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_p_i_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_p_i_s)8 int ostei_g_p_i_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_p_i_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__g_p_i_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_i_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__h_s_i_s = work + (SIMINT_NSHELL_SIMD * 420);
30 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*1008);
31 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
32 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_p_s = primwork + 12;
33 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_d_s = primwork + 45;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_f_s = primwork + 105;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_g_s = primwork + 195;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_h_s = primwork + 315;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_i_s = primwork + 462;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_d_s = primwork + 630;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_f_s = primwork + 720;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_g_s = primwork + 870;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_h_s = primwork + 1095;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_i_s = primwork + 1410;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_f_s = primwork + 1830;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_g_s = primwork + 2070;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_h_s = primwork + 2430;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_i_s = primwork + 2934;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_g_s = primwork + 3606;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_h_s = primwork + 4056;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_i_s = primwork + 4686;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_h_s = primwork + 5526;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_i_s = primwork + 6156;
52 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_i_s = primwork + 6996;
53 double * const hrrwork = (double *)(primwork + 7584);
54
55
56 // Create constants
57 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
58 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
59 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
60 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
61 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
62 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
63 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
64
65
66 ////////////////////////////////////////
67 // Loop over shells and primitives
68 ////////////////////////////////////////
69
70 real_abcd = 0;
71 istart = 0;
72 for(ab = 0; ab < P.nshell12_clip; ++ab)
73 {
74 const int iend = istart + P.nprim12[ab];
75
76 cd = 0;
77 jstart = 0;
78
79 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
80 {
81 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
82 int jend = jstart;
83 for(i = 0; i < nshellbatch; i++)
84 jend += Q.nprim12[cd+i];
85
86 // Clear the beginning of the workspace (where we are accumulating integrals)
87 memset(work, 0, SIMINT_NSHELL_SIMD * 1008 * sizeof(double));
88 abcd = 0;
89
90
91 for(i = istart; i < iend; ++i)
92 {
93 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
94
95 if(check_screen)
96 {
97 // Skip this whole thing if always insignificant
98 if((P.screen[i] * Q.screen_max) < screen_tol)
99 continue;
100 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
101 }
102
103 icd = 0;
104 iprimcd = 0;
105 nprim_icd = Q.nprim12[cd];
106 double * restrict PRIM_PTR_INT__g_s_i_s = INT__g_s_i_s + abcd * 420;
107 double * restrict PRIM_PTR_INT__h_s_i_s = INT__h_s_i_s + abcd * 588;
108
109
110
111 // Load these one per loop over i
112 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
113 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
114 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
115
116 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
117
118 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
119 {
120 // calculate the shell offsets
121 // these are the offset from the shell pointed to by cd
122 // for each element
123 int shelloffsets[SIMINT_SIMD_LEN] = {0};
124 int lastoffset = 0;
125 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
126
127 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
128 {
129 // Handle if the first element of the vector is a new shell
130 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
131 {
132 nprim_icd += Q.nprim12[cd + (++icd)];
133 PRIM_PTR_INT__g_s_i_s += 420;
134 PRIM_PTR_INT__h_s_i_s += 588;
135 }
136 iprimcd++;
137 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
138 {
139 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
140 {
141 shelloffsets[n] = shelloffsets[n-1] + 1;
142 lastoffset++;
143 nprim_icd += Q.nprim12[cd + (++icd)];
144 }
145 else
146 shelloffsets[n] = shelloffsets[n-1];
147 iprimcd++;
148 }
149 }
150 else
151 iprimcd += SIMINT_SIMD_LEN;
152
153 // Do we have to compute this vector (or has it been screened out)?
