal prods 1) expsums)
3390		       (t (expandterms prods (fixexpand expsums)))))
3391     down (cond ((null negsums)
3392		 (cond ((equal 1 negprods) (return prods))
3393		       ((mplusp prods)
3394		        (return (expandterms (power negprods -1) (cdr prods))))
3395		       (t (return (let ((expandflag t))
3396				    (mul2 prods (power negprods -1)))))))
3397		(t
3398		 (setq expnegsums (car negsums))
3399		 (mapc #'(lambda (c)
3400			   (setq expnegsums (expandsums expnegsums c)))
3401		       (cdr negsums))))
3402     (setq expnegsums (expandterms negprods (fixexpand expnegsums)))
3403     (return (if (mplusp prods)
3404		 (expandterms (list '(mexpt simp) expnegsums -1) (cdr prods))
3405		 (let ((expandflag t))
3406		   (mul2 prods (list '(mexpt simp) expnegsums -1)))))))
3407
3408(defun expand1 (exp $expop $expon)
3409  (unless (and (integerp $expop) (> $expop -1))
3410    (merror (intl:gettext "expand: expop must be a nonnegative integer; found: ~M") $expop))
3411  (unless (and (integerp $expon) (> $expon -1))
3412    (merror (intl:gettext "expand: expon must be a nonnegative integer; found: ~M") $expon))
3413  (resimplify (specrepcheck exp)))
3414
3415(defmfun $expand (exp &optional (expop $maxposex) (expon $maxnegex))
3416  (expand1 exp expop expon))
3417
3418(defun fixexpand (a)
3419  (if (not (mplusp a))
3420      (ncons a)
3421      (cdr a)))
3422
3423(defun simpnrt (x *n)			; computes X^(1/*N)
3424  (prog (*in *out varlist genvar $factorflag $dontfactor)
3425     (setq $factorflag t)
3426     (newvar x)
3427     (setq x (ratrep* x))
3428     (when (equal (cadr x) 0) (return 0))
3429     (setq x (ratfact (cdr x) 'psqfr))
3430     (simpnrt1 (mapcar #'pdis x))
3431     (setq *out (if *out (muln *out nil) 1))
3432     (setq *in (cond (*in
3433		      (setq *in (muln *in nil))
3434		      (nrthk *in *n))
3435		     (t 1)))
3436     (return (let (($%emode t))
3437	       (simplifya (list '(mtimes) *in *out)
3438			  (not (or (atom *in)
3439				   (atom (cadr *in))
3440				   (member (caaadr *in) '(mplus mtimes rat) :test #'eq))))))))
3441
3442(defun simpnrt1 (x)
3443  (do ((x x (cddr x)) (y))
3444      ((null x))
3445    (cond ((not (equal 1 (setq y (gcd (cadr x) *n))))
3446	   (push (simpnrt (list '(mexpt) (car x) (quotient (cadr x) y))
3447			  (quotient *n y))
3448		 *out))
3449	  ((and (equal (cadr x) 1) (integerp (car x)) (plusp (car x))
3450		(setq y (pnthrootp (car x) *n)))
3451	   (push y *out))
3452	  (t
3453	   (unless (> *n (abs (cadr x)))
3454	     (push (list '(mexpt) (car x) (quotient (cadr x) *n)) *out))
3455	   (push (list '(mexpt) (car x) (rem (cadr x) *n)) *in)))))
3456
3457(defun nrthk (in *n)
3458  (cond ((equal in 1)
3459	 1)
3460	((equal in -1)
3461	 (cond ((equal *n 2)
3462		'$%i)
3463	       ((eq $domain '$real)
3464		(if (even *n)
3465		    (nrthk2 -1 *n)
3466		    -1))
3467	       ($m1pbranch
3468		(let (($%emode t))
3469		  (power* '$%e (list '(mtimes) (list '(rat) 1 *n) '$%pi '$%i))))
3470	       (t
3471		(nrthk2 -1 *n))))
3472	((or (and wflag (eq ($asksign in) '$neg))
