1 #include "simint/boys/boys.h"
2 #include "simint/ostei/gen/ostei_generated.h"
3 #include "simint/vectorization/vectorization.h"
4 #include <math.h>
5 #include <string.h>
6
7
ostei_p_d_f_d(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__p_d_f_d)8 int ostei_p_d_f_d(struct simint_multi_shellpair const P,
9 struct simint_multi_shellpair const Q,
10 double screen_tol,
11 double * const restrict work,
12 double * const restrict INT__p_d_f_d)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__p_d_f_d);
17 int ab, cd, abcd;
18 int istart, jstart;
19 int iprimcd, nprim_icd, icd;
20 const int check_screen = (screen_tol > 0.0);
21 int i, j;
22 int n;
23 int not_screened;
24 int real_abcd;
25 int iket;
26 int ibra;
27
28 // partition workspace
29 double * const INT__s_d_f_s = work + (SIMINT_NSHELL_SIMD * 0);
30 double * const INT__s_d_g_s = work + (SIMINT_NSHELL_SIMD * 60);
31 double * const INT__s_d_h_s = work + (SIMINT_NSHELL_SIMD * 150);
32 double * const INT__s_f_f_s = work + (SIMINT_NSHELL_SIMD * 276);
33 double * const INT__s_f_g_s = work + (SIMINT_NSHELL_SIMD * 376);
34 double * const INT__s_f_h_s = work + (SIMINT_NSHELL_SIMD * 526);
35 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*736);
36 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_p_s = primwork + 9;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_d_s = primwork + 33;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_f_s = primwork + 75;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_g_s = primwork + 135;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_h_s = primwork + 210;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__s_p_p_s = primwork + 294;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__s_p_d_s = primwork + 321;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__s_p_f_s = primwork + 375;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__s_p_g_s = primwork + 465;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__s_p_h_s = primwork + 600;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__s_d_d_s = primwork + 789;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__s_d_f_s = primwork + 861;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__s_d_g_s = primwork + 981;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__s_d_h_s = primwork + 1161;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__s_f_f_s = primwork + 1413;
52 SIMINT_DBLTYPE * const restrict PRIM_INT__s_f_g_s = primwork + 1513;
53 SIMINT_DBLTYPE * const restrict PRIM_INT__s_f_h_s = primwork + 1663;
54 double * const hrrwork = (double *)(primwork + 1873);
55 double * const HRR_INT__p_d_f_s = hrrwork + 0;
56 double * const HRR_INT__p_d_f_p = hrrwork + 180;
57 double * const HRR_INT__p_d_g_s = hrrwork + 720;
58 double * const HRR_INT__p_d_g_p = hrrwork + 990;
59 double * const HRR_INT__p_d_h_s = hrrwork + 1800;
60
61
62 // Create constants
63 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
64 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
65 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
66 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
67 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
68 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
69
70
71 ////////////////////////////////////////
72 // Loop over shells and primitives
73 ////////////////////////////////////////
74
75 real_abcd = 0;
76 istart = 0;
77 for(ab = 0; ab < P.nshell12_clip; ++ab)
78 {
79 const int iend = istart + P.nprim12[ab];
80
81 cd = 0;
82 jstart = 0;
83
84 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
85 {
86 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
87 int jend = jstart;
88 for(i = 0; i < nshellbatch; i++)
89 jend += Q.nprim12[cd+i];
90
91 // Clear the beginning of the workspace (where we are accumulating integrals)
92 memset(work, 0, SIMINT_NSHELL_SIMD * 736 * sizeof(double));
93 abcd = 0;
94
95
96 for(i = istart; i < iend; ++i)
97 {
98 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
99
100 if(check_screen)
101 {
102 // Skip this whole thing if always insignificant
103 if((P.screen[i] * Q.screen_max) < screen_tol)
104 continue;
105 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
106 }
107
108 icd = 0;
109 iprimcd = 0;
110 nprim_icd = Q.nprim12[cd];
111 double * restrict PRIM_PTR_INT__s_d_f_s = INT__s_d_f_s + abcd * 60;
112 double * restrict PRIM_PTR_INT__s_d_g_s = INT__s_d_g_s + abcd * 90;
113 double * restrict PRIM_PTR_INT__s_d_h_s = INT__s_d_h_s + abcd * 126;
114 double * restrict PRIM_PTR_INT__s_f_f_s = INT__s_f_f_s + abcd * 100;
115 double * restrict PRIM_PTR_INT__s_f_g_s = INT__s_f_g_s + abcd * 150;
116 double * restrict PRIM_PTR_INT__s_f_h_s = INT__s_f_h_s + abcd * 210;
117
118
119
120 // Load these one per loop over i
121 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
122 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
123 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
124
125 const SIMINT_DBLTYPE P_PB[3] = { SIMINT_DBLSET1(P.PB_x[i]), SIMINT_DBLSET1(P.PB_y[i]), SIMINT_DBLSET1(P.PB_z[i]) };
126
127 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
128 {
129 // calculate the shell offsets
130 // these are the offset from the shell pointed to by cd
131 // for each element
132 int shelloffsets[SIMINT_SIMD_LEN] = {0};
133 int lastoffset = 0;
134 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
135
136 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
137 {
138 // Handle if the first element of the vector is a new shell
139 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
140 {
141 nprim_icd += Q.nprim12[cd + (++icd)];
142 PRIM_PTR_INT__s_d_f_s += 60;
143 PRIM_PTR_INT__s_d_g_s += 90;
144 PRIM_PTR_INT__s_d_h_s += 126;
145 PRIM_PTR_INT__s_f_f_s += 100;
146 PRIM_PTR_INT__s_f_g_s += 150;
147 PRIM_PTR_INT__s_f_h_s += 210;
148 }
149 iprimcd++;
150 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
151 {
152 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
153 {
154 shelloffsets[n] = shelloffsets[n-1] + 1;
155 lastoffset++;
156 nprim_icd += Q.nprim12[cd + (++icd)];
157 }
158 else
159 shelloffsets[n] = shelloffsets[n-1];
160 iprimcd++;
161 }
162 }
163 else
164 iprimcd += SIMINT_SIMD_LEN;
165
166 // Do we have to compute this vector (or has it been screened out)?
