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_d_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__f_d_i_s)8 int ostei_f_d_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__f_d_i_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__f_d_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__f_s_i_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__g_s_i_s = work + (SIMINT_NSHELL_SIMD * 280);
30 double * const INT__h_s_i_s = work + (SIMINT_NSHELL_SIMD * 700);
31 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*1288);
32 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
33 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_p_s = primwork + 12;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_d_s = primwork + 45;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_f_s = primwork + 105;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_g_s = primwork + 195;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_h_s = primwork + 315;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_i_s = primwork + 462;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_d_s = primwork + 630;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_f_s = primwork + 720;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_g_s = primwork + 870;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_h_s = primwork + 1095;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_i_s = primwork + 1410;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_f_s = primwork + 1830;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_g_s = primwork + 2070;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_h_s = primwork + 2430;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_i_s = primwork + 2934;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_g_s = primwork + 3606;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_h_s = primwork + 4056;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_i_s = primwork + 4686;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_h_s = primwork + 5526;
52 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_i_s = primwork + 6156;
53 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_i_s = primwork + 6996;
54 double * const hrrwork = (double *)(primwork + 7584);
55 double * const HRR_INT__f_p_i_s = hrrwork + 0;
56 double * const HRR_INT__g_p_i_s = hrrwork + 840;
57
58
59 // Create constants
60 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
61 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
62 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
63 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
64 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
65 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
66 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
67
68
69 ////////////////////////////////////////
70 // Loop over shells and primitives
71 ////////////////////////////////////////
72
73 real_abcd = 0;
74 istart = 0;
75 for(ab = 0; ab < P.nshell12_clip; ++ab)
76 {
77 const int iend = istart + P.nprim12[ab];
78
79 cd = 0;
80 jstart = 0;
81
82 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
83 {
84 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
85 int jend = jstart;
86 for(i = 0; i < nshellbatch; i++)
87 jend += Q.nprim12[cd+i];
88
89 // Clear the beginning of the workspace (where we are accumulating integrals)
90 memset(work, 0, SIMINT_NSHELL_SIMD * 1288 * sizeof(double));
91 abcd = 0;
92
93
94 for(i = istart; i < iend; ++i)
95 {
96 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
97
98 if(check_screen)
99 {
100 // Skip this whole thing if always insignificant
101 if((P.screen[i] * Q.screen_max) < screen_tol)
102 continue;
103 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
104 }
105
106 icd = 0;
107 iprimcd = 0;
108 nprim_icd = Q.nprim12[cd];
109 double * restrict PRIM_PTR_INT__f_s_i_s = INT__f_s_i_s + abcd * 280;
110 double * restrict PRIM_PTR_INT__g_s_i_s = INT__g_s_i_s + abcd * 420;
111 double * restrict PRIM_PTR_INT__h_s_i_s = INT__h_s_i_s + abcd * 588;
112
113
114
115 // Load these one per loop over i
116 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
117 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
118 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
119
120 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
121
122 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
123 {
124 // calculate the shell offsets
125 // these are the offset from the shell pointed to by cd
126 // for each element
127 int shelloffsets[SIMINT_SIMD_LEN] = {0};
128 int lastoffset = 0;
129 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
130
131 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
132 {
133 // Handle if the first element of the vector is a new shell
134 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
135 {
136 nprim_icd += Q.nprim12[cd + (++icd)];
137 PRIM_PTR_INT__f_s_i_s += 280;
138 PRIM_PTR_INT__g_s_i_s += 420;
139 PRIM_PTR_INT__h_s_i_s += 588;
140 }
141 iprimcd++;
142 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
143 {
144 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
145 {
146 shelloffsets[n] = shelloffsets[n-1] + 1;
147 lastoffset++;
148 nprim_icd += Q.nprim12[cd + (++icd)];
149 }
150 else
151 shelloffsets[n] = shelloffsets[n-1];
152 iprimcd++;
153 }
154 }
155 else
156 iprimcd += SIMINT_SIMD_LEN;
157
158 // Do we have to compute this vector (or has it been screened out)?