3473	     (and (mnump in) (equal ($sign in) '$neg)))
3474	 (nrthk1 (mul2* -1 in) *n))
3475	(t
3476	 (nrthk2 in *n))))
3477
3478(defun nrthk1 (in *n)			; computes (-IN)^(1/*N)
3479  (if $radexpand
3480      (mul2 (nrthk2 in *n) (nrthk -1 *n))
3481      (nrthk2 (mul2* -1 in) *n)))
3482
3483(defun nrthk2 (in *n)
3484  (power* in (list '(rat) 1 *n)))	; computes IN^(1/*N)
3485
3486;; The following was formerly in SININT.  This code was placed here because
3487;; SININT is now an out-of-core file on MC, and this code is needed in-core
3488;; because of the various calls to it. - BMT & JPG
3489
3490(declare-top (special var $ratfac ratform context))
3491
3492(defmfun $integrate (expr x &optional lo hi)
3493  (let ($ratfac)
3494    (if (not hi)
3495	(with-new-context (context)
3496	  (if (member '%risch *nounl* :test #'eq)
3497	      (rischint expr x)
3498	      (sinint expr x)))
3499	($defint expr x lo hi))))
3500
3501(defun ratp (a var)
3502  (cond ((atom a) t)
3503	((member (caar a) '(mplus mtimes) :test #'eq)
3504	 (do ((l (cdr a) (cdr l))) ((null l) t)
3505	   (or (ratp (car l) var) (return nil))))
3506	((eq (caar a) 'mexpt)
3507	 (if (free (cadr a) var)
3508	     (free (caddr a) var)
3509	     (and (integerp (caddr a)) (ratp (cadr a) var))))
3510	(t (free a var))))
3511
3512(defun ratnumerator (r)
3513  (cond ((atom r) r)
3514	((atom (cdr r)) (car r))
3515	((numberp (cadr r)) r)
3516	(t (car r))))
3517
3518(defun ratdenominator (r)
3519  (cond ((atom r) 1)
3520	((atom (cdr r)) (cdr r))
3521	((numberp (cadr r)) 1)
3522	(t (cdr r))))
3523
3524(declare-top (special var))
3525
3526;; (BPROG U V) appears to return A and B (as ((A1 . A2) B1 . B2) with A = A1/A2, B = B1/B2)
3527;; such that B/U + A/V = 1/(U*V), where U, V are polynomials represented as a list of
3528;; exponents and coefficients, (<gensym> E1 C1 E2 C2 ...) = C1*<gensym>^E1 + C2*<gensym>^E2 + ....
3529;; Example:
3530;; (%i73) partfrac ((2*x^2-3)/(x^4-3*x^2+2), x);
3531;; 1. Trace: (PARTFRAC '((#:X16910 2 2 0 -3) #:X16910 4 1 2 -3 0 2) '#:X16910)
3532;; 2. Trace: (BPROG '(#:X16910 2 1 0 -2) '(#:X16910 2 1 0 -1))
3533;; 2. Trace: BPROG ==> ((-1 . 1) 1 . 1)
3534;; 2. Trace: (BPROG '(#:X16910 1 1 0 1) '(#:X16910 1 1 0 -1))
3535;; 2. Trace: BPROG ==> ((1 . 2) -1 . 2)
3536;; 2. Trace: (BPROG '(#:X16910 1 1 0 -1) '1)
3537;; 2. Trace: BPROG ==> ((0 . 1) 1 . 1)
3538;; 1. Trace: PARTFRAC ==> ((0 . 1) ((1 . 2) (#:X16910 1 1 0 -1) 1) ((-1 . 2) (#:X16910 1 1 0 1) 1) ((1 . 1) (#:X16910 2 1 0 -2) 1))
3539;; (%o73) 1/(x^2-2)-1/(2*(x+1))+1/(2*(x-1))
3540
3541(defun bprog (r s)
3542  (prog (p1b p2b coef1r coef2r coef1s coef2s f1 f2 a egcd state seen-state)
3543     (setq r (ratfix r))
3544     (setq s (ratfix s))
3545     (setq coef2r (setq coef1s 0))
3546     (setq coef2s (setq coef1r 1))
3547     (setq a 1 egcd 1)
3548     (setq p1b (car r))
3549     (unless (zerop (pdegree p1b var)) (setq egcd (pgcdexpon p1b)))
3550     (setq p2b (car s))
3551	 (setq seen-state nil)
3552     (unless (or (zerop (pdegree p2b var)) (= egcd 1))
3553       (setq egcd (gcd egcd (pgcdexpon p2b)))