167 // (not_screened != 0 means we have to do this vector)
168 if(check_screen)
169 {
170 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
171 if(vmax < screen_tol)
172 {
173 PRIM_PTR_INT__s_d_f_s += lastoffset*60;
174 PRIM_PTR_INT__s_d_g_s += lastoffset*90;
175 PRIM_PTR_INT__s_d_h_s += lastoffset*126;
176 PRIM_PTR_INT__s_f_f_s += lastoffset*100;
177 PRIM_PTR_INT__s_f_g_s += lastoffset*150;
178 PRIM_PTR_INT__s_f_h_s += lastoffset*210;
179 continue;
180 }
181 }
182
183 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
184 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
185 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
186 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
187
188
189 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
190 SIMINT_DBLTYPE PQ[3];
191 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
192 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
193 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
194 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
195 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
196 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
197
198 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
199 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
200 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
201 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
202 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
203 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
204 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
205
206 // NOTE: Minus sign!
207 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
208 SIMINT_DBLTYPE aop_PQ[3];
209 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
210 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
211 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
212
213 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
214 SIMINT_DBLTYPE aoq_PQ[3];
215 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
216 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
217 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
218 // Put a minus sign here so we don't have to in RR routines
219 a_over_q = SIMINT_NEG(a_over_q);
220
221
222 //////////////////////////////////////////////
223 // Fjt function section
224 // Maximum v value: 8
225 //////////////////////////////////////////////
226 // The parameter to the Fjt function
227 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
228
229
230 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
231
232
233 boys_F_split(PRIM_INT__s_s_s_s, F_x, 8);
234 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
235 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
236 for(n = 0; n <= 8; n++)
237 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
238
239 //////////////////////////////////////////////
240 // Primitive integrals: Vertical recurrance
241 //////////////////////////////////////////////
242
243 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
244 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
245 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
246 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
247 const SIMINT_DBLTYPE vrr_const_3_over_2q = SIMINT_MUL(const_3, one_over_2q);
248 const SIMINT_DBLTYPE vrr_const_4_over_2q = SIMINT_MUL(const_4, one_over_2q);
249 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
250 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
251 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
252 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
253 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
254
255
256
257 // Forming PRIM_INT__s_s_p_s[8 * 3];
258 for(n = 0; n < 8; ++n) // loop over orders of auxiliary function
259 {
260
261 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
262 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_p_s[n * 3 + 0]);
263
