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_f_p_h_p(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__f_p_h_p)8 int ostei_f_p_h_p(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__f_p_h_p)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__f_p_h_p);
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__f_s_h_s = work + (SIMINT_NSHELL_SIMD * 0);
30 double * const INT__f_s_i_s = work + (SIMINT_NSHELL_SIMD * 210);
31 double * const INT__g_s_h_s = work + (SIMINT_NSHELL_SIMD * 490);
32 double * const INT__g_s_i_s = work + (SIMINT_NSHELL_SIMD * 805);
33 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*1225);
34 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_p_s = primwork + 11;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_d_s = primwork + 41;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_f_s = primwork + 95;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_g_s = primwork + 175;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_h_s = primwork + 280;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_i_s = primwork + 406;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_d_s = primwork + 546;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_f_s = primwork + 618;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_g_s = primwork + 738;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_h_s = primwork + 918;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_i_s = primwork + 1170;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_f_s = primwork + 1506;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_g_s = primwork + 1686;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_h_s = primwork + 1956;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_i_s = primwork + 2334;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_g_s = primwork + 2838;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_h_s = primwork + 3138;
52 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_i_s = primwork + 3558;
53 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_h_s = primwork + 4118;
54 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_i_s = primwork + 4433;
55 double * const hrrwork = (double *)(primwork + 4853);
56 double * const HRR_INT__f_p_h_s = hrrwork + 0;
57 double * const HRR_INT__f_p_i_s = hrrwork + 630;
58
59
60 // Create constants
61 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
62 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
63 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
64 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
65 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
66 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
67 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
68
69
70 ////////////////////////////////////////
71 // Loop over shells and primitives
72 ////////////////////////////////////////
73
74 real_abcd = 0;
75 istart = 0;
76 for(ab = 0; ab < P.nshell12_clip; ++ab)
77 {
78 const int iend = istart + P.nprim12[ab];
79
80 cd = 0;
81 jstart = 0;
82
83 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
84 {
85 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
86 int jend = jstart;
87 for(i = 0; i < nshellbatch; i++)
88 jend += Q.nprim12[cd+i];
89
90 // Clear the beginning of the workspace (where we are accumulating integrals)
91 memset(work, 0, SIMINT_NSHELL_SIMD * 1225 * sizeof(double));
92 abcd = 0;
93
94
95 for(i = istart; i < iend; ++i)
96 {
97 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
98
99 if(check_screen)
100 {
101 // Skip this whole thing if always insignificant
102 if((P.screen[i] * Q.screen_max) < screen_tol)
103 continue;
104 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
105 }
106
107 icd = 0;
108 iprimcd = 0;
109 nprim_icd = Q.nprim12[cd];
110 double * restrict PRIM_PTR_INT__f_s_h_s = INT__f_s_h_s + abcd * 210;
111 double * restrict PRIM_PTR_INT__f_s_i_s = INT__f_s_i_s + abcd * 280;
112 double * restrict PRIM_PTR_INT__g_s_h_s = INT__g_s_h_s + abcd * 315;
113 double * restrict PRIM_PTR_INT__g_s_i_s = INT__g_s_i_s + abcd * 420;
114
115
116
117 // Load these one per loop over i
118 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
119 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
120 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
121
122 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
123
124 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
125 {
126 // calculate the shell offsets
127 // these are the offset from the shell pointed to by cd
128 // for each element
129 int shelloffsets[SIMINT_SIMD_LEN] = {0};
130 int lastoffset = 0;
131 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
132
133 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
134 {
135 // Handle if the first element of the vector is a new shell
136 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
137 {
138 nprim_icd += Q.nprim12[cd + (++icd)];
139 PRIM_PTR_INT__f_s_h_s += 210;
140 PRIM_PTR_INT__f_s_i_s += 280;
141 PRIM_PTR_INT__g_s_h_s += 315;
142 PRIM_PTR_INT__g_s_i_s += 420;
143 }
144 iprimcd++;
145 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
146 {
147 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
148 {
149 shelloffsets[n] = shelloffsets[n-1] + 1;
150 lastoffset++;
151 nprim_icd += Q.nprim12[cd + (++icd)];
152 }
153 else
154 shelloffsets[n] = shelloffsets[n-1];
155 iprimcd++;
156 }
157 }
158 else
159 iprimcd += SIMINT_SIMD_LEN;
160
161 // Do we have to compute this vector (or has it been screened out)?