159 // (not_screened != 0 means we have to do this vector)
160 if(check_screen)
161 {
162 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
163 if(vmax < screen_tol)
164 {
165 PRIM_PTR_INT__f_s_i_s += lastoffset*280;
166 PRIM_PTR_INT__g_s_i_s += lastoffset*420;
167 PRIM_PTR_INT__h_s_i_s += lastoffset*588;
168 continue;
169 }
170 }
171
172 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
173 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
174 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
175 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
176
177
178 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
179 SIMINT_DBLTYPE PQ[3];
180 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
181 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
182 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
183 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
184 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
185 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
186
187 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
188 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
189 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
190 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
191 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
192 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
193 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
194
195 // NOTE: Minus sign!
196 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
197 SIMINT_DBLTYPE aop_PQ[3];
198 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
199 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
200 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
201
202 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
203 SIMINT_DBLTYPE aoq_PQ[3];
204 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
205 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
206 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
207 // Put a minus sign here so we don't have to in RR routines
208 a_over_q = SIMINT_NEG(a_over_q);
209
210
211 //////////////////////////////////////////////
212 // Fjt function section
213 // Maximum v value: 11
214 //////////////////////////////////////////////
215 // The parameter to the Fjt function
216 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
217
218
219 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
220
221
222 boys_F_split(PRIM_INT__s_s_s_s, F_x, 11);
223 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
224 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
225 for(n = 0; n <= 11; n++)
226 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
227
228 //////////////////////////////////////////////
229 // Primitive integrals: Vertical recurrance
230 //////////////////////////////////////////////
231
232 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
233 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
234 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
235 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
236 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
237 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
238 const SIMINT_DBLTYPE vrr_const_3_over_2q = SIMINT_MUL(const_3, one_over_2q);
239 const SIMINT_DBLTYPE vrr_const_4_over_2q = SIMINT_MUL(const_4, one_over_2q);
240 const SIMINT_DBLTYPE vrr_const_5_over_2q = SIMINT_MUL(const_5, one_over_2q);
241 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
242 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
243 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
244 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
245 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
246 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
247
248
249
250 // Forming PRIM_INT__s_s_p_s[11 * 3];
251 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
252 {
253
254 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
255 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]);
256
257 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
258 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]);
259
260 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
261 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]);
262
263 }
264
265
266
267 // Forming PRIM_INT__s_s_d_s[10 * 6];
268 for(n = 0; n < 10; ++n) // loop over orders of auxiliary function
269 {
270
271 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_p_s[n * 3 + 0]);
272 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]);
273 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]);
274
275 PRIM_INT__s_s_d_s[n * 6 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 0]);
276 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]);
277
278 PRIM_INT__s_s_d_s[n * 6 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 0]);