264 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
265 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_p_s[n * 3 + 1]);
266
267 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
268 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_p_s[n * 3 + 2]);
269
270 }
271
272
273
274 // Forming PRIM_INT__s_s_d_s[7 * 6];
275 for(n = 0; n < 7; ++n) // loop over orders of auxiliary function
276 {
277
278 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_p_s[n * 3 + 0]);
279 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_s_d_s[n * 6 + 0]);
280 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_FMADD( vrr_const_1_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__s_s_d_s[n * 6 + 0]);
281
282 PRIM_INT__s_s_d_s[n * 6 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 0]);
283 PRIM_INT__s_s_d_s[n * 6 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_s_d_s[n * 6 + 1]);
284
285 PRIM_INT__s_s_d_s[n * 6 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 0]);
286 PRIM_INT__s_s_d_s[n * 6 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_s_d_s[n * 6 + 2]);
287
288 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 1]);
289 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_s_d_s[n * 6 + 3]);
290 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_FMADD( vrr_const_1_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__s_s_d_s[n * 6 + 3]);
291
292 PRIM_INT__s_s_d_s[n * 6 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 1]);
293 PRIM_INT__s_s_d_s[n * 6 + 4] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_s_d_s[n * 6 + 4]);
294
295 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 2]);
296 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__s_s_d_s[n * 6 + 5]);
297 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_FMADD( vrr_const_1_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__s_s_d_s[n * 6 + 5]);
298
299 }
300
301
302
303 // Forming PRIM_INT__s_s_f_s[6 * 10];
304 for(n = 0; n < 6; ++n) // loop over orders of auxiliary function
305 {
306
307 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
308 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__s_s_f_s[n * 10 + 0]);
309 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_FMADD( vrr_const_2_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_s_p_s[n * 3 + 0]), PRIM_INT__s_s_f_s[n * 10 + 0]);
310
311 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
312 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__s_s_f_s[n * 10 + 1]);
313
314 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
315 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__s_s_f_s[n * 10 + 2]);
316
317 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
318 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__s_s_f_s[n * 10 + 3]);
319
320 PRIM_INT__s_s_f_s[n * 10 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
321 PRIM_INT__s_s_f_s[n * 10 + 4] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 1], PRIM_INT__s_s_f_s[n * 10 + 4]);
322
323 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
324 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__s_s_f_s[n * 10 + 5]);
325
326 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
327 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__s_s_f_s[n * 10 + 6]);
328 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_FMADD( vrr_const_2_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_s_p_s[n * 3 + 1]), PRIM_INT__s_s_f_s[n * 10 + 6]);
329
330 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
331 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__s_s_f_s[n * 10 + 7]);
332
333 PRIM_INT__s_s_f_s[n * 10 + 8] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
334 PRIM_INT__s_s_f_s[n * 10 + 8] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__s_s_f_s[n * 10 + 8]);
335
336 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
337 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__s_s_f_s[n * 10 + 9]);