162 // (not_screened != 0 means we have to do this vector)
163 if(check_screen)
164 {
165 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
166 if(vmax < screen_tol)
167 {
168 PRIM_PTR_INT__f_s_h_s += lastoffset*210;
169 PRIM_PTR_INT__f_s_i_s += lastoffset*280;
170 PRIM_PTR_INT__g_s_h_s += lastoffset*315;
171 PRIM_PTR_INT__g_s_i_s += lastoffset*420;
172 continue;
173 }
174 }
175
176 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
177 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
178 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
179 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
180
181
182 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
183 SIMINT_DBLTYPE PQ[3];
184 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
185 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
186 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
187 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
188 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
189 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
190
191 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
192 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
193 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
194 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
195 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
196 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
197 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
198
199 // NOTE: Minus sign!
200 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
201 SIMINT_DBLTYPE aop_PQ[3];
202 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
203 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
204 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
205
206 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
207 SIMINT_DBLTYPE aoq_PQ[3];
208 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
209 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
210 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
211 // Put a minus sign here so we don't have to in RR routines
212 a_over_q = SIMINT_NEG(a_over_q);
213
214
215 //////////////////////////////////////////////
216 // Fjt function section
217 // Maximum v value: 10
218 //////////////////////////////////////////////
219 // The parameter to the Fjt function
220 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
221
222
223 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
224
225
226 boys_F_split(PRIM_INT__s_s_s_s, F_x, 10);
227 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
228 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
229 for(n = 0; n <= 10; n++)
230 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
231
232 //////////////////////////////////////////////
233 // Primitive integrals: Vertical recurrance
234 //////////////////////////////////////////////
235
236 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
237 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
238 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
239 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
240 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
241 const SIMINT_DBLTYPE vrr_const_3_over_2q = SIMINT_MUL(const_3, one_over_2q);
242 const SIMINT_DBLTYPE vrr_const_4_over_2q = SIMINT_MUL(const_4, one_over_2q);
243 const SIMINT_DBLTYPE vrr_const_5_over_2q = SIMINT_MUL(const_5, one_over_2q);
244 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
245 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
246 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
247 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
248 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
249 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
250
251
252
253 // Forming PRIM_INT__s_s_p_s[10 * 3];
254 for(n = 0; n < 10; ++n) // loop over orders of auxiliary function
255 {
256
257 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
258 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]);
259
260 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
261 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]);
262
263 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
264 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]);
265
266 }
267
268
269
