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_h_f_s_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__h_f_s_s)8 int ostei_h_f_s_s(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__h_f_s_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__h_f_s_s);
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
27 // partition workspace
28 double * const INT__h_s_s_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__i_s_s_s = work + (SIMINT_NSHELL_SIMD * 21);
30 double * const INT__k_s_s_s = work + (SIMINT_NSHELL_SIMD * 49);
31 double * const INT__l_s_s_s = work + (SIMINT_NSHELL_SIMD * 85);
32 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*130);
33 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_s_s = primwork + 9;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 33;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 75;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 135;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 210;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 294;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_s_s = primwork + 378;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_s_s = primwork + 450;
42 double * const hrrwork = (double *)(primwork + 495);
43 double * const HRR_INT__h_p_s_s = hrrwork + 0;
44 double * const HRR_INT__h_d_s_s = hrrwork + 63;
45 double * const HRR_INT__i_p_s_s = hrrwork + 189;
46 double * const HRR_INT__i_d_s_s = hrrwork + 273;
47 double * const HRR_INT__k_p_s_s = hrrwork + 441;
48
49
50 // Create constants
51 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
52 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
53 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
54 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
55 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
56 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
57 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
58 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
59
60
61 ////////////////////////////////////////
62 // Loop over shells and primitives
63 ////////////////////////////////////////
64
65 real_abcd = 0;
66 istart = 0;
67 for(ab = 0; ab < P.nshell12_clip; ++ab)
68 {
69 const int iend = istart + P.nprim12[ab];
70
71 cd = 0;
72 jstart = 0;
73
74 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
75 {
76 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
77 int jend = jstart;
78 for(i = 0; i < nshellbatch; i++)
79 jend += Q.nprim12[cd+i];
80
81 // Clear the beginning of the workspace (where we are accumulating integrals)
82 memset(work, 0, SIMINT_NSHELL_SIMD * 130 * sizeof(double));
83 abcd = 0;
84
85
86 for(i = istart; i < iend; ++i)
87 {
88 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
89
90 if(check_screen)
91 {
92 // Skip this whole thing if always insignificant
93 if((P.screen[i] * Q.screen_max) < screen_tol)
94 continue;
95 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
96 }
97
98 icd = 0;
99 iprimcd = 0;
100 nprim_icd = Q.nprim12[cd];
101 double * restrict PRIM_PTR_INT__h_s_s_s = INT__h_s_s_s + abcd * 21;
102 double * restrict PRIM_PTR_INT__i_s_s_s = INT__i_s_s_s + abcd * 28;
103 double * restrict PRIM_PTR_INT__k_s_s_s = INT__k_s_s_s + abcd * 36;
104 double * restrict PRIM_PTR_INT__l_s_s_s = INT__l_s_s_s + abcd * 45;
105
106
107
108 // Load these one per loop over i
109 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
110 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
111 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
112
113 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
114
115 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
116 {
117 // calculate the shell offsets
118 // these are the offset from the shell pointed to by cd
119 // for each element
120 int shelloffsets[SIMINT_SIMD_LEN] = {0};
121 int lastoffset = 0;
122 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
123
124 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
125 {
126 // Handle if the first element of the vector is a new shell
127 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
128 {
129 nprim_icd += Q.nprim12[cd + (++icd)];
130 PRIM_PTR_INT__h_s_s_s += 21;
131 PRIM_PTR_INT__i_s_s_s += 28;
132 PRIM_PTR_INT__k_s_s_s += 36;
133 PRIM_PTR_INT__l_s_s_s += 45;
134 }
135 iprimcd++;
136 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
137 {
138 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
139 {
140 shelloffsets[n] = shelloffsets[n-1] + 1;
141 lastoffset++;
142 nprim_icd += Q.nprim12[cd + (++icd)];
143 }
144 else
145 shelloffsets[n] = shelloffsets[n-1];
146 iprimcd++;
147 }
148 }
149 else
150 iprimcd += SIMINT_SIMD_LEN;
151
152 // Do we have to compute this vector (or has it been screened out)?