338 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_FMADD( vrr_const_2_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__s_s_p_s[n * 3 + 2]), PRIM_INT__s_s_f_s[n * 10 + 9]);
339
340 }
341
342
343 VRR_J_s_p_f_s(
344 PRIM_INT__s_p_f_s,
345 PRIM_INT__s_s_f_s,
346 PRIM_INT__s_s_d_s,
347 P_PB,
348 aop_PQ,
349 one_over_2pq,
350 3);
351
352
353
354 // Forming PRIM_INT__s_p_d_s[3 * 18];
355 for(n = 0; n < 3; ++n) // loop over orders of auxiliary function
356 {
357
358 PRIM_INT__s_p_d_s[n * 18 + 0] = SIMINT_MUL(P_PB[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
359 PRIM_INT__s_p_d_s[n * 18 + 0] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__s_p_d_s[n * 18 + 0]);
360 PRIM_INT__s_p_d_s[n * 18 + 0] = SIMINT_FMADD( vrr_const_2_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_p_d_s[n * 18 + 0]);
361
362 PRIM_INT__s_p_d_s[n * 18 + 1] = SIMINT_MUL(P_PB[0], PRIM_INT__s_s_d_s[n * 6 + 1]);
363 PRIM_INT__s_p_d_s[n * 18 + 1] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 1], PRIM_INT__s_p_d_s[n * 18 + 1]);
364 PRIM_INT__s_p_d_s[n * 18 + 1] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_p_d_s[n * 18 + 1]);
365
366 PRIM_INT__s_p_d_s[n * 18 + 2] = SIMINT_MUL(P_PB[0], PRIM_INT__s_s_d_s[n * 6 + 2]);
367 PRIM_INT__s_p_d_s[n * 18 + 2] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 2], PRIM_INT__s_p_d_s[n * 18 + 2]);
368 PRIM_INT__s_p_d_s[n * 18 + 2] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__s_p_d_s[n * 18 + 2]);
369
370 PRIM_INT__s_p_d_s[n * 18 + 3] = SIMINT_MUL(P_PB[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
371 PRIM_INT__s_p_d_s[n * 18 + 3] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__s_p_d_s[n * 18 + 3]);
372
373 PRIM_INT__s_p_d_s[n * 18 + 4] = SIMINT_MUL(P_PB[0], PRIM_INT__s_s_d_s[n * 6 + 4]);
374 PRIM_INT__s_p_d_s[n * 18 + 4] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 4], PRIM_INT__s_p_d_s[n * 18 + 4]);
375
376 PRIM_INT__s_p_d_s[n * 18 + 5] = SIMINT_MUL(P_PB[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
377 PRIM_INT__s_p_d_s[n * 18 + 5] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__s_p_d_s[n * 18 + 5]);
378
379 PRIM_INT__s_p_d_s[n * 18 + 6] = SIMINT_MUL(P_PB[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
380 PRIM_INT__s_p_d_s[n * 18 + 6] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__s_p_d_s[n * 18 + 6]);
381
382 PRIM_INT__s_p_d_s[n * 18 + 7] = SIMINT_MUL(P_PB[1], PRIM_INT__s_s_d_s[n * 6 + 1]);
383 PRIM_INT__s_p_d_s[n * 18 + 7] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 1], PRIM_INT__s_p_d_s[n * 18 + 7]);
384 PRIM_INT__s_p_d_s[n * 18 + 7] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_p_d_s[n * 18 + 7]);
385
386 PRIM_INT__s_p_d_s[n * 18 + 8] = SIMINT_MUL(P_PB[1], PRIM_INT__s_s_d_s[n * 6 + 2]);
387 PRIM_INT__s_p_d_s[n * 18 + 8] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 2], PRIM_INT__s_p_d_s[n * 18 + 8]);
388
389 PRIM_INT__s_p_d_s[n * 18 + 9] = SIMINT_MUL(P_PB[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
390 PRIM_INT__s_p_d_s[n * 18 + 9] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__s_p_d_s[n * 18 + 9]);
391 PRIM_INT__s_p_d_s[n * 18 + 9] = SIMINT_FMADD( vrr_const_2_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_p_d_s[n * 18 + 9]);
392
393 PRIM_INT__s_p_d_s[n * 18 + 10] = SIMINT_MUL(P_PB[1], PRIM_INT__s_s_d_s[n * 6 + 4]);
394 PRIM_INT__s_p_d_s[n * 18 + 10] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 4], PRIM_INT__s_p_d_s[n * 18 + 10]);
395 PRIM_INT__s_p_d_s[n * 18 + 10] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__s_p_d_s[n * 18 + 10]);
396
397 PRIM_INT__s_p_d_s[n * 18 + 11] = SIMINT_MUL(P_PB[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
398 PRIM_INT__s_p_d_s[n * 18 + 11] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__s_p_d_s[n * 18 + 11]);
399
400 PRIM_INT__s_p_d_s[n * 18 + 12] = SIMINT_MUL(P_PB[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
