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_g_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__g_f_s_s)8 int ostei_g_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__g_f_s_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__g_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__g_s_s_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__h_s_s_s = work + (SIMINT_NSHELL_SIMD * 15);
30 double * const INT__i_s_s_s = work + (SIMINT_NSHELL_SIMD * 36);
31 double * const INT__k_s_s_s = work + (SIMINT_NSHELL_SIMD * 64);
32 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*100);
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 + 8;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 29;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 65;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 115;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 175;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 238;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_s_s = primwork + 294;
41 double * const hrrwork = (double *)(primwork + 330);
42 double * const HRR_INT__g_p_s_s = hrrwork + 0;
43 double * const HRR_INT__g_d_s_s = hrrwork + 45;
44 double * const HRR_INT__h_p_s_s = hrrwork + 135;
45 double * const HRR_INT__h_d_s_s = hrrwork + 198;
46 double * const HRR_INT__i_p_s_s = hrrwork + 324;
47
48
49 // Create constants
50 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
51 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
52 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
53 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
54 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
55 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
56 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
57
58
59 ////////////////////////////////////////
60 // Loop over shells and primitives
61 ////////////////////////////////////////
62
63 real_abcd = 0;
64 istart = 0;
65 for(ab = 0; ab < P.nshell12_clip; ++ab)
66 {
67 const int iend = istart + P.nprim12[ab];
68
69 cd = 0;
70 jstart = 0;
71
72 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
73 {
74 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
75 int jend = jstart;
76 for(i = 0; i < nshellbatch; i++)
77 jend += Q.nprim12[cd+i];
78
79 // Clear the beginning of the workspace (where we are accumulating integrals)
80 memset(work, 0, SIMINT_NSHELL_SIMD * 100 * sizeof(double));
81 abcd = 0;
82
83
84 for(i = istart; i < iend; ++i)
85 {
86 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
87
88 if(check_screen)
89 {
90 // Skip this whole thing if always insignificant
91 if((P.screen[i] * Q.screen_max) < screen_tol)
92 continue;
93 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
94 }
95
96 icd = 0;
97 iprimcd = 0;
98 nprim_icd = Q.nprim12[cd];
99 double * restrict PRIM_PTR_INT__g_s_s_s = INT__g_s_s_s + abcd * 15;
100 double * restrict PRIM_PTR_INT__h_s_s_s = INT__h_s_s_s + abcd * 21;
101 double * restrict PRIM_PTR_INT__i_s_s_s = INT__i_s_s_s + abcd * 28;
102 double * restrict PRIM_PTR_INT__k_s_s_s = INT__k_s_s_s + abcd * 36;
103
104
105
106 // Load these one per loop over i
107 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
108 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
109 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
110
111 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
112
113 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
114 {
115 // calculate the shell offsets
116 // these are the offset from the shell pointed to by cd
117 // for each element
118 int shelloffsets[SIMINT_SIMD_LEN] = {0};
119 int lastoffset = 0;
120 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
121
122 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
123 {
124 // Handle if the first element of the vector is a new shell
125 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
126 {
127 nprim_icd += Q.nprim12[cd + (++icd)];
128 PRIM_PTR_INT__g_s_s_s += 15;
129 PRIM_PTR_INT__h_s_s_s += 21;
130 PRIM_PTR_INT__i_s_s_s += 28;
131 PRIM_PTR_INT__k_s_s_s += 36;
132 }
133 iprimcd++;
134 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
135 {
136 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
137 {
138 shelloffsets[n] = shelloffsets[n-1] + 1;
139 lastoffset++;
140 nprim_icd += Q.nprim12[cd + (++icd)];
141 }
142 else
143 shelloffsets[n] = shelloffsets[n-1];
144 iprimcd++;
145 }
146 }
147 else
148 iprimcd += SIMINT_SIMD_LEN;
149
150 // Do we have to compute this vector (or has it been screened out)?
151 // (not_screened != 0 means we have to do this vector)
152 if(check_screen)
153 {
154 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
155 if(vmax < screen_tol)
156 {
157 PRIM_PTR_INT__g_s_s_s += lastoffset*15;
158 PRIM_PTR_INT__h_s_s_s += lastoffset*21;
159 PRIM_PTR_INT__i_s_s_s += lastoffset*28;
160 PRIM_PTR_INT__k_s_s_s += lastoffset*36;
161 continue;
162 }
163 }
164
165 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
166 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
167 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
168 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
169
170
171 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
172 SIMINT_DBLTYPE PQ[3];
173 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
174 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
175 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
176 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
177 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
178 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
179
180 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
181 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
182 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
183 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
184 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
185 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
186
187 // NOTE: Minus sign!