270 // Forming PRIM_INT__s_s_d_s[9 * 6];
271 for(n = 0; n < 9; ++n) // loop over orders of auxiliary function
272 {
273
274 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_p_s[n * 3 + 0]);
275 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]);
276 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]);
277
278 PRIM_INT__s_s_d_s[n * 6 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 0]);
279 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]);
280
281 PRIM_INT__s_s_d_s[n * 6 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 0]);
282 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]);
283
284 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 1]);
285 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]);
286 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]);
287
288 PRIM_INT__s_s_d_s[n * 6 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 1]);
289 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]);
290
291 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 2]);
292 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]);
293 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]);
294
295 }
296
297
298
299 // Forming PRIM_INT__s_s_f_s[8 * 10];
300 for(n = 0; n < 8; ++n) // loop over orders of auxiliary function
301 {
302
303 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
304 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]);
305 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]);
306
307 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
308 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]);
309
310 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
311 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]);
312
313 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
314 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]);
315
316 PRIM_INT__s_s_f_s[n * 10 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
317 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]);
318
319 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
320 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]);
321
322 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
323 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]);
324 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]);
325
326 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
327 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]);
328
329 PRIM_INT__s_s_f_s[n * 10 + 8] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
330 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]);
331
332 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
333 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]);
334 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]);
335
336 }
337
338
339 VRR_K_s_s_g_s(
340 PRIM_INT__s_s_g_s,
341 PRIM_INT__s_s_f_s,
342 PRIM_INT__s_s_d_s,
343 Q_PA,
344 a_over_q,
345 aoq_PQ,
346 one_over_2q,
347 7);
348
349
350 VRR_K_s_s_h_s(
351 PRIM_INT__s_s_h_s,
352 PRIM_INT__s_s_g_s,
353 PRIM_INT__s_s_f_s,
354 Q_PA,
355 a_over_q,
356 aoq_PQ,
357 one_over_2q,
358 6);
359
360
361 ostei_general_vrr_I(1, 0, 5, 0, 4,
362 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
363 PRIM_INT__s_s_h_s, NULL, NULL, PRIM_INT__s_s_g_s, NULL, PRIM_INT__p_s_h_s);
364
365
366 VRR_I_p_s_g_s(
367 PRIM_INT__p_s_g_s,
368 PRIM_INT__s_s_g_s,
369 PRIM_INT__s_s_f_s,
370 P_PA,
371 aop_PQ,
372 one_over_2pq,
373 4);
374
375
376 ostei_general_vrr_I(2, 0, 5, 0, 3,
377 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
378 PRIM_INT__p_s_h_s, PRIM_INT__s_s_h_s, NULL, PRIM_INT__p_s_g_s, NULL, PRIM_INT__d_s_h_s);
379
380
381 VRR_I_p_s_f_s(
382 PRIM_INT__p_s_f_s,
383 PRIM_INT__s_s_f_s,
384 PRIM_INT__s_s_d_s,
385 P_PA,
386 aop_PQ,
387 one_over_2pq,
388 4);
389
390
391 ostei_general_vrr_I(2, 0, 4, 0, 3,
392 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
393 PRIM_INT__p_s_g_s, PRIM_INT__s_s_g_s, NULL, PRIM_INT__p_s_f_s, NULL, PRIM_INT__d_s_g_s);
394
395
396 ostei_general_vrr_I(3, 0, 5, 0, 2,
397 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
398 PRIM_INT__d_s_h_s, PRIM_INT__p_s_h_s, NULL, PRIM_INT__d_s_g_s, NULL, PRIM_INT__f_s_h_s);
399
400
401 ostei_general_vrr1_K(6, 5,
402 one_over_2q, a_over_q, aoq_PQ, Q_PA,
403 PRIM_INT__s_s_h_s, PRIM_INT__s_s_g_s, PRIM_INT__s_s_i_s);
404
405