401 PRIM_INT__s_p_d_s[n * 18 + 12] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__s_p_d_s[n * 18 + 12]);
402
403 PRIM_INT__s_p_d_s[n * 18 + 13] = SIMINT_MUL(P_PB[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
404 PRIM_INT__s_p_d_s[n * 18 + 13] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 1], PRIM_INT__s_p_d_s[n * 18 + 13]);
405
406 PRIM_INT__s_p_d_s[n * 18 + 14] = SIMINT_MUL(P_PB[2], PRIM_INT__s_s_d_s[n * 6 + 2]);
407 PRIM_INT__s_p_d_s[n * 18 + 14] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 2], PRIM_INT__s_p_d_s[n * 18 + 14]);
408 PRIM_INT__s_p_d_s[n * 18 + 14] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_p_d_s[n * 18 + 14]);
409
410 PRIM_INT__s_p_d_s[n * 18 + 15] = SIMINT_MUL(P_PB[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
411 PRIM_INT__s_p_d_s[n * 18 + 15] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__s_p_d_s[n * 18 + 15]);
412
413 PRIM_INT__s_p_d_s[n * 18 + 16] = SIMINT_MUL(P_PB[2], PRIM_INT__s_s_d_s[n * 6 + 4]);
414 PRIM_INT__s_p_d_s[n * 18 + 16] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 4], PRIM_INT__s_p_d_s[n * 18 + 16]);
415 PRIM_INT__s_p_d_s[n * 18 + 16] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_p_d_s[n * 18 + 16]);
416
417 PRIM_INT__s_p_d_s[n * 18 + 17] = SIMINT_MUL(P_PB[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
418 PRIM_INT__s_p_d_s[n * 18 + 17] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__s_p_d_s[n * 18 + 17]);
419 PRIM_INT__s_p_d_s[n * 18 + 17] = SIMINT_FMADD( vrr_const_2_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__s_p_d_s[n * 18 + 17]);
420
421 }
422
423
424 VRR_J_s_d_f_s(
425 PRIM_INT__s_d_f_s,
426 PRIM_INT__s_p_f_s,
427 PRIM_INT__s_s_f_s,
428 PRIM_INT__s_p_d_s,
429 P_PB,
430 a_over_p,
431 aop_PQ,
432 one_over_2p,
433 one_over_2pq,
434 2);
435
436
437 VRR_K_s_s_g_s(
438 PRIM_INT__s_s_g_s,
439 PRIM_INT__s_s_f_s,
440 PRIM_INT__s_s_d_s,
441 Q_PA,
442 a_over_q,
443 aoq_PQ,
444 one_over_2q,
445 5);
446
447
448 VRR_J_s_p_g_s(
449 PRIM_INT__s_p_g_s,
450 PRIM_INT__s_s_g_s,
451 PRIM_INT__s_s_f_s,
452 P_PB,
453 aop_PQ,
454 one_over_2pq,
455 3);
456
457
458 ostei_general_vrr_J(0, 2, 4, 0, 2,
459 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PB,
460 PRIM_INT__s_p_g_s, NULL, PRIM_INT__s_s_g_s, PRIM_INT__s_p_f_s, NULL, PRIM_INT__s_d_g_s);
461
462
463
464 // Forming PRIM_INT__s_p_p_s[3 * 9];
465 for(n = 0; n < 3; ++n) // loop over orders of auxiliary function
466 {
467
468 PRIM_INT__s_p_p_s[n * 9 + 0] = SIMINT_MUL(P_PB[0], PRIM_INT__s_s_p_s[n * 3 + 0]);
469 PRIM_INT__s_p_p_s[n * 9 + 0] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_p_p_s[n * 9 + 0]);
470 PRIM_INT__s_p_p_s[n * 9 + 0] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_p_p_s[n * 9 + 0]);
471
472 PRIM_INT__s_p_p_s[n * 9 + 1] = SIMINT_MUL(P_PB[0], PRIM_INT__s_s_p_s[n * 3 + 1]);
473 PRIM_INT__s_p_p_s[n * 9 + 1] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_p_p_s[n * 9 + 1]);
474
475 PRIM_INT__s_p_p_s[n * 9 + 2] = SIMINT_MUL(P_PB[0], PRIM_INT__s_s_p_s[n * 3 + 2]);
476 PRIM_INT__s_p_p_s[n * 9 + 2] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__s_p_p_s[n * 9 + 2]);
477
478 PRIM_INT__s_p_p_s[n * 9 + 3] = SIMINT_MUL(P_PB[1], PRIM_INT__s_s_p_s[n * 3 + 0]);
479 PRIM_INT__s_p_p_s[n * 9 + 3] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_p_p_s[n * 9 + 3]);
480
481 PRIM_INT__s_p_p_s[n * 9 + 4] = SIMINT_MUL(P_PB[1], PRIM_INT__s_s_p_s[n * 3 + 1]);
482 PRIM_INT__s_p_p_s[n * 9 + 4] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_p_p_s[n * 9 + 4]);
483 PRIM_INT__s_p_p_s[n * 9 + 4] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_p_p_s[n * 9 + 4]);
484
485 PRIM_INT__s_p_p_s[n * 9 + 5] = SIMINT_MUL(P_PB[1], PRIM_INT__s_s_p_s[n * 3 + 2]);
486 PRIM_INT__s_p_p_s[n * 9 + 5] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__s_p_p_s[n * 9 + 5]);
487
488 PRIM_INT__s_p_p_s[n * 9 + 6] = SIMINT_MUL(P_PB[2], PRIM_INT__s_s_p_s[n * 3 + 0]);