153 // (not_screened != 0 means we have to do this vector)
154 if(check_screen)
155 {
156 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
157 if(vmax < screen_tol)
158 {
159 PRIM_PTR_INT__h_s_s_s += lastoffset*21;
160 PRIM_PTR_INT__i_s_s_s += lastoffset*28;
161 PRIM_PTR_INT__k_s_s_s += lastoffset*36;
162 PRIM_PTR_INT__l_s_s_s += lastoffset*45;
163 continue;
164 }
165 }
166
167 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
168 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
169 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
170 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
171
172
173 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
174 SIMINT_DBLTYPE PQ[3];
175 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
176 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
177 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
178 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
179 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
180 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
181
182 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
183 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
184 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
185 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
186 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
187 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
188
189 // NOTE: Minus sign!
190 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
191 SIMINT_DBLTYPE aop_PQ[3];
192 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
193 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
194 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
195
196
197 //////////////////////////////////////////////
198 // Fjt function section
199 // Maximum v value: 8
200 //////////////////////////////////////////////
201 // The parameter to the Fjt function
202 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
203
204
205 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
206
207
208 boys_F_split(PRIM_INT__s_s_s_s, F_x, 8);
209 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
210 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
211 for(n = 0; n <= 8; n++)
212 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
213
214 //////////////////////////////////////////////
215 // Primitive integrals: Vertical recurrance
216 //////////////////////////////////////////////
217
218 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
219 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
220 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
221 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
222 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
223 const SIMINT_DBLTYPE vrr_const_6_over_2p = SIMINT_MUL(const_6, one_over_2p);
224 const SIMINT_DBLTYPE vrr_const_7_over_2p = SIMINT_MUL(const_7, one_over_2p);
225
226
227
228 // Forming PRIM_INT__p_s_s_s[8 * 3];
229 for(n = 0; n < 8; ++n) // loop over orders of auxiliary function
230 {
231
232 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
233 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__p_s_s_s[n * 3 + 0]);
234
235 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
236 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__p_s_s_s[n * 3 + 1]);
237
238 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
239 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__p_s_s_s[n * 3 + 2]);
240
241 }
242
243
244
245 // Forming PRIM_INT__d_s_s_s[7 * 6];
246 for(n = 0; n < 7; ++n) // loop over orders of auxiliary function
247 {
248
249 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
250 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__d_s_s_s[n * 6 + 0]);
