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_d_p(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_d_p)8 int ostei_g_p_d_p(struct simint_multi_shellpair const P,
9 struct simint_multi_shellpair const Q,
10 double screen_tol,
11 double * const restrict work,
12 double * const restrict INT__g_p_d_p)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__g_p_d_p);
17 int ab, cd, abcd;
18 int istart, jstart;
19 int iprimcd, nprim_icd, icd;
20 const int check_screen = (screen_tol > 0.0);
21 int i, j;
22 int n;
23 int not_screened;
24 int real_abcd;
25 int iket;
26 int ibra;
27
28 // partition workspace
29 double * const INT__g_s_d_s = work + (SIMINT_NSHELL_SIMD * 0);
30 double * const INT__g_s_f_s = work + (SIMINT_NSHELL_SIMD * 90);
31 double * const INT__h_s_d_s = work + (SIMINT_NSHELL_SIMD * 240);
32 double * const INT__h_s_f_s = work + (SIMINT_NSHELL_SIMD * 366);
33 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*576);
34 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_s_s = primwork + 9;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 33;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_p_s = primwork + 75;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 129;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_p_s = primwork + 189;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_d_s = primwork + 279;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 399;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_p_s = primwork + 474;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_d_s = primwork + 609;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_f_s = primwork + 789;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 939;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_p_s = primwork + 1023;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_d_s = primwork + 1212;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_f_s = primwork + 1464;
49 double * const hrrwork = (double *)(primwork + 1674);
50 double * const HRR_INT__g_p_d_s = hrrwork + 0;
51 double * const HRR_INT__g_p_f_s = hrrwork + 270;
52
53
54 // Create constants
55 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
56 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
57 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
58 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
59 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
60 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
61
62
63 ////////////////////////////////////////
64 // Loop over shells and primitives
65 ////////////////////////////////////////
66
67 real_abcd = 0;
68 istart = 0;
69 for(ab = 0; ab < P.nshell12_clip; ++ab)
70 {
71 const int iend = istart + P.nprim12[ab];
72
73 cd = 0;
74 jstart = 0;
75
76 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
77 {
78 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
79 int jend = jstart;
80 for(i = 0; i < nshellbatch; i++)
81 jend += Q.nprim12[cd+i];
82
83 // Clear the beginning of the workspace (where we are accumulating integrals)
84 memset(work, 0, SIMINT_NSHELL_SIMD * 576 * sizeof(double));
85 abcd = 0;
86
87
88 for(i = istart; i < iend; ++i)
89 {
90 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
91
92 if(check_screen)
93 {
94 // Skip this whole thing if always insignificant
95 if((P.screen[i] * Q.screen_max) < screen_tol)
96 continue;
97 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
98 }
99
100 icd = 0;
101 iprimcd = 0;
102 nprim_icd = Q.nprim12[cd];
103 double * restrict PRIM_PTR_INT__g_s_d_s = INT__g_s_d_s + abcd * 90;
104 double * restrict PRIM_PTR_INT__g_s_f_s = INT__g_s_f_s + abcd * 150;
105 double * restrict PRIM_PTR_INT__h_s_d_s = INT__h_s_d_s + abcd * 126;
106 double * restrict PRIM_PTR_INT__h_s_f_s = INT__h_s_f_s + abcd * 210;
107
108
109
110 // Load these one per loop over i
111 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
112 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
113 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
114
115 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
116
117 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
118 {
119 // calculate the shell offsets
120 // these are the offset from the shell pointed to by cd
121 // for each element
122 int shelloffsets[SIMINT_SIMD_LEN] = {0};
123 int lastoffset = 0;
124 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
125
126 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
127 {
128 // Handle if the first element of the vector is a new shell
129 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
130 {
131 nprim_icd += Q.nprim12[cd + (++icd)];
132 PRIM_PTR_INT__g_s_d_s += 90;
133 PRIM_PTR_INT__g_s_f_s += 150;
134 PRIM_PTR_INT__h_s_d_s += 126;
135 PRIM_PTR_INT__h_s_f_s += 210;
136 }
137 iprimcd++;
138 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
139 {
140 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
141 {
142 shelloffsets[n] = shelloffsets[n-1] + 1;
143 lastoffset++;
144 nprim_icd += Q.nprim12[cd + (++icd)];
145 }
146 else
147 shelloffsets[n] = shelloffsets[n-1];
148 iprimcd++;
149 }
150 }
151 else
152 iprimcd += SIMINT_SIMD_LEN;
153
154 // Do we have to compute this vector (or has it been screened out)?