489 PRIM_INT__s_p_p_s[n * 9 + 6] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_p_p_s[n * 9 + 6]);
490
491 PRIM_INT__s_p_p_s[n * 9 + 7] = SIMINT_MUL(P_PB[2], PRIM_INT__s_s_p_s[n * 3 + 1]);
492 PRIM_INT__s_p_p_s[n * 9 + 7] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_p_p_s[n * 9 + 7]);
493
494 PRIM_INT__s_p_p_s[n * 9 + 8] = SIMINT_MUL(P_PB[2], PRIM_INT__s_s_p_s[n * 3 + 2]);
495 PRIM_INT__s_p_p_s[n * 9 + 8] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__s_p_p_s[n * 9 + 8]);
496 PRIM_INT__s_p_p_s[n * 9 + 8] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_p_p_s[n * 9 + 8]);
497
498 }
499
500
501 VRR_J_s_d_d_s(
502 PRIM_INT__s_d_d_s,
503 PRIM_INT__s_p_d_s,
504 PRIM_INT__s_s_d_s,
505 PRIM_INT__s_p_p_s,
506 P_PB,
507 a_over_p,
508 aop_PQ,
509 one_over_2p,
510 one_over_2pq,
511 2);
512
513
514 ostei_general_vrr_J(0, 3, 3, 0, 1,
515 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PB,
516 PRIM_INT__s_d_f_s, NULL, PRIM_INT__s_p_f_s, PRIM_INT__s_d_d_s, NULL, PRIM_INT__s_f_f_s);
517
518
519 VRR_K_s_s_h_s(
520 PRIM_INT__s_s_h_s,
521 PRIM_INT__s_s_g_s,
522 PRIM_INT__s_s_f_s,
523 Q_PA,
524 a_over_q,
525 aoq_PQ,
526 one_over_2q,
527 4);
528
529
530 ostei_general_vrr_J(0, 1, 5, 0, 3,
531 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PB,
532 PRIM_INT__s_s_h_s, NULL, NULL, PRIM_INT__s_s_g_s, NULL, PRIM_INT__s_p_h_s);
533
534
535 ostei_general_vrr_J(0, 2, 5, 0, 2,
536 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PB,
537 PRIM_INT__s_p_h_s, NULL, PRIM_INT__s_s_h_s, PRIM_INT__s_p_g_s, NULL, PRIM_INT__s_d_h_s);
538
539
540 ostei_general_vrr_J(0, 3, 4, 0, 1,
541 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PB,
542 PRIM_INT__s_d_g_s, NULL, PRIM_INT__s_p_g_s, PRIM_INT__s_d_f_s, NULL, PRIM_INT__s_f_g_s);
543
544
545 ostei_general_vrr_J(0, 3, 5, 0, 1,
546 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PB,
547 PRIM_INT__s_d_h_s, NULL, PRIM_INT__s_p_h_s, PRIM_INT__s_d_g_s, NULL, PRIM_INT__s_f_h_s);
548
549
550
551
552 ////////////////////////////////////
553 // Accumulate contracted integrals
554 ////////////////////////////////////
555 if(lastoffset == 0)
556 {
557 contract_all(60, PRIM_INT__s_d_f_s, PRIM_PTR_INT__s_d_f_s);
558 contract_all(90, PRIM_INT__s_d_g_s, PRIM_PTR_INT__s_d_g_s);
559 contract_all(126, PRIM_INT__s_d_h_s, PRIM_PTR_INT__s_d_h_s);
560 contract_all(100, PRIM_INT__s_f_f_s, PRIM_PTR_INT__s_f_f_s);
561 contract_all(150, PRIM_INT__s_f_g_s, PRIM_PTR_INT__s_f_g_s);
562 contract_all(210, PRIM_INT__s_f_h_s, PRIM_PTR_INT__s_f_h_s);
563 }
564 else
565 {
566 contract(60, shelloffsets, PRIM_INT__s_d_f_s, PRIM_PTR_INT__s_d_f_s);
567 contract(90, shelloffsets, PRIM_INT__s_d_g_s, PRIM_PTR_INT__s_d_g_s);
568 contract(126, shelloffsets, PRIM_INT__s_d_h_s, PRIM_PTR_INT__s_d_h_s);
569 contract(100, shelloffsets, PRIM_INT__s_f_f_s, PRIM_PTR_INT__s_f_f_s);
570 contract(150, shelloffsets, PRIM_INT__s_f_g_s, PRIM_PTR_INT__s_f_g_s);
571 contract(210, shelloffsets, PRIM_INT__s_f_h_s, PRIM_PTR_INT__s_f_h_s);
572 PRIM_PTR_INT__s_d_f_s += lastoffset*60;
573 PRIM_PTR_INT__s_d_g_s += lastoffset*90;
574 PRIM_PTR_INT__s_d_h_s += lastoffset*126;
575 PRIM_PTR_INT__s_f_f_s += lastoffset*100;
576 PRIM_PTR_INT__s_f_g_s += lastoffset*150;
577 PRIM_PTR_INT__s_f_h_s += lastoffset*210;
578 }
579
580 } // close loop over j
581 } // close loop over i
582
583 //Advance to the next batch
584 jstart = SIMINT_SIMD_ROUND(jend);
585
586 //////////////////////////////////////////////
587 // Contracted integrals: Horizontal recurrance
588 //////////////////////////////////////////////
589
590
591 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
592
593
594 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
595 {
596 const double hCD[3] = { Q.AB_x[cd+abcd], Q.AB_y[cd+abcd], Q.AB_z[cd+abcd] };
597
598 // set up HRR pointers
599 double const * restrict HRR_INT__s_d_f_s = INT__s_d_f_s + abcd * 60;