188 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
189 SIMINT_DBLTYPE aop_PQ[3];
190 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
191 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
192 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
193
194
195 //////////////////////////////////////////////
196 // Fjt function section
197 // Maximum v value: 7
198 //////////////////////////////////////////////
199 // The parameter to the Fjt function
200 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
201
202
203 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
204
205
206 boys_F_split(PRIM_INT__s_s_s_s, F_x, 7);
207 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
208 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
209 for(n = 0; n <= 7; n++)
210 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
211
212 //////////////////////////////////////////////
213 // Primitive integrals: Vertical recurrance
214 //////////////////////////////////////////////
215
216 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
217 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
218 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
219 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
220 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
221 const SIMINT_DBLTYPE vrr_const_6_over_2p = SIMINT_MUL(const_6, one_over_2p);
222
223
224
225 // Forming PRIM_INT__p_s_s_s[7 * 3];
226 for(n = 0; n < 7; ++n) // loop over orders of auxiliary function
227 {
228
229 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
230 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]);
231
232 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
233 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]);
234
235 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
236 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]);
237
238 }
239
240
241
242 // Forming PRIM_INT__d_s_s_s[6 * 6];
243 for(n = 0; n < 6; ++n) // loop over orders of auxiliary function
244 {
245
246 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
247 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]);
248 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]);
249
250 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
251 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]);
252 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]);
253
254 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
255 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]);
256 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]);
257
258 }
259
260
261
262 // Forming PRIM_INT__f_s_s_s[5 * 10];
263 for(n = 0; n < 5; ++n) // loop over orders of auxiliary function
264 {
265
266 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
267 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]);
268 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]);
269
270 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
271 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]);
272
273 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
274 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]);
275
276 PRIM_INT__f_s_s_s[n * 10 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
277 PRIM_INT__f_s_s_s[n * 10 + 3] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__d_s_s_s[(n+1) * 6 + 3], PRIM_INT__f_s_s_s[n * 10 + 3]);
278
279 PRIM_INT__f_s_s_s[n * 10 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
280 PRIM_INT__f_s_s_s[n * 10 + 5] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__d_s_s_s[(n+1) * 6 + 5], PRIM_INT__f_s_s_s[n * 10 + 5]);
281
282 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
283 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]);
284 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]);
285
286 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
287 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]);
288
289 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
290 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]);
291 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]);
292
293 }
294
295
296 VRR_I_g_s_s_s(
297 PRIM_INT__g_s_s_s,
298 PRIM_INT__f_s_s_s,
299 PRIM_INT__d_s_s_s,
300 P_PA,
301 a_over_p,