406 ostei_general_vrr_I(1, 0, 6, 0, 4,
407 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
408 PRIM_INT__s_s_i_s, NULL, NULL, PRIM_INT__s_s_h_s, NULL, PRIM_INT__p_s_i_s);
409
410
411 ostei_general_vrr_I(2, 0, 6, 0, 3,
412 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
413 PRIM_INT__p_s_i_s, PRIM_INT__s_s_i_s, NULL, PRIM_INT__p_s_h_s, NULL, PRIM_INT__d_s_i_s);
414
415
416 ostei_general_vrr_I(3, 0, 6, 0, 2,
417 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
418 PRIM_INT__d_s_i_s, PRIM_INT__p_s_i_s, NULL, PRIM_INT__d_s_h_s, NULL, PRIM_INT__f_s_i_s);
419
420
421
422 // Forming PRIM_INT__p_s_d_s[4 * 18];
423 for(n = 0; n < 4; ++n) // loop over orders of auxiliary function
424 {
425
426 PRIM_INT__p_s_d_s[n * 18 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
427 PRIM_INT__p_s_d_s[n * 18 + 0] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__p_s_d_s[n * 18 + 0]);
428 PRIM_INT__p_s_d_s[n * 18 + 0] = SIMINT_FMADD( vrr_const_2_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__p_s_d_s[n * 18 + 0]);
429
430 PRIM_INT__p_s_d_s[n * 18 + 1] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 1]);
431 PRIM_INT__p_s_d_s[n * 18 + 1] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 1], PRIM_INT__p_s_d_s[n * 18 + 1]);
432 PRIM_INT__p_s_d_s[n * 18 + 1] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__p_s_d_s[n * 18 + 1]);
433
434 PRIM_INT__p_s_d_s[n * 18 + 2] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 2]);
435 PRIM_INT__p_s_d_s[n * 18 + 2] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 2], PRIM_INT__p_s_d_s[n * 18 + 2]);
436 PRIM_INT__p_s_d_s[n * 18 + 2] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__p_s_d_s[n * 18 + 2]);
437
438 PRIM_INT__p_s_d_s[n * 18 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
439 PRIM_INT__p_s_d_s[n * 18 + 3] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__p_s_d_s[n * 18 + 3]);
440
441 PRIM_INT__p_s_d_s[n * 18 + 4] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 4]);
442 PRIM_INT__p_s_d_s[n * 18 + 4] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 4], PRIM_INT__p_s_d_s[n * 18 + 4]);
443
444 PRIM_INT__p_s_d_s[n * 18 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
445 PRIM_INT__p_s_d_s[n * 18 + 5] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__p_s_d_s[n * 18 + 5]);
446
447 PRIM_INT__p_s_d_s[n * 18 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
448 PRIM_INT__p_s_d_s[n * 18 + 6] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__p_s_d_s[n * 18 + 6]);
449
450 PRIM_INT__p_s_d_s[n * 18 + 7] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 1]);
451 PRIM_INT__p_s_d_s[n * 18 + 7] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 1], PRIM_INT__p_s_d_s[n * 18 + 7]);
452 PRIM_INT__p_s_d_s[n * 18 + 7] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__p_s_d_s[n * 18 + 7]);
453
454 PRIM_INT__p_s_d_s[n * 18 + 8] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 2]);
455 PRIM_INT__p_s_d_s[n * 18 + 8] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 2], PRIM_INT__p_s_d_s[n * 18 + 8]);
456
457 PRIM_INT__p_s_d_s[n * 18 + 9] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
458 PRIM_INT__p_s_d_s[n * 18 + 9] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__p_s_d_s[n * 18 + 9]);
459 PRIM_INT__p_s_d_s[n * 18 + 9] = SIMINT_FMADD( vrr_const_2_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__p_s_d_s[n * 18 + 9]);
460
461 PRIM_INT__p_s_d_s[n * 18 + 10] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 4]);
462 PRIM_INT__p_s_d_s[n * 18 + 10] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 4], PRIM_INT__p_s_d_s[n * 18 + 10]);
463 PRIM_INT__p_s_d_s[n * 18 + 10] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__p_s_d_s[n * 18 + 10]);
464
465 PRIM_INT__p_s_d_s[n * 18 + 11] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
466 PRIM_INT__p_s_d_s[n * 18 + 11] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__p_s_d_s[n * 18 + 11]);
467
468 PRIM_INT__p_s_d_s[n * 18 + 12] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
469 PRIM_INT__p_s_d_s[n * 18 + 12] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__p_s_d_s[n * 18 + 12]);
470
471 PRIM_INT__p_s_d_s[n * 18 + 13] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