600 double const * restrict HRR_INT__s_d_g_s = INT__s_d_g_s + abcd * 90;
601 double const * restrict HRR_INT__s_d_h_s = INT__s_d_h_s + abcd * 126;
602 double const * restrict HRR_INT__s_f_f_s = INT__s_f_f_s + abcd * 100;
603 double const * restrict HRR_INT__s_f_g_s = INT__s_f_g_s + abcd * 150;
604 double const * restrict HRR_INT__s_f_h_s = INT__s_f_h_s + abcd * 210;
605 double * restrict HRR_INT__p_d_f_d = INT__p_d_f_d + real_abcd * 1080;
606
607 // form INT__p_d_f_s
608 for(iket = 0; iket < 10; ++iket)
609 {
610 HRR_INT__p_d_f_s[0 * 10 + iket] = HRR_INT__s_f_f_s[0 * 10 + iket] - ( hAB[0] * HRR_INT__s_d_f_s[0 * 10 + iket] );
611
612 HRR_INT__p_d_f_s[1 * 10 + iket] = HRR_INT__s_f_f_s[1 * 10 + iket] - ( hAB[0] * HRR_INT__s_d_f_s[1 * 10 + iket] );
613
614 HRR_INT__p_d_f_s[2 * 10 + iket] = HRR_INT__s_f_f_s[2 * 10 + iket] - ( hAB[0] * HRR_INT__s_d_f_s[2 * 10 + iket] );
615
616 HRR_INT__p_d_f_s[3 * 10 + iket] = HRR_INT__s_f_f_s[3 * 10 + iket] - ( hAB[0] * HRR_INT__s_d_f_s[3 * 10 + iket] );
617
618 HRR_INT__p_d_f_s[4 * 10 + iket] = HRR_INT__s_f_f_s[4 * 10 + iket] - ( hAB[0] * HRR_INT__s_d_f_s[4 * 10 + iket] );
619
620 HRR_INT__p_d_f_s[5 * 10 + iket] = HRR_INT__s_f_f_s[5 * 10 + iket] - ( hAB[0] * HRR_INT__s_d_f_s[5 * 10 + iket] );
621
622 HRR_INT__p_d_f_s[6 * 10 + iket] = HRR_INT__s_f_f_s[1 * 10 + iket] - ( hAB[1] * HRR_INT__s_d_f_s[0 * 10 + iket] );
623
624 HRR_INT__p_d_f_s[7 * 10 + iket] = HRR_INT__s_f_f_s[3 * 10 + iket] - ( hAB[1] * HRR_INT__s_d_f_s[1 * 10 + iket] );
625
626 HRR_INT__p_d_f_s[8 * 10 + iket] = HRR_INT__s_f_f_s[4 * 10 + iket] - ( hAB[1] * HRR_INT__s_d_f_s[2 * 10 + iket] );
627
628 HRR_INT__p_d_f_s[9 * 10 + iket] = HRR_INT__s_f_f_s[6 * 10 + iket] - ( hAB[1] * HRR_INT__s_d_f_s[3 * 10 + iket] );
629
630 HRR_INT__p_d_f_s[10 * 10 + iket] = HRR_INT__s_f_f_s[7 * 10 + iket] - ( hAB[1] * HRR_INT__s_d_f_s[4 * 10 + iket] );
631
632 HRR_INT__p_d_f_s[11 * 10 + iket] = HRR_INT__s_f_f_s[8 * 10 + iket] - ( hAB[1] * HRR_INT__s_d_f_s[5 * 10 + iket] );
633
634 HRR_INT__p_d_f_s[12 * 10 + iket] = HRR_INT__s_f_f_s[2 * 10 + iket] - ( hAB[2] * HRR_INT__s_d_f_s[0 * 10 + iket] );
635
636 HRR_INT__p_d_f_s[13 * 10 + iket] = HRR_INT__s_f_f_s[4 * 10 + iket] - ( hAB[2] * HRR_INT__s_d_f_s[1 * 10 + iket] );
637
638 HRR_INT__p_d_f_s[14 * 10 + iket] = HRR_INT__s_f_f_s[5 * 10 + iket] - ( hAB[2] * HRR_INT__s_d_f_s[2 * 10 + iket] );
639
640 HRR_INT__p_d_f_s[15 * 10 + iket] = HRR_INT__s_f_f_s[7 * 10 + iket] - ( hAB[2] * HRR_INT__s_d_f_s[3 * 10 + iket] );
641
642 HRR_INT__p_d_f_s[16 * 10 + iket] = HRR_INT__s_f_f_s[8 * 10 + iket] - ( hAB[2] * HRR_INT__s_d_f_s[4 * 10 + iket] );
643
644 HRR_INT__p_d_f_s[17 * 10 + iket] = HRR_INT__s_f_f_s[9 * 10 + iket] - ( hAB[2] * HRR_INT__s_d_f_s[5 * 10 + iket] );
645
646 }
647
648
649 // form INT__p_d_g_s
650 for(iket = 0; iket < 15; ++iket)
651 {
652 HRR_INT__p_d_g_s[0 * 15 + iket] = HRR_INT__s_f_g_s[0 * 15 + iket] - ( hAB[0] * HRR_INT__s_d_g_s[0 * 15 + iket] );
653
654 HRR_INT__p_d_g_s[1 * 15 + iket] = HRR_INT__s_f_g_s[1 * 15 + iket] - ( hAB[0] * HRR_INT__s_d_g_s[1 * 15 + iket] );
655
656 HRR_INT__p_d_g_s[2 * 15 + iket] = HRR_INT__s_f_g_s[2 * 15 + iket] - ( hAB[0] * HRR_INT__s_d_g_s[2 * 15 + iket] );
657
658 HRR_INT__p_d_g_s[3 * 15 + iket] = HRR_INT__s_f_g_s[3 * 15 + iket] - ( hAB[0] * HRR_INT__s_d_g_s[3 * 15 + iket] );
659
660 HRR_INT__p_d_g_s[4 * 15 + iket] = HRR_INT__s_f_g_s[4 * 15 + iket] - ( hAB[0] * HRR_INT__s_d_g_s[4 * 15 + iket] );
661
662 HRR_INT__p_d_g_s[5 * 15 + iket] = HRR_INT__s_f_g_s[5 * 15 + iket] - ( hAB[0] * HRR_INT__s_d_g_s[5 * 15 + iket] );
663
664 HRR_INT__p_d_g_s[6 * 15 + iket] = HRR_INT__s_f_g_s[1 * 15 + iket] - ( hAB[1] * HRR_INT__s_d_g_s[0 * 15 + iket] );
665
666 HRR_INT__p_d_g_s[7 * 15 + iket] = HRR_INT__s_f_g_s[3 * 15 + iket] - ( hAB[1] * HRR_INT__s_d_g_s[1 * 15 + iket] );
667
668 HRR_INT__p_d_g_s[8 * 15 + iket] = HRR_INT__s_f_g_s[4 * 15 + iket] - ( hAB[1] * HRR_INT__s_d_g_s[2 * 15 + iket] );
669