279 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]);
280
281 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 1]);
282 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]);
283 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]);
284
285 PRIM_INT__s_s_d_s[n * 6 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 1]);
286 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]);
287
288 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 2]);
289 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]);
290 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]);
291
292 }
293
294
295
296 // Forming PRIM_INT__s_s_f_s[9 * 10];
297 for(n = 0; n < 9; ++n) // loop over orders of auxiliary function
298 {
299
300 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
301 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]);
302 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]);
303
304 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
305 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]);
306
307 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
308 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]);
309
310 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
311 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]);
312
313 PRIM_INT__s_s_f_s[n * 10 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
314 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]);
315
316 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
317 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]);
318
319 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
320 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]);
321 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]);
322
323 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
324 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]);
325
326 PRIM_INT__s_s_f_s[n * 10 + 8] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
327 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]);
328
329 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
330 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]);
331 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]);
332
333 }
334
335
336 VRR_K_s_s_g_s(
337 PRIM_INT__s_s_g_s,
338 PRIM_INT__s_s_f_s,
339 PRIM_INT__s_s_d_s,
340 Q_PA,
341 a_over_q,
342 aoq_PQ,
343 one_over_2q,
344 8);
345
346
347 VRR_K_s_s_h_s(
348 PRIM_INT__s_s_h_s,
349 PRIM_INT__s_s_g_s,
350 PRIM_INT__s_s_f_s,
351 Q_PA,
352 a_over_q,
353 aoq_PQ,
354 one_over_2q,
355 7);
356
357
358 ostei_general_vrr1_K(6, 6,
359 one_over_2q, a_over_q, aoq_PQ, Q_PA,
360 PRIM_INT__s_s_h_s, PRIM_INT__s_s_g_s, PRIM_INT__s_s_i_s);
361
362
363 ostei_general_vrr_I(1, 0, 6, 0, 5,
364 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
365 PRIM_INT__s_s_i_s, NULL, NULL, PRIM_INT__s_s_h_s, NULL, PRIM_INT__p_s_i_s);
366
367
368 ostei_general_vrr_I(1, 0, 5, 0, 5,
369 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
370 PRIM_INT__s_s_h_s, NULL, NULL, PRIM_INT__s_s_g_s, NULL, PRIM_INT__p_s_h_s);
371
372
373 ostei_general_vrr_I(2, 0, 6, 0, 4,
374 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
375 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);
376
377
378 VRR_I_p_s_g_s(
379 PRIM_INT__p_s_g_s,
380 PRIM_INT__s_s_g_s,
381 PRIM_INT__s_s_f_s,
382 P_PA,
383 aop_PQ,
384 one_over_2pq,
385 5);
386
387
388 ostei_general_vrr_I(2, 0, 5, 0, 4,
389 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
390 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);
391
392
393 ostei_general_vrr_I(3, 0, 6, 0, 3,
394 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
395 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);
396
397
398 VRR_I_p_s_f_s(
399 PRIM_INT__p_s_f_s,
400 PRIM_INT__s_s_f_s,
401 PRIM_INT__s_s_d_s,
402 P_PA,
403 aop_PQ,
404 one_over_2pq,
405 5);
406
407
408 ostei_general_vrr_I(2, 0, 4, 0, 4,
409 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
410 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);
411
412
413 ostei_general_vrr_I(3, 0, 5, 0, 3,
414 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
415 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);
416
417
418 ostei_general_vrr_I(4, 0, 6, 0, 2,
419 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
420 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);
421
422
423
424 // Forming PRIM_INT__p_s_d_s[5 * 18];
425 for(n = 0; n < 5; ++n) // loop over orders of auxiliary function
426 {
427
428 PRIM_INT__p_s_d_s[n * 18 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
429 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]);
430 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]);
431
432 PRIM_INT__p_s_d_s[n * 18 + 1] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 1]);
433 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]);
434 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]);
435