472 PRIM_INT__p_s_d_s[n * 18 + 13] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 1], PRIM_INT__p_s_d_s[n * 18 + 13]);
473
474 PRIM_INT__p_s_d_s[n * 18 + 14] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 2]);
475 PRIM_INT__p_s_d_s[n * 18 + 14] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 2], PRIM_INT__p_s_d_s[n * 18 + 14]);
476 PRIM_INT__p_s_d_s[n * 18 + 14] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__p_s_d_s[n * 18 + 14]);
477
478 PRIM_INT__p_s_d_s[n * 18 + 15] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
479 PRIM_INT__p_s_d_s[n * 18 + 15] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__p_s_d_s[n * 18 + 15]);
480
481 PRIM_INT__p_s_d_s[n * 18 + 16] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 4]);
482 PRIM_INT__p_s_d_s[n * 18 + 16] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 4], PRIM_INT__p_s_d_s[n * 18 + 16]);
483 PRIM_INT__p_s_d_s[n * 18 + 16] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__p_s_d_s[n * 18 + 16]);
484
485 PRIM_INT__p_s_d_s[n * 18 + 17] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
486 PRIM_INT__p_s_d_s[n * 18 + 17] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__p_s_d_s[n * 18 + 17]);
487 PRIM_INT__p_s_d_s[n * 18 + 17] = SIMINT_FMADD( vrr_const_2_over_2pq, PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__p_s_d_s[n * 18 + 17]);
488
489 }
490
491
492 VRR_I_d_s_f_s(
493 PRIM_INT__d_s_f_s,
494 PRIM_INT__p_s_f_s,
495 PRIM_INT__s_s_f_s,
496 PRIM_INT__p_s_d_s,
497 P_PA,
498 a_over_p,
499 aop_PQ,
500 one_over_2p,
501 one_over_2pq,
502 3);
503
504
505 ostei_general_vrr_I(3, 0, 4, 0, 2,
506 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
507 PRIM_INT__d_s_g_s, PRIM_INT__p_s_g_s, NULL, PRIM_INT__d_s_f_s, NULL, PRIM_INT__f_s_g_s);
508
509
510 ostei_general_vrr_I(4, 0, 5, 0, 1,
511 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
512 PRIM_INT__f_s_h_s, PRIM_INT__d_s_h_s, NULL, PRIM_INT__f_s_g_s, NULL, PRIM_INT__g_s_h_s);
513
514
515 ostei_general_vrr_I(4, 0, 6, 0, 1,
516 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
517 PRIM_INT__f_s_i_s, PRIM_INT__d_s_i_s, NULL, PRIM_INT__f_s_h_s, NULL, PRIM_INT__g_s_i_s);
518
519
520
521
522 ////////////////////////////////////
523 // Accumulate contracted integrals
524 ////////////////////////////////////
525 if(lastoffset == 0)
526 {
527 contract_all(210, PRIM_INT__f_s_h_s, PRIM_PTR_INT__f_s_h_s);
528 contract_all(280, PRIM_INT__f_s_i_s, PRIM_PTR_INT__f_s_i_s);
529 contract_all(315, PRIM_INT__g_s_h_s, PRIM_PTR_INT__g_s_h_s);
530 contract_all(420, PRIM_INT__g_s_i_s, PRIM_PTR_INT__g_s_i_s);
531 }
532 else
533 {
534 contract(210, shelloffsets, PRIM_INT__f_s_h_s, PRIM_PTR_INT__f_s_h_s);
535 contract(280, shelloffsets, PRIM_INT__f_s_i_s, PRIM_PTR_INT__f_s_i_s);
536 contract(315, shelloffsets, PRIM_INT__g_s_h_s, PRIM_PTR_INT__g_s_h_s);
537 contract(420, shelloffsets, PRIM_INT__g_s_i_s, PRIM_PTR_INT__g_s_i_s);
538 PRIM_PTR_INT__f_s_h_s += lastoffset*210;
539 PRIM_PTR_INT__f_s_i_s += lastoffset*280;
540 PRIM_PTR_INT__g_s_h_s += lastoffset*315;
541 PRIM_PTR_INT__g_s_i_s += lastoffset*420;
542 }
543
544 } // close loop over j
545 } // close loop over i
546
547 //Advance to the next batch
548 jstart = SIMINT_SIMD_ROUND(jend);
549
550 //////////////////////////////////////////////
551 // Contracted integrals: Horizontal recurrance
552 //////////////////////////////////////////////
553
554
555 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
556
557
558 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
559 {
560 const double hCD[3] = { Q.AB_x[cd+abcd], Q.AB_y[cd+abcd], Q.AB_z[cd+abcd] };
561
562 // set up HRR pointers
563 double const * restrict HRR_INT__f_s_h_s = INT__f_s_h_s + abcd * 210;
564 double const * restrict HRR_INT__f_s_i_s = INT__f_s_i_s + abcd * 280;
565 double const * restrict HRR_INT__g_s_h_s = INT__g_s_h_s + abcd * 315;
566 double const * restrict HRR_INT__g_s_i_s = INT__g_s_i_s + abcd * 420;
567 double * restrict HRR_INT__f_p_h_p = INT__f_p_h_p + real_abcd * 1890;
568
569 // form INT__f_p_h_s
570 HRR_J_f_p(
571 HRR_INT__f_p_h_s,
572 HRR_INT__f_s_h_s,
573 HRR_INT__g_s_h_s,
574 hAB, 21);
575
576 // form INT__f_p_i_s
577 HRR_J_f_p(
578 HRR_INT__f_p_i_s,
579 HRR_INT__f_s_i_s,
580 HRR_INT__g_s_i_s,
581 hAB, 28);
582
583 // form INT__f_p_h_p
584 ostei_general_hrr_L(3, 1, 5, 1, hCD, HRR_INT__f_p_i_s, HRR_INT__f_p_h_s, HRR_INT__f_p_h_p);