251 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_FMADD( vrr_const_1_over_2p, SIMINT_FMADD(a_over_p, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__d_s_s_s[n * 6 + 0]);
252
253 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
254 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__p_s_s_s[(n+1) * 3 + 1], PRIM_INT__d_s_s_s[n * 6 + 3]);
255 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_FMADD( vrr_const_1_over_2p, SIMINT_FMADD(a_over_p, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__d_s_s_s[n * 6 + 3]);
256
257 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
258 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__p_s_s_s[(n+1) * 3 + 2], PRIM_INT__d_s_s_s[n * 6 + 5]);
259 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_FMADD( vrr_const_1_over_2p, SIMINT_FMADD(a_over_p, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__d_s_s_s[n * 6 + 5]);
260
261 }
262
263
264
265 // Forming PRIM_INT__f_s_s_s[6 * 10];
266 for(n = 0; n < 6; ++n) // loop over orders of auxiliary function
267 {
268
269 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
270 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__d_s_s_s[(n+1) * 6 + 0], PRIM_INT__f_s_s_s[n * 10 + 0]);
271 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_FMADD( vrr_const_2_over_2p, SIMINT_FMADD(a_over_p, PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__p_s_s_s[n * 3 + 0]), PRIM_INT__f_s_s_s[n * 10 + 0]);
272
273 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
274 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__d_s_s_s[(n+1) * 6 + 0], PRIM_INT__f_s_s_s[n * 10 + 1]);
275
276 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
277 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__d_s_s_s[(n+1) * 6 + 0], PRIM_INT__f_s_s_s[n * 10 + 2]);
278
279 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
280 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__d_s_s_s[(n+1) * 6 + 3], PRIM_INT__f_s_s_s[n * 10 + 6]);
281 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_FMADD( vrr_const_2_over_2p, SIMINT_FMADD(a_over_p, PRIM_INT__p_s_s_s[(n+1) * 3 + 1], PRIM_INT__p_s_s_s[n * 3 + 1]), PRIM_INT__f_s_s_s[n * 10 + 6]);
282
283 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
284 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__d_s_s_s[(n+1) * 6 + 3], PRIM_INT__f_s_s_s[n * 10 + 7]);
285
286 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
287 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__d_s_s_s[(n+1) * 6 + 5], PRIM_INT__f_s_s_s[n * 10 + 9]);
288 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_FMADD( vrr_const_2_over_2p, SIMINT_FMADD(a_over_p, PRIM_INT__p_s_s_s[(n+1) * 3 + 2], PRIM_INT__p_s_s_s[n * 3 + 2]), PRIM_INT__f_s_s_s[n * 10 + 9]);
289
290 }
291
292
293 VRR_I_g_s_s_s(
294 PRIM_INT__g_s_s_s,
295 PRIM_INT__f_s_s_s,
296 PRIM_INT__d_s_s_s,
297 P_PA,
298 a_over_p,
299 aop_PQ,
300 one_over_2p,
301 5);
302
303
304 VRR_I_h_s_s_s(
305 PRIM_INT__h_s_s_s,
306 PRIM_INT__g_s_s_s,
307 PRIM_INT__f_s_s_s,
308 P_PA,
309 a_over_p,
310 aop_PQ,
311 one_over_2p,
312 4);
313
314
315 ostei_general_vrr1_I(6, 3,
316 one_over_2p, a_over_p, aop_PQ, P_PA,
317 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
318
319
320 ostei_general_vrr1_I(7, 2,
321 one_over_2p, a_over_p, aop_PQ, P_PA,
322 PRIM_INT__i_s_s_s, PRIM_INT__h_s_s_s, PRIM_INT__k_s_s_s);
323
324
325 ostei_general_vrr1_I(8, 1,
326 one_over_2p, a_over_p, aop_PQ, P_PA,
327 PRIM_INT__k_s_s_s, PRIM_INT__i_s_s_s, PRIM_INT__l_s_s_s);
328
329
330
331
332 ////////////////////////////////////
333 // Accumulate contracted integrals
334 ////////////////////////////////////
335 if(lastoffset == 0)
336 {