302 aop_PQ,
303 one_over_2p,
304 4);
305
306
307 VRR_I_h_s_s_s(
308 PRIM_INT__h_s_s_s,
309 PRIM_INT__g_s_s_s,
310 PRIM_INT__f_s_s_s,
311 P_PA,
312 a_over_p,
313 aop_PQ,
314 one_over_2p,
315 3);
316
317
318 ostei_general_vrr1_I(6, 2,
319 one_over_2p, a_over_p, aop_PQ, P_PA,
320 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
321
322
323 ostei_general_vrr1_I(7, 1,
324 one_over_2p, a_over_p, aop_PQ, P_PA,
325 PRIM_INT__i_s_s_s, PRIM_INT__h_s_s_s, PRIM_INT__k_s_s_s);
326
327
328
329
330 ////////////////////////////////////
331 // Accumulate contracted integrals
332 ////////////////////////////////////
333 if(lastoffset == 0)
334 {
335 contract_all(15, PRIM_INT__g_s_s_s, PRIM_PTR_INT__g_s_s_s);
336 contract_all(21, PRIM_INT__h_s_s_s, PRIM_PTR_INT__h_s_s_s);
337 contract_all(28, PRIM_INT__i_s_s_s, PRIM_PTR_INT__i_s_s_s);
338 contract_all(36, PRIM_INT__k_s_s_s, PRIM_PTR_INT__k_s_s_s);
339 }
340 else
341 {
342 contract(15, shelloffsets, PRIM_INT__g_s_s_s, PRIM_PTR_INT__g_s_s_s);
343 contract(21, shelloffsets, PRIM_INT__h_s_s_s, PRIM_PTR_INT__h_s_s_s);
344 contract(28, shelloffsets, PRIM_INT__i_s_s_s, PRIM_PTR_INT__i_s_s_s);
345 contract(36, shelloffsets, PRIM_INT__k_s_s_s, PRIM_PTR_INT__k_s_s_s);
346 PRIM_PTR_INT__g_s_s_s += lastoffset*15;
347 PRIM_PTR_INT__h_s_s_s += lastoffset*21;
348 PRIM_PTR_INT__i_s_s_s += lastoffset*28;
349 PRIM_PTR_INT__k_s_s_s += lastoffset*36;
350 }
351
352 } // close loop over j
353 } // close loop over i
354
355 //Advance to the next batch
356 jstart = SIMINT_SIMD_ROUND(jend);
357
358 //////////////////////////////////////////////
359 // Contracted integrals: Horizontal recurrance
360 //////////////////////////////////////////////
361
362
363 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
364
365
366 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
367 {
368
369 // set up HRR pointers
370 double const * restrict HRR_INT__g_s_s_s = INT__g_s_s_s + abcd * 15;
371 double const * restrict HRR_INT__h_s_s_s = INT__h_s_s_s + abcd * 21;
372 double const * restrict HRR_INT__i_s_s_s = INT__i_s_s_s + abcd * 28;
373 double const * restrict HRR_INT__k_s_s_s = INT__k_s_s_s + abcd * 36;
374 double * restrict HRR_INT__g_f_s_s = INT__g_f_s_s + real_abcd * 150;
375
376 // form INT__g_p_s_s
377 HRR_J_g_p(
378 HRR_INT__g_p_s_s,
379 HRR_INT__g_s_s_s,
380 HRR_INT__h_s_s_s,
381 hAB, 1);
382
383 // form INT__h_p_s_s
384 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);
385
386 // form INT__i_p_s_s
387 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);
388
389 // form INT__g_d_s_s
390 ostei_general_hrr_J(4, 2, 0, 0, hAB, HRR_INT__h_p_s_s, HRR_INT__g_p_s_s, HRR_INT__g_d_s_s);
391
392 // form INT__h_d_s_s
393 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);
394
395 // form INT__g_f_s_s
396 ostei_general_hrr_J(4, 3, 0, 0, hAB, HRR_INT__h_d_s_s, HRR_INT__g_d_s_s, HRR_INT__g_f_s_s);
397
398
399 } // close HRR loop
400
401
402 } // close loop cdbatch
403
404 istart = iend;
405 } // close loop over ab
406
407 return P.nshell12_clip * Q.nshell12_clip;
408 }
409
ostei_f_g_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_g_s_s)410 int ostei_f_g_s_s(struct simint_multi_shellpair const P,
411 struct simint_multi_shellpair const Q,
412 double screen_tol,
413 double * const restrict work,
414 double * const restrict INT__f_g_s_s)
415 {
416 double P_AB[3*P.nshell12];
417 struct simint_multi_shellpair P_tmp = P;
418 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
419 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
420 P_tmp.AB_x = P_AB;
421 P_tmp.AB_y = P_AB + P.nshell12;
422 P_tmp.AB_z = P_AB + 2*P.nshell12;
423
424 for(int i = 0; i < P.nshell12; i++)
425 {
426 P_tmp.AB_x[i] = -P.AB_x[i];
427 P_tmp.AB_y[i] = -P.AB_y[i];
428 P_tmp.AB_z[i] = -P.AB_z[i];
429 }
430
431 int ret = ostei_g_f_s_s(P_tmp, Q, screen_tol, work, INT__f_g_s_s);
432 double buffer[150] SIMINT_ALIGN_ARRAY_DBL;
433
434 for(int q = 0; q < ret; q++)
435 {
436 int idx = 0;
437 for(int a = 0; a < 10; ++a)
438 for(int b = 0; b < 15; ++b)
439 for(int c = 0; c < 1; ++c)
440 for(int d = 0; d < 1; ++d)
441 buffer[idx++] = INT__f_g_s_s[q*150+b*10+a*1+c*1+d];
442
443 memcpy(INT__f_g_s_s+q*150, buffer, 150*sizeof(double));
444 }
445
446 return ret;
447 }
448
449