154 // (not_screened != 0 means we have to do this vector)
155 if(check_screen)
156 {
157 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
158 if(vmax < screen_tol)
159 {
160 PRIM_PTR_INT__g_s_i_s += lastoffset*420;
161 PRIM_PTR_INT__h_s_i_s += lastoffset*588;
162 continue;
163 }
164 }
165
166 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
167 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
168 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
169 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
170
171
172 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
173 SIMINT_DBLTYPE PQ[3];
174 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
175 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
176 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
177 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
178 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
179 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
180
181 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
182 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
183 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
184 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
185 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
186 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
187 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
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 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
197 SIMINT_DBLTYPE aoq_PQ[3];
198 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
199 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
200 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
201 // Put a minus sign here so we don't have to in RR routines
202 a_over_q = SIMINT_NEG(a_over_q);
203
204
205 //////////////////////////////////////////////
206 // Fjt function section
207 // Maximum v value: 11
208 //////////////////////////////////////////////
209 // The parameter to the Fjt function
210 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
211
212
213 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
214
215
216 boys_F_split(PRIM_INT__s_s_s_s, F_x, 11);
217 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
218 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
219 for(n = 0; n <= 11; n++)
220 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
221
222 //////////////////////////////////////////////
223 // Primitive integrals: Vertical recurrance
224 //////////////////////////////////////////////
225
226 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
227 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
228 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
229 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
230 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
231 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
232 const SIMINT_DBLTYPE vrr_const_3_over_2q = SIMINT_MUL(const_3, one_over_2q);
233 const SIMINT_DBLTYPE vrr_const_4_over_2q = SIMINT_MUL(const_4, one_over_2q);
234 const SIMINT_DBLTYPE vrr_const_5_over_2q = SIMINT_MUL(const_5, one_over_2q);
235 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
236 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
237 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
238 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
239 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
240 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
241
242
243
244 // Forming PRIM_INT__s_s_p_s[11 * 3];
245 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
246 {
247
248 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
249 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]);
250
251 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
252 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]);
253
254 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
255 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]);
256
257 }
258
259
260
261 // Forming PRIM_INT__s_s_d_s[10 * 6];
262 for(n = 0; n < 10; ++n) // loop over orders of auxiliary function
263 {
264
265 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_p_s[n * 3 + 0]);
266 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]);
267 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]);
268
269 PRIM_INT__s_s_d_s[n * 6 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 0]);
270 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]);
271
272 PRIM_INT__s_s_d_s[n * 6 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 0]);
273 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]);
274
275 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 1]);