155 // (not_screened != 0 means we have to do this vector)
156 if(check_screen)
157 {
158 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
159 if(vmax < screen_tol)
160 {
161 PRIM_PTR_INT__g_s_d_s += lastoffset*90;
162 PRIM_PTR_INT__g_s_f_s += lastoffset*150;
163 PRIM_PTR_INT__h_s_d_s += lastoffset*126;
164 PRIM_PTR_INT__h_s_f_s += lastoffset*210;
165 continue;
166 }
167 }
168
169 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
170 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
171 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
172 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
173
174
175 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
176 SIMINT_DBLTYPE PQ[3];
177 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
178 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
179 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
180 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
181 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
182 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
183
184 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
185 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
186 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
187 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
188 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
189 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
190 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
191
192 // NOTE: Minus sign!
193 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
194 SIMINT_DBLTYPE aop_PQ[3];
195 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
196 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
197 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
198
199 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
200 SIMINT_DBLTYPE aoq_PQ[3];
201 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
202 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
203 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
204 // Put a minus sign here so we don't have to in RR routines
205 a_over_q = SIMINT_NEG(a_over_q);
206
207
208 //////////////////////////////////////////////
209 // Fjt function section
210 // Maximum v value: 8
211 //////////////////////////////////////////////
212 // The parameter to the Fjt function
213 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
214
215
216 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
217
218
219 boys_F_split(PRIM_INT__s_s_s_s, F_x, 8);
220 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
221 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
222 for(n = 0; n <= 8; n++)
223 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
224
225 //////////////////////////////////////////////
226 // Primitive integrals: Vertical recurrance
227 //////////////////////////////////////////////
228
229 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
230 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
231 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
232 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
233 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
234 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, 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
241
242
243 // Forming PRIM_INT__p_s_s_s[8 * 3];
244 for(n = 0; n < 8; ++n) // loop over orders of auxiliary function
245 {
246
247 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
248 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]);
249
250 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
251 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]);
252
253 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
254 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]);
255
256 }
257
258
259
260 // Forming PRIM_INT__d_s_s_s[7 * 6];
261 for(n = 0; n < 7; ++n) // loop over orders of auxiliary function
262 {
263
264 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
265 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]);
266 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]);
267
268 PRIM_INT__d_s_s_s[n * 6 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 0]);
269 PRIM_INT__d_s_s_s[n * 6 + 1] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__d_s_s_s[n * 6 + 1]);
270
271 PRIM_INT__d_s_s_s[n * 6 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 0]);
272 PRIM_INT__d_s_s_s[n * 6 + 2] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__d_s_s_s[n * 6 + 2]);
273
274 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
275 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]);
276 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]);
277
278 PRIM_INT__d_s_s_s[n * 6 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 1]);
279 PRIM_INT__d_s_s_s[n * 6 + 4] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__p_s_s_s[(n+1) * 3 + 1], PRIM_INT__d_s_s_s[n * 6 + 4]);