670 HRR_INT__p_d_g_s[9 * 15 + iket] = HRR_INT__s_f_g_s[6 * 15 + iket] - ( hAB[1] * HRR_INT__s_d_g_s[3 * 15 + iket] );
671
672 HRR_INT__p_d_g_s[10 * 15 + iket] = HRR_INT__s_f_g_s[7 * 15 + iket] - ( hAB[1] * HRR_INT__s_d_g_s[4 * 15 + iket] );
673
674 HRR_INT__p_d_g_s[11 * 15 + iket] = HRR_INT__s_f_g_s[8 * 15 + iket] - ( hAB[1] * HRR_INT__s_d_g_s[5 * 15 + iket] );
675
676 HRR_INT__p_d_g_s[12 * 15 + iket] = HRR_INT__s_f_g_s[2 * 15 + iket] - ( hAB[2] * HRR_INT__s_d_g_s[0 * 15 + iket] );
677
678 HRR_INT__p_d_g_s[13 * 15 + iket] = HRR_INT__s_f_g_s[4 * 15 + iket] - ( hAB[2] * HRR_INT__s_d_g_s[1 * 15 + iket] );
679
680 HRR_INT__p_d_g_s[14 * 15 + iket] = HRR_INT__s_f_g_s[5 * 15 + iket] - ( hAB[2] * HRR_INT__s_d_g_s[2 * 15 + iket] );
681
682 HRR_INT__p_d_g_s[15 * 15 + iket] = HRR_INT__s_f_g_s[7 * 15 + iket] - ( hAB[2] * HRR_INT__s_d_g_s[3 * 15 + iket] );
683
684 HRR_INT__p_d_g_s[16 * 15 + iket] = HRR_INT__s_f_g_s[8 * 15 + iket] - ( hAB[2] * HRR_INT__s_d_g_s[4 * 15 + iket] );
685
686 HRR_INT__p_d_g_s[17 * 15 + iket] = HRR_INT__s_f_g_s[9 * 15 + iket] - ( hAB[2] * HRR_INT__s_d_g_s[5 * 15 + iket] );
687
688 }
689
690
691 // form INT__p_d_h_s
692 for(iket = 0; iket < 21; ++iket)
693 {
694 HRR_INT__p_d_h_s[0 * 21 + iket] = HRR_INT__s_f_h_s[0 * 21 + iket] - ( hAB[0] * HRR_INT__s_d_h_s[0 * 21 + iket] );
695
696 HRR_INT__p_d_h_s[1 * 21 + iket] = HRR_INT__s_f_h_s[1 * 21 + iket] - ( hAB[0] * HRR_INT__s_d_h_s[1 * 21 + iket] );
697
698 HRR_INT__p_d_h_s[2 * 21 + iket] = HRR_INT__s_f_h_s[2 * 21 + iket] - ( hAB[0] * HRR_INT__s_d_h_s[2 * 21 + iket] );
699
700 HRR_INT__p_d_h_s[3 * 21 + iket] = HRR_INT__s_f_h_s[3 * 21 + iket] - ( hAB[0] * HRR_INT__s_d_h_s[3 * 21 + iket] );
701
702 HRR_INT__p_d_h_s[4 * 21 + iket] = HRR_INT__s_f_h_s[4 * 21 + iket] - ( hAB[0] * HRR_INT__s_d_h_s[4 * 21 + iket] );
703
704 HRR_INT__p_d_h_s[5 * 21 + iket] = HRR_INT__s_f_h_s[5 * 21 + iket] - ( hAB[0] * HRR_INT__s_d_h_s[5 * 21 + iket] );
705
706 HRR_INT__p_d_h_s[6 * 21 + iket] = HRR_INT__s_f_h_s[1 * 21 + iket] - ( hAB[1] * HRR_INT__s_d_h_s[0 * 21 + iket] );
707
708 HRR_INT__p_d_h_s[7 * 21 + iket] = HRR_INT__s_f_h_s[3 * 21 + iket] - ( hAB[1] * HRR_INT__s_d_h_s[1 * 21 + iket] );
709
710 HRR_INT__p_d_h_s[8 * 21 + iket] = HRR_INT__s_f_h_s[4 * 21 + iket] - ( hAB[1] * HRR_INT__s_d_h_s[2 * 21 + iket] );
711
712 HRR_INT__p_d_h_s[9 * 21 + iket] = HRR_INT__s_f_h_s[6 * 21 + iket] - ( hAB[1] * HRR_INT__s_d_h_s[3 * 21 + iket] );
713
714 HRR_INT__p_d_h_s[10 * 21 + iket] = HRR_INT__s_f_h_s[7 * 21 + iket] - ( hAB[1] * HRR_INT__s_d_h_s[4 * 21 + iket] );
715
716 HRR_INT__p_d_h_s[11 * 21 + iket] = HRR_INT__s_f_h_s[8 * 21 + iket] - ( hAB[1] * HRR_INT__s_d_h_s[5 * 21 + iket] );
717
718 HRR_INT__p_d_h_s[12 * 21 + iket] = HRR_INT__s_f_h_s[2 * 21 + iket] - ( hAB[2] * HRR_INT__s_d_h_s[0 * 21 + iket] );
719
720 HRR_INT__p_d_h_s[13 * 21 + iket] = HRR_INT__s_f_h_s[4 * 21 + iket] - ( hAB[2] * HRR_INT__s_d_h_s[1 * 21 + iket] );
721
722 HRR_INT__p_d_h_s[14 * 21 + iket] = HRR_INT__s_f_h_s[5 * 21 + iket] - ( hAB[2] * HRR_INT__s_d_h_s[2 * 21 + iket] );
723
724 HRR_INT__p_d_h_s[15 * 21 + iket] = HRR_INT__s_f_h_s[7 * 21 + iket] - ( hAB[2] * HRR_INT__s_d_h_s[3 * 21 + iket] );
725
726 HRR_INT__p_d_h_s[16 * 21 + iket] = HRR_INT__s_f_h_s[8 * 21 + iket] - ( hAB[2] * HRR_INT__s_d_h_s[4 * 21 + iket] );
727
728 HRR_INT__p_d_h_s[17 * 21 + iket] = HRR_INT__s_f_h_s[9 * 21 + iket] - ( hAB[2] * HRR_INT__s_d_h_s[5 * 21 + iket] );
729
730 }
731
732
733 // form INT__p_d_f_p
734 HRR_L_f_p(
735 HRR_INT__p_d_f_p,
736 HRR_INT__p_d_f_s,
737 HRR_INT__p_d_g_s,
738 hCD, 18);
739
740 // form INT__p_d_g_p
741 HRR_L_g_p(
742 HRR_INT__p_d_g_p,
743 HRR_INT__p_d_g_s,
744 HRR_INT__p_d_h_s,
745 hCD, 18);
746
747 // form INT__p_d_f_d
748 HRR_L_f_d(
749 HRR_INT__p_d_f_d,
750 HRR_INT__p_d_f_p,
751 HRR_INT__p_d_g_p,
752 hCD, 18);
753
754
755 } // close HRR loop
756
757
758 } // close loop cdbatch
759
760 istart = iend;
761 } // close loop over ab
762
763 return P.nshell12_clip * Q.nshell12_clip;
764 }
765
766