337 contract_all(21, PRIM_INT__h_s_s_s, PRIM_PTR_INT__h_s_s_s);
338 contract_all(28, PRIM_INT__i_s_s_s, PRIM_PTR_INT__i_s_s_s);
339 contract_all(36, PRIM_INT__k_s_s_s, PRIM_PTR_INT__k_s_s_s);
340 contract_all(45, PRIM_INT__l_s_s_s, PRIM_PTR_INT__l_s_s_s);
341 }
342 else
343 {
344 contract(21, shelloffsets, PRIM_INT__h_s_s_s, PRIM_PTR_INT__h_s_s_s);
345 contract(28, shelloffsets, PRIM_INT__i_s_s_s, PRIM_PTR_INT__i_s_s_s);
346 contract(36, shelloffsets, PRIM_INT__k_s_s_s, PRIM_PTR_INT__k_s_s_s);
347 contract(45, shelloffsets, PRIM_INT__l_s_s_s, PRIM_PTR_INT__l_s_s_s);
348 PRIM_PTR_INT__h_s_s_s += lastoffset*21;
349 PRIM_PTR_INT__i_s_s_s += lastoffset*28;
350 PRIM_PTR_INT__k_s_s_s += lastoffset*36;
351 PRIM_PTR_INT__l_s_s_s += lastoffset*45;
352 }
353
354 } // close loop over j
355 } // close loop over i
356
357 //Advance to the next batch
358 jstart = SIMINT_SIMD_ROUND(jend);
359
360 //////////////////////////////////////////////
361 // Contracted integrals: Horizontal recurrance
362 //////////////////////////////////////////////
363
364
365 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
366
367
368 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
369 {
370
371 // set up HRR pointers
372 double const * restrict HRR_INT__h_s_s_s = INT__h_s_s_s + abcd * 21;
373 double const * restrict HRR_INT__i_s_s_s = INT__i_s_s_s + abcd * 28;
374 double const * restrict HRR_INT__k_s_s_s = INT__k_s_s_s + abcd * 36;
375 double const * restrict HRR_INT__l_s_s_s = INT__l_s_s_s + abcd * 45;
376 double * restrict HRR_INT__h_f_s_s = INT__h_f_s_s + real_abcd * 210;
377
378 // form INT__h_p_s_s
379 ostei_general_hrr_J(5, 1, 0, 0, hAB, HRR_INT__i_s_s_s, HRR_INT__h_s_s_s, HRR_INT__h_p_s_s);
380
381 // form INT__i_p_s_s
382 ostei_general_hrr_J(6, 1, 0, 0, hAB, HRR_INT__k_s_s_s, HRR_INT__i_s_s_s, HRR_INT__i_p_s_s);
383
384 // form INT__k_p_s_s
385 ostei_general_hrr_J(7, 1, 0, 0, hAB, HRR_INT__l_s_s_s, HRR_INT__k_s_s_s, HRR_INT__k_p_s_s);
386
387 // form INT__h_d_s_s
388 ostei_general_hrr_J(5, 2, 0, 0, hAB, HRR_INT__i_p_s_s, HRR_INT__h_p_s_s, HRR_INT__h_d_s_s);
389
390 // form INT__i_d_s_s
391 ostei_general_hrr_J(6, 2, 0, 0, hAB, HRR_INT__k_p_s_s, HRR_INT__i_p_s_s, HRR_INT__i_d_s_s);
392
393 // form INT__h_f_s_s
394 ostei_general_hrr_J(5, 3, 0, 0, hAB, HRR_INT__i_d_s_s, HRR_INT__h_d_s_s, HRR_INT__h_f_s_s);
395
396
397 } // close HRR loop
398
399
400 } // close loop cdbatch
401
402 istart = iend;
403 } // close loop over ab
404
405 return P.nshell12_clip * Q.nshell12_clip;
406 }
407
ostei_f_h_s_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__f_h_s_s)408 int ostei_f_h_s_s(struct simint_multi_shellpair const P,
409 struct simint_multi_shellpair const Q,
410 double screen_tol,
411 double * const restrict work,
412 double * const restrict INT__f_h_s_s)
413 {
414 double P_AB[3*P.nshell12];
415 struct simint_multi_shellpair P_tmp = P;
416 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
417 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
418 P_tmp.AB_x = P_AB;
419 P_tmp.AB_y = P_AB + P.nshell12;
420 P_tmp.AB_z = P_AB + 2*P.nshell12;
421
422 for(int i = 0; i < P.nshell12; i++)
423 {
424 P_tmp.AB_x[i] = -P.AB_x[i];
425 P_tmp.AB_y[i] = -P.AB_y[i];
426 P_tmp.AB_z[i] = -P.AB_z[i];
427 }
428
429 int ret = ostei_h_f_s_s(P_tmp, Q, screen_tol, work, INT__f_h_s_s);
430 double buffer[210] SIMINT_ALIGN_ARRAY_DBL;
431
432 for(int q = 0; q < ret; q++)
433 {
434 int idx = 0;
435 for(int a = 0; a < 10; ++a)
436 for(int b = 0; b < 21; ++b)
437 for(int c = 0; c < 1; ++c)
438 for(int d = 0; d < 1; ++d)
439 buffer[idx++] = INT__f_h_s_s[q*210+b*10+a*1+c*1+d];
440
441 memcpy(INT__f_h_s_s+q*210, buffer, 210*sizeof(double));
442 }
443
444 return ret;
445 }
446
447