276 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]);
277 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]);
278
279 PRIM_INT__s_s_d_s[n * 6 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 1]);
280 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]);
281
282 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 2]);
283 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]);
284 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]);
285
286 }
287
288
289
290 // Forming PRIM_INT__s_s_f_s[9 * 10];
291 for(n = 0; n < 9; ++n) // loop over orders of auxiliary function
292 {
293
294 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
295 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]);
296 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]);
297
298 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
299 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]);
300
301 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
302 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]);
303
304 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
305 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]);
306
307 PRIM_INT__s_s_f_s[n * 10 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
308 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]);
309
310 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
311 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]);
312
313 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
314 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]);
315 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]);
316
317 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
318 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]);
319
320 PRIM_INT__s_s_f_s[n * 10 + 8] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
321 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]);
322
323 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
324 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]);
325 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]);
326
327 }
328
329
330 VRR_K_s_s_g_s(
331 PRIM_INT__s_s_g_s,
332 PRIM_INT__s_s_f_s,
333 PRIM_INT__s_s_d_s,
334 Q_PA,
335 a_over_q,
336 aoq_PQ,
337 one_over_2q,
338 8);
339
340
341 VRR_K_s_s_h_s(
342 PRIM_INT__s_s_h_s,
343 PRIM_INT__s_s_g_s,
344 PRIM_INT__s_s_f_s,
345 Q_PA,
346 a_over_q,
347 aoq_PQ,
348 one_over_2q,
349 7);
350
351
352 ostei_general_vrr1_K(6, 6,
353 one_over_2q, a_over_q, aoq_PQ, Q_PA,
354 PRIM_INT__s_s_h_s, PRIM_INT__s_s_g_s, PRIM_INT__s_s_i_s);
355
356
357 ostei_general_vrr_I(1, 0, 6, 0, 5,
358 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
359 PRIM_INT__s_s_i_s, NULL, NULL, PRIM_INT__s_s_h_s, NULL, PRIM_INT__p_s_i_s);
360
361
362 ostei_general_vrr_I(1, 0, 5, 0, 5,
363 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
364 PRIM_INT__s_s_h_s, NULL, NULL, PRIM_INT__s_s_g_s, NULL, PRIM_INT__p_s_h_s);
365
366
367 ostei_general_vrr_I(2, 0, 6, 0, 4,
368 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
369 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);
370
371
372 VRR_I_p_s_g_s(
373 PRIM_INT__p_s_g_s,
374 PRIM_INT__s_s_g_s,
375 PRIM_INT__s_s_f_s,
376 P_PA,
377 aop_PQ,
378 one_over_2pq,
379 5);
380
381
382 ostei_general_vrr_I(2, 0, 5, 0, 4,
383 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
384 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);
385
386
387 ostei_general_vrr_I(3, 0, 6, 0, 3,
388 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
389 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);
390
391
392 VRR_I_p_s_f_s(
393 PRIM_INT__p_s_f_s,
394 PRIM_INT__s_s_f_s,
395 PRIM_INT__s_s_d_s,
396 P_PA,
397 aop_PQ,
398 one_over_2pq,
399 5);
400
401
402 ostei_general_vrr_I(2, 0, 4, 0, 4,
403 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
404 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);
405
406
407 ostei_general_vrr_I(3, 0, 5, 0, 3,
408 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
409 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);
410
411
412 ostei_general_vrr_I(4, 0, 6, 0, 2,
413 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
414 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);
415
416
417
418 // Forming PRIM_INT__p_s_d_s[5 * 18];
419 for(n = 0; n < 5; ++n) // loop over orders of auxiliary function
420 {
421
422 PRIM_INT__p_s_d_s[n * 18 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
423 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]);
424 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]);
425
426 PRIM_INT__p_s_d_s[n * 18 + 1] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 1]);
427 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]);