3554       (setq p1b (pexpon*// p1b egcd nil)
3555	     p2b (pexpon*// p2b egcd nil)))
3556     b1   (cond ((< (pdegree p1b var) (pdegree p2b var))
3557		 (rotatef p1b p2b)
3558		 (rotatef coef1r coef2r)
3559		 (rotatef coef1s coef2s)))
3560     (when (zerop (pdegree p2b var))
3561       (unless (zerop (pdegree coef2r var))
3562	 (setq coef2r (pexpon*// coef2r egcd t)))
3563       (unless (zerop (pdegree coef2s var))
3564	 (setq coef2s (pexpon*// coef2s egcd t)))
3565       (return (cons (ratreduce (ptimes (cdr r) coef2r) p2b)
3566		     (ratreduce (ptimes (cdr s) coef2s) p2b))))
3567     (setq f1 (psquorem1 (cdr p1b) (cdr p2b) t))
3568     (setq f2 (psimp var (cadr f1)))
3569     (setq p1b (pquotientchk (psimp var (caddr f1)) a))
3570     (setq f1 (car f1))
3571     (setq coef1r (pquotientchk (pdifference (ptimes f1 coef1r)
3572					     (ptimes f2 coef2r))
3573				a))
3574     (setq coef1s (pquotientchk (pdifference (ptimes f1 coef1s)
3575					     (ptimes f2 coef2s))
3576				a))
3577     (setq a f1)
3578	 ;; Catch an endless loop by keeping track of (p1b, p2b) combinations seen.
3579	 ;; Without this, rat(1/(x^(2/3)+1)) with algebraic = true loops forever.
3580	 (when (member (setq state (cons p1b p2b)) seen-state :test #'equal)
3581	   (rat-error (intl:gettext "BPROG: Failed to apply Bezout's identity")))
3582	 (push state seen-state)
3583     (go b1)))
3584
3585(defun ratdifference (a b) (ratplus a (ratminus b)))
3586
3587(defun ratpl (a b) (ratplus (ratfix a) (ratfix b)))
3588
3589(defun ratti (a b c) (rattimes (ratfix a) (ratfix b) c))
3590
3591(defun ratqu (a b) (ratquotient (ratfix a) (ratfix b)))
3592
3593(defun ratfix (a) (cond ((equal a (ratnumerator a)) (cons a 1)) (t a)))
3594
3595(defun ratdivide (f g)
3596  (destructuring-let* (((fnum . fden) (ratfix f))
3597		       ((gnum . gden) (ratfix g))
3598		       ((q r) (pdivide fnum gnum)))
3599    (cons (ratqu (ratti q gden t) fden)
3600	  (ratqu r fden))))
3601
3602(defun polcoef (l n) (cond ((or (atom l) (pointergp var (car l)))
3603			      (cond ((equal n 0) l) (t 0)))
3604			     (t (ptterm (cdr l) n))))
3605
3606(defun disrep (l) (cond ((equal (ratnumerator l) l)
3607			 ($ratdisrep (cons ratform (cons l 1))))
3608			(t ($ratdisrep (cons ratform l)))))
3609
3610(declare-top (unspecial var))
3611
3612
3613;; The following was formerly in MATRUN.  This code was placed here because
3614;; MATRUN is now an out-of-core file on MC, and this code is needed in-core
3615;; so that MACSYMA SAVE files will work. - JPG
3616
3617(defun matcherr ()
3618  (throw 'match nil))
3619
3620(defun kar (x) (if (atom x) (matcherr) (car x)))
3621
3622(defun kaar (x) (kar (kar x)))
3623
3624(defun kdr (x) (if (atom x) (matcherr) (cdr x)))
3625
3626(defun simpargs1 (a vestigial c)
3627  (declare (ignore vestigial))
3628  (simpargs a c))
3629
3630(defun *kar (x)
3631  (if (not (atom x)) (car x)))
3632
3633(defquote retlist (&rest l)
3634  (cons '(mlist simp)
3635	(mapcar #'(lambda (z) (list '(mequal simp) z (meval z))) l)))
3636
3637(defun nthkdr (x c)
3638  (if (zerop c) x (nthkdr (kdr x) (1- c))))
3639