436 PRIM_INT__p_s_d_s[n * 18 + 2] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 2]);
437 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]);
438 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]);
439
440 PRIM_INT__p_s_d_s[n * 18 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
441 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]);
442
443 PRIM_INT__p_s_d_s[n * 18 + 4] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 4]);
444 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]);
445
446 PRIM_INT__p_s_d_s[n * 18 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
447 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]);
448
449 PRIM_INT__p_s_d_s[n * 18 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
450 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]);
451
452 PRIM_INT__p_s_d_s[n * 18 + 7] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 1]);
453 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]);
454 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]);
455
456 PRIM_INT__p_s_d_s[n * 18 + 8] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 2]);
457 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]);
458
459 PRIM_INT__p_s_d_s[n * 18 + 9] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
460 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]);
461 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]);
462
463 PRIM_INT__p_s_d_s[n * 18 + 10] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 4]);
464 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]);
465 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]);
466
467 PRIM_INT__p_s_d_s[n * 18 + 11] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
468 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]);
469
470 PRIM_INT__p_s_d_s[n * 18 + 12] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
471 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]);
472
473 PRIM_INT__p_s_d_s[n * 18 + 13] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
474 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]);
475
476 PRIM_INT__p_s_d_s[n * 18 + 14] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 2]);
477 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]);
478 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]);
479
480 PRIM_INT__p_s_d_s[n * 18 + 15] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
481 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]);
482
483 PRIM_INT__p_s_d_s[n * 18 + 16] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 4]);
484 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]);
485 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]);
486
487 PRIM_INT__p_s_d_s[n * 18 + 17] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
488 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]);
489 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]);
490
491 }
492
493
494 VRR_I_d_s_f_s(
495 PRIM_INT__d_s_f_s,
496 PRIM_INT__p_s_f_s,
497 PRIM_INT__s_s_f_s,
498 PRIM_INT__p_s_d_s,
499 P_PA,
500 a_over_p,
501 aop_PQ,
502 one_over_2p,
503 one_over_2pq,
504 4);
505
506
507 ostei_general_vrr_I(3, 0, 4, 0, 3,
508 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
509 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);
510
511
512 ostei_general_vrr_I(4, 0, 5, 0, 2,
513 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
514 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);
515
516
517 ostei_general_vrr_I(5, 0, 6, 0, 1,
518 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
519 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);
520
521
522
523
524 ////////////////////////////////////
525 // Accumulate contracted integrals
526 ////////////////////////////////////
527 if(lastoffset == 0)
528 {
529 contract_all(280, PRIM_INT__f_s_i_s, PRIM_PTR_INT__f_s_i_s);
530 contract_all(420, PRIM_INT__g_s_i_s, PRIM_PTR_INT__g_s_i_s);
531 contract_all(588, PRIM_INT__h_s_i_s, PRIM_PTR_INT__h_s_i_s);
532 }
533 else
534 {
535 contract(280, shelloffsets, PRIM_INT__f_s_i_s, PRIM_PTR_INT__f_s_i_s);
536 contract(420, shelloffsets, PRIM_INT__g_s_i_s, PRIM_PTR_INT__g_s_i_s);
537 contract(588, shelloffsets, PRIM_INT__h_s_i_s, PRIM_PTR_INT__h_s_i_s);
538 PRIM_PTR_INT__f_s_i_s += lastoffset*280;
539 PRIM_PTR_INT__g_s_i_s += lastoffset*420;
540 PRIM_PTR_INT__h_s_i_s += lastoffset*588;
541 }
542
543 } // close loop over j
544 } // close loop over i
545
546 //Advance to the next batch
547 jstart = SIMINT_SIMD_ROUND(jend);
548
549 //////////////////////////////////////////////
550 // Contracted integrals: Horizontal recurrance
551 //////////////////////////////////////////////
552
553
554 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
555
556
557 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
558 {
559
560 // set up HRR pointers
561 double const * restrict HRR_INT__f_s_i_s = INT__f_s_i_s + abcd * 280;