428 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]);
429
430 PRIM_INT__p_s_d_s[n * 18 + 2] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 2]);
431 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]);
432 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]);
433
434 PRIM_INT__p_s_d_s[n * 18 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
435 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]);
436
437 PRIM_INT__p_s_d_s[n * 18 + 4] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 4]);
438 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]);
439
440 PRIM_INT__p_s_d_s[n * 18 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
441 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]);
442
443 PRIM_INT__p_s_d_s[n * 18 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
444 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]);
445
446 PRIM_INT__p_s_d_s[n * 18 + 7] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 1]);
447 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]);
448 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]);
449
450 PRIM_INT__p_s_d_s[n * 18 + 8] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 2]);
451 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]);
452
453 PRIM_INT__p_s_d_s[n * 18 + 9] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
454 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]);
455 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]);
456
457 PRIM_INT__p_s_d_s[n * 18 + 10] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 4]);
458 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]);
459 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]);
460
461 PRIM_INT__p_s_d_s[n * 18 + 11] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
462 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]);
463
464 PRIM_INT__p_s_d_s[n * 18 + 12] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
465 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]);
466
467 PRIM_INT__p_s_d_s[n * 18 + 13] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
468 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]);
469
470 PRIM_INT__p_s_d_s[n * 18 + 14] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 2]);
471 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]);
472 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]);
473
474 PRIM_INT__p_s_d_s[n * 18 + 15] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
475 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]);
476
477 PRIM_INT__p_s_d_s[n * 18 + 16] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 4]);
478 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]);
479 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]);
480
481 PRIM_INT__p_s_d_s[n * 18 + 17] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
482 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]);
483 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]);
484
485 }
486
487
488 VRR_I_d_s_f_s(
489 PRIM_INT__d_s_f_s,
490 PRIM_INT__p_s_f_s,
491 PRIM_INT__s_s_f_s,
492 PRIM_INT__p_s_d_s,
493 P_PA,
494 a_over_p,
495 aop_PQ,
496 one_over_2p,
497 one_over_2pq,
498 4);
499
500
501 ostei_general_vrr_I(3, 0, 4, 0, 3,
502 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
503 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);
504
505
506 ostei_general_vrr_I(4, 0, 5, 0, 2,
507 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
508 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);
509
510
511 ostei_general_vrr_I(5, 0, 6, 0, 1,
512 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
513 PRIM_INT__g_s_i_s, PRIM_INT__f_s_i_s, NULL, PRIM_INT__g_s_h_s, NULL, PRIM_INT__h_s_i_s);
514
515
516
517
518 ////////////////////////////////////
519 // Accumulate contracted integrals
520 ////////////////////////////////////
521 if(lastoffset == 0)
522 {
523 contract_all(420, PRIM_INT__g_s_i_s, PRIM_PTR_INT__g_s_i_s);
524 contract_all(588, PRIM_INT__h_s_i_s, PRIM_PTR_INT__h_s_i_s);
525 }
526 else
527 {
528 contract(420, shelloffsets, PRIM_INT__g_s_i_s, PRIM_PTR_INT__g_s_i_s);
529 contract(588, shelloffsets, PRIM_INT__h_s_i_s, PRIM_PTR_INT__h_s_i_s);
530 PRIM_PTR_INT__g_s_i_s += lastoffset*420;
531 PRIM_PTR_INT__h_s_i_s += lastoffset*588;
532 }
533
534 } // close loop over j
535 } // close loop over i
536
537 //Advance to the next batch
538 jstart = SIMINT_SIMD_ROUND(jend);
539
540 //////////////////////////////////////////////
541 // Contracted integrals: Horizontal recurrance
542 //////////////////////////////////////////////
543
544
545 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
546
547
548 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