585
586
587 } // close HRR loop
588
589
590 } // close loop cdbatch
591
592 istart = iend;
593 } // close loop over ab
594
595 return P.nshell12_clip * Q.nshell12_clip;
596 }
597
ostei_p_f_h_p(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__p_f_h_p)598 int ostei_p_f_h_p(struct simint_multi_shellpair const P,
599 struct simint_multi_shellpair const Q,
600 double screen_tol,
601 double * const restrict work,
602 double * const restrict INT__p_f_h_p)
603 {
604 double P_AB[3*P.nshell12];
605 struct simint_multi_shellpair P_tmp = P;
606 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
607 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
608 P_tmp.AB_x = P_AB;
609 P_tmp.AB_y = P_AB + P.nshell12;
610 P_tmp.AB_z = P_AB + 2*P.nshell12;
611
612 for(int i = 0; i < P.nshell12; i++)
613 {
614 P_tmp.AB_x[i] = -P.AB_x[i];
615 P_tmp.AB_y[i] = -P.AB_y[i];
616 P_tmp.AB_z[i] = -P.AB_z[i];
617 }
618
619 int ret = ostei_f_p_h_p(P_tmp, Q, screen_tol, work, INT__p_f_h_p);
620 double buffer[1890] SIMINT_ALIGN_ARRAY_DBL;
621
622 for(int q = 0; q < ret; q++)
623 {
624 int idx = 0;
625 for(int a = 0; a < 3; ++a)
626 for(int b = 0; b < 10; ++b)
627 for(int c = 0; c < 21; ++c)
628 for(int d = 0; d < 3; ++d)
629 buffer[idx++] = INT__p_f_h_p[q*1890+b*189+a*63+c*3+d];
630
631 memcpy(INT__p_f_h_p+q*1890, buffer, 1890*sizeof(double));
632 }
633
634 return ret;
635 }
636
ostei_f_p_p_h(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__f_p_p_h)637 int ostei_f_p_p_h(struct simint_multi_shellpair const P,
638 struct simint_multi_shellpair const Q,
639 double screen_tol,
640 double * const restrict work,
641 double * const restrict INT__f_p_p_h)
642 {
643 double Q_AB[3*Q.nshell12];
644 struct simint_multi_shellpair Q_tmp = Q;
645 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
646 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
647 Q_tmp.AB_x = Q_AB;
648 Q_tmp.AB_y = Q_AB + Q.nshell12;
649 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
650
651 for(int i = 0; i < Q.nshell12; i++)
652 {
653 Q_tmp.AB_x[i] = -Q.AB_x[i];
654 Q_tmp.AB_y[i] = -Q.AB_y[i];
655 Q_tmp.AB_z[i] = -Q.AB_z[i];
656 }
657
658 int ret = ostei_f_p_h_p(P, Q_tmp, screen_tol, work, INT__f_p_p_h);
659 double buffer[1890] SIMINT_ALIGN_ARRAY_DBL;
660
661 for(int q = 0; q < ret; q++)
662 {
663 int idx = 0;
664 for(int a = 0; a < 10; ++a)
665 for(int b = 0; b < 3; ++b)
666 for(int c = 0; c < 3; ++c)
667 for(int d = 0; d < 21; ++d)
668 buffer[idx++] = INT__f_p_p_h[q*1890+a*189+b*63+d*3+c];
669
670 memcpy(INT__f_p_p_h+q*1890, buffer, 1890*sizeof(double));
671 }
672
673 return ret;
674 }
675
ostei_p_f_p_h(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__p_f_p_h)676 int ostei_p_f_p_h(struct simint_multi_shellpair const P,
677 struct simint_multi_shellpair const Q,
678 double screen_tol,
679 double * const restrict work,
680 double * const restrict INT__p_f_p_h)
681 {
682 double P_AB[3*P.nshell12];
683 struct simint_multi_shellpair P_tmp = P;
684 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
685 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
686 P_tmp.AB_x = P_AB;
687 P_tmp.AB_y = P_AB + P.nshell12;
688 P_tmp.AB_z = P_AB + 2*P.nshell12;
689
690 for(int i = 0; i < P.nshell12; i++)
691 {
692 P_tmp.AB_x[i] = -P.AB_x[i];
693 P_tmp.AB_y[i] = -P.AB_y[i];
694 P_tmp.AB_z[i] = -P.AB_z[i];
695 }
696
697 double Q_AB[3*Q.nshell12];
698 struct simint_multi_shellpair Q_tmp = Q;
699 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
700 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
701 Q_tmp.AB_x = Q_AB;
702 Q_tmp.AB_y = Q_AB + Q.nshell12;
703 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
704
705 for(int i = 0; i < Q.nshell12; i++)
706 {
707 Q_tmp.AB_x[i] = -Q.AB_x[i];
708 Q_tmp.AB_y[i] = -Q.AB_y[i];
709 Q_tmp.AB_z[i] = -Q.AB_z[i];
710 }
711
712 int ret = ostei_f_p_h_p(P_tmp, Q_tmp, screen_tol, work, INT__p_f_p_h);
713 double buffer[1890] SIMINT_ALIGN_ARRAY_DBL;
714
715 for(int q = 0; q < ret; q++)
716 {
717 int idx = 0;
718 for(int a = 0; a < 3; ++a)
719 for(int b = 0; b < 10; ++b)
720 for(int c = 0; c < 3; ++c)
721 for(int d = 0; d < 21; ++d)
722 buffer[idx++] = INT__p_f_p_h[q*1890+b*189+a*63+d*3+c];
723
724 memcpy(INT__p_f_p_h+q*1890, buffer, 1890*sizeof(double));
725 }
726
727 return ret;
728 }
729
730