562 double const * restrict HRR_INT__g_s_i_s = INT__g_s_i_s + abcd * 420;
563 double const * restrict HRR_INT__h_s_i_s = INT__h_s_i_s + abcd * 588;
564 double * restrict HRR_INT__f_d_i_s = INT__f_d_i_s + real_abcd * 1680;
565
566 // form INT__f_p_i_s
567 HRR_J_f_p(
568 HRR_INT__f_p_i_s,
569 HRR_INT__f_s_i_s,
570 HRR_INT__g_s_i_s,
571 hAB, 28);
572
573 // form INT__g_p_i_s
574 HRR_J_g_p(
575 HRR_INT__g_p_i_s,
576 HRR_INT__g_s_i_s,
577 HRR_INT__h_s_i_s,
578 hAB, 28);
579
580 // form INT__f_d_i_s
581 HRR_J_f_d(
582 HRR_INT__f_d_i_s,
583 HRR_INT__f_p_i_s,
584 HRR_INT__g_p_i_s,
585 hAB, 28);
586
587
588 } // close HRR loop
589
590
591 } // close loop cdbatch
592
593 istart = iend;
594 } // close loop over ab
595
596 return P.nshell12_clip * Q.nshell12_clip;
597 }
598
ostei_d_f_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__d_f_i_s)599 int ostei_d_f_i_s(struct simint_multi_shellpair const P,
600 struct simint_multi_shellpair const Q,
601 double screen_tol,
602 double * const restrict work,
603 double * const restrict INT__d_f_i_s)
604 {
605 double P_AB[3*P.nshell12];
606 struct simint_multi_shellpair P_tmp = P;
607 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
608 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
609 P_tmp.AB_x = P_AB;
610 P_tmp.AB_y = P_AB + P.nshell12;
611 P_tmp.AB_z = P_AB + 2*P.nshell12;
612
613 for(int i = 0; i < P.nshell12; i++)
614 {
615 P_tmp.AB_x[i] = -P.AB_x[i];
616 P_tmp.AB_y[i] = -P.AB_y[i];
617 P_tmp.AB_z[i] = -P.AB_z[i];
618 }
619
620 int ret = ostei_f_d_i_s(P_tmp, Q, screen_tol, work, INT__d_f_i_s);
621 double buffer[1680] SIMINT_ALIGN_ARRAY_DBL;
622
623 for(int q = 0; q < ret; q++)
624 {
625 int idx = 0;
626 for(int a = 0; a < 6; ++a)
627 for(int b = 0; b < 10; ++b)
628 for(int c = 0; c < 28; ++c)
629 for(int d = 0; d < 1; ++d)
630 buffer[idx++] = INT__d_f_i_s[q*1680+b*168+a*28+c*1+d];
631
632 memcpy(INT__d_f_i_s+q*1680, buffer, 1680*sizeof(double));
633 }
634
635 return ret;
636 }
637
ostei_f_d_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__f_d_s_i)638 int ostei_f_d_s_i(struct simint_multi_shellpair const P,
639 struct simint_multi_shellpair const Q,
640 double screen_tol,
641 double * const restrict work,
642 double * const restrict INT__f_d_s_i)
643 {
644 double Q_AB[3*Q.nshell12];
645 struct simint_multi_shellpair Q_tmp = Q;
646 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
647 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
648 Q_tmp.AB_x = Q_AB;
649 Q_tmp.AB_y = Q_AB + Q.nshell12;
650 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
651
652 for(int i = 0; i < Q.nshell12; i++)
653 {
654 Q_tmp.AB_x[i] = -Q.AB_x[i];
655 Q_tmp.AB_y[i] = -Q.AB_y[i];
656 Q_tmp.AB_z[i] = -Q.AB_z[i];
657 }
658
659 int ret = ostei_f_d_i_s(P, Q_tmp, screen_tol, work, INT__f_d_s_i);
660 double buffer[1680] SIMINT_ALIGN_ARRAY_DBL;
661
662 for(int q = 0; q < ret; q++)
663 {
664 int idx = 0;
665 for(int a = 0; a < 10; ++a)
666 for(int b = 0; b < 6; ++b)
667 for(int c = 0; c < 1; ++c)
668 for(int d = 0; d < 28; ++d)
669 buffer[idx++] = INT__f_d_s_i[q*1680+a*168+b*28+d*1+c];
670
671 memcpy(INT__f_d_s_i+q*1680, buffer, 1680*sizeof(double));
672 }
673
674 return ret;
675 }
676
ostei_d_f_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__d_f_s_i)677 int ostei_d_f_s_i(struct simint_multi_shellpair const P,
678 struct simint_multi_shellpair const Q,
679 double screen_tol,
680 double * const restrict work,
681 double * const restrict INT__d_f_s_i)
682 {
683 double P_AB[3*P.nshell12];
684 struct simint_multi_shellpair P_tmp = P;
685 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
686 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
687 P_tmp.AB_x = P_AB;
688 P_tmp.AB_y = P_AB + P.nshell12;
689 P_tmp.AB_z = P_AB + 2*P.nshell12;
690
691 for(int i = 0; i < P.nshell12; i++)
692 {
693 P_tmp.AB_x[i] = -P.AB_x[i];
694 P_tmp.AB_y[i] = -P.AB_y[i];
695 P_tmp.AB_z[i] = -P.AB_z[i];
696 }
697
698 double Q_AB[3*Q.nshell12];
699 struct simint_multi_shellpair Q_tmp = Q;
700 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
701 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
702 Q_tmp.AB_x = Q_AB;
703 Q_tmp.AB_y = Q_AB + Q.nshell12;
704 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
705
706 for(int i = 0; i < Q.nshell12; i++)
707 {
708 Q_tmp.AB_x[i] = -Q.AB_x[i];
709 Q_tmp.AB_y[i] = -Q.AB_y[i];
710 Q_tmp.AB_z[i] = -Q.AB_z[i];
711 }
712
713 int ret = ostei_f_d_i_s(P_tmp, Q_tmp, screen_tol, work, INT__d_f_s_i);
714 double buffer[1680] SIMINT_ALIGN_ARRAY_DBL;
715
716 for(int q = 0; q < ret; q++)
717 {
718 int idx = 0;
719 for(int a = 0; a < 6; ++a)
720 for(int b = 0; b < 10; ++b)
721 for(int c = 0; c < 1; ++c)
722 for(int d = 0; d < 28; ++d)
723 buffer[idx++] = INT__d_f_s_i[q*1680+b*168+a*28+d*1+c];
724
725 memcpy(INT__d_f_s_i+q*1680, buffer, 1680*sizeof(double));
726 }
727
728 return ret;
729 }
730
731