549 {
550
551 // set up HRR pointers
552 double const * restrict HRR_INT__g_s_i_s = INT__g_s_i_s + abcd * 420;
553 double const * restrict HRR_INT__h_s_i_s = INT__h_s_i_s + abcd * 588;
554 double * restrict HRR_INT__g_p_i_s = INT__g_p_i_s + real_abcd * 1260;
555
556 // form INT__g_p_i_s
557 HRR_J_g_p(
558 HRR_INT__g_p_i_s,
559 HRR_INT__g_s_i_s,
560 HRR_INT__h_s_i_s,
561 hAB, 28);
562
563
564 } // close HRR loop
565
566
567 } // close loop cdbatch
568
569 istart = iend;
570 } // close loop over ab
571
572 return P.nshell12_clip * Q.nshell12_clip;
573 }
574
ostei_p_g_i_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__p_g_i_s)575 int ostei_p_g_i_s(struct simint_multi_shellpair const P,
576 struct simint_multi_shellpair const Q,
577 double screen_tol,
578 double * const restrict work,
579 double * const restrict INT__p_g_i_s)
580 {
581 double P_AB[3*P.nshell12];
582 struct simint_multi_shellpair P_tmp = P;
583 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
584 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
585 P_tmp.AB_x = P_AB;
586 P_tmp.AB_y = P_AB + P.nshell12;
587 P_tmp.AB_z = P_AB + 2*P.nshell12;
588
589 for(int i = 0; i < P.nshell12; i++)
590 {
591 P_tmp.AB_x[i] = -P.AB_x[i];
592 P_tmp.AB_y[i] = -P.AB_y[i];
593 P_tmp.AB_z[i] = -P.AB_z[i];
594 }
595
596 int ret = ostei_g_p_i_s(P_tmp, Q, screen_tol, work, INT__p_g_i_s);
597 double buffer[1260] SIMINT_ALIGN_ARRAY_DBL;
598
599 for(int q = 0; q < ret; q++)
600 {
601 int idx = 0;
602 for(int a = 0; a < 3; ++a)
603 for(int b = 0; b < 15; ++b)
604 for(int c = 0; c < 28; ++c)
605 for(int d = 0; d < 1; ++d)
606 buffer[idx++] = INT__p_g_i_s[q*1260+b*84+a*28+c*1+d];
607
608 memcpy(INT__p_g_i_s+q*1260, buffer, 1260*sizeof(double));
609 }
610
611 return ret;
612 }
613
ostei_g_p_s_i(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__g_p_s_i)614 int ostei_g_p_s_i(struct simint_multi_shellpair const P,
615 struct simint_multi_shellpair const Q,
616 double screen_tol,
617 double * const restrict work,
618 double * const restrict INT__g_p_s_i)
619 {
620 double Q_AB[3*Q.nshell12];
621 struct simint_multi_shellpair Q_tmp = Q;
622 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
623 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
624 Q_tmp.AB_x = Q_AB;
625 Q_tmp.AB_y = Q_AB + Q.nshell12;
626 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
627
628 for(int i = 0; i < Q.nshell12; i++)
629 {
630 Q_tmp.AB_x[i] = -Q.AB_x[i];
631 Q_tmp.AB_y[i] = -Q.AB_y[i];
632 Q_tmp.AB_z[i] = -Q.AB_z[i];
633 }
634
635 int ret = ostei_g_p_i_s(P, Q_tmp, screen_tol, work, INT__g_p_s_i);
636 double buffer[1260] SIMINT_ALIGN_ARRAY_DBL;
637
638 for(int q = 0; q < ret; q++)
639 {
640 int idx = 0;
641 for(int a = 0; a < 15; ++a)
642 for(int b = 0; b < 3; ++b)
643 for(int c = 0; c < 1; ++c)
644 for(int d = 0; d < 28; ++d)
645 buffer[idx++] = INT__g_p_s_i[q*1260+a*84+b*28+d*1+c];
646
647 memcpy(INT__g_p_s_i+q*1260, buffer, 1260*sizeof(double));
648 }
649
650 return ret;
651 }
652
ostei_p_g_s_i(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__p_g_s_i)653 int ostei_p_g_s_i(struct simint_multi_shellpair const P,
654 struct simint_multi_shellpair const Q,
655 double screen_tol,
656 double * const restrict work,
657 double * const restrict INT__p_g_s_i)
658 {
659 double P_AB[3*P.nshell12];
660 struct simint_multi_shellpair P_tmp = P;
661 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
662 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
663 P_tmp.AB_x = P_AB;
664 P_tmp.AB_y = P_AB + P.nshell12;
665 P_tmp.AB_z = P_AB + 2*P.nshell12;
666
667 for(int i = 0; i < P.nshell12; i++)
668 {
669 P_tmp.AB_x[i] = -P.AB_x[i];
670 P_tmp.AB_y[i] = -P.AB_y[i];
671 P_tmp.AB_z[i] = -P.AB_z[i];
672 }
673
674 double Q_AB[3*Q.nshell12];
675 struct simint_multi_shellpair Q_tmp = Q;
676 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
677 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
678 Q_tmp.AB_x = Q_AB;
679 Q_tmp.AB_y = Q_AB + Q.nshell12;
680 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
681
682 for(int i = 0; i < Q.nshell12; i++)
683 {
684 Q_tmp.AB_x[i] = -Q.AB_x[i];
685 Q_tmp.AB_y[i] = -Q.AB_y[i];
686 Q_tmp.AB_z[i] = -Q.AB_z[i];
687 }
688
689 int ret = ostei_g_p_i_s(P_tmp, Q_tmp, screen_tol, work, INT__p_g_s_i);
690 double buffer[1260] SIMINT_ALIGN_ARRAY_DBL;
691
692 for(int q = 0; q < ret; q++)
693 {
694 int idx = 0;
695 for(int a = 0; a < 3; ++a)
696 for(int b = 0; b < 15; ++b)
697 for(int c = 0; c < 1; ++c)
698 for(int d = 0; d < 28; ++d)
699 buffer[idx++] = INT__p_g_s_i[q*1260+b*84+a*28+d*1+c];
700
701 memcpy(INT__p_g_s_i+q*1260, buffer, 1260*sizeof(double));
702 }
703
704 return ret;
705 }
706
707