280
281 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
282 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]);
283 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]);
284
285 }
286
287
288
289 // Forming PRIM_INT__f_s_s_s[6 * 10];
290 for(n = 0; n < 6; ++n) // loop over orders of auxiliary function
291 {
292
293 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
294 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]);
295 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]);
296
297 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
298 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]);
299
300 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
301 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]);
302
303 PRIM_INT__f_s_s_s[n * 10 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
304 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]);
305
306 PRIM_INT__f_s_s_s[n * 10 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 1]);
307 PRIM_INT__f_s_s_s[n * 10 + 4] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__d_s_s_s[(n+1) * 6 + 1], PRIM_INT__f_s_s_s[n * 10 + 4]);
308
309 PRIM_INT__f_s_s_s[n * 10 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
310 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]);
311
312 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
313 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]);
314 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]);
315
316 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
317 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]);
318
319 PRIM_INT__f_s_s_s[n * 10 + 8] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 5]);
320 PRIM_INT__f_s_s_s[n * 10 + 8] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__d_s_s_s[(n+1) * 6 + 5], PRIM_INT__f_s_s_s[n * 10 + 8]);
321
322 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
323 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]);
324 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]);
325
326 }
327
328
329 VRR_I_g_s_s_s(
330 PRIM_INT__g_s_s_s,
331 PRIM_INT__f_s_s_s,
332 PRIM_INT__d_s_s_s,
333 P_PA,
334 a_over_p,
335 aop_PQ,
336 one_over_2p,
337 5);
338
339
340 VRR_K_g_s_p_s(
341 PRIM_INT__g_s_p_s,
342 PRIM_INT__g_s_s_s,
343 PRIM_INT__f_s_s_s,
344 Q_PA,
345 aoq_PQ,
346 one_over_2pq,
347 3);
348
349
350 VRR_K_f_s_p_s(
351 PRIM_INT__f_s_p_s,
352 PRIM_INT__f_s_s_s,
353 PRIM_INT__d_s_s_s,
354 Q_PA,
355 aoq_PQ,
356 one_over_2pq,
357 3);
358
359
360 ostei_general_vrr_K(4, 0, 2, 0, 2,
361 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
362 PRIM_INT__g_s_p_s, PRIM_INT__g_s_s_s, NULL, PRIM_INT__f_s_p_s, NULL, PRIM_INT__g_s_d_s);
363
364
365
366 // Forming PRIM_INT__d_s_p_s[3 * 18];
367 for(n = 0; n < 3; ++n) // loop over orders of auxiliary function
368 {
369
370 PRIM_INT__d_s_p_s[n * 18 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
371 PRIM_INT__d_s_p_s[n * 18 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__d_s_s_s[(n+1) * 6 + 0], PRIM_INT__d_s_p_s[n * 18 + 0]);
372 PRIM_INT__d_s_p_s[n * 18 + 0] = SIMINT_FMADD( vrr_const_2_over_2pq, PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__d_s_p_s[n * 18 + 0]);
373
374 PRIM_INT__d_s_p_s[n * 18 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
375 PRIM_INT__d_s_p_s[n * 18 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__d_s_s_s[(n+1) * 6 + 0], PRIM_INT__d_s_p_s[n * 18 + 1]);
376
377 PRIM_INT__d_s_p_s[n * 18 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
378 PRIM_INT__d_s_p_s[n * 18 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__d_s_s_s[(n+1) * 6 + 0], PRIM_INT__d_s_p_s[n * 18 + 2]);
379
380 PRIM_INT__d_s_p_s[n * 18 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 1]);
381 PRIM_INT__d_s_p_s[n * 18 + 3] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__d_s_s_s[(n+1) * 6 + 1], PRIM_INT__d_s_p_s[n * 18 + 3]);
382 PRIM_INT__d_s_p_s[n * 18 + 3] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__p_s_s_s[(n+1) * 3 + 1], PRIM_INT__d_s_p_s[n * 18 + 3]);
383
384 PRIM_INT__d_s_p_s[n * 18 + 4] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 1]);
385 PRIM_INT__d_s_p_s[n * 18 + 4] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__d_s_s_s[(n+1) * 6 + 1], PRIM_INT__d_s_p_s[n * 18 + 4]);
386 PRIM_INT__d_s_p_s[n * 18 + 4] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__d_s_p_s[n * 18 + 4]);
387
388 PRIM_INT__d_s_p_s[n * 18 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 1]);
389 PRIM_INT__d_s_p_s[n * 18 + 5] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__d_s_s_s[(n+1) * 6 + 1], PRIM_INT__d_s_p_s[n * 18 + 5]);
390
391 PRIM_INT__d_s_p_s[n * 18 + 6] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 2]);
392 PRIM_INT__d_s_p_s[n * 18 + 6] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__d_s_s_s[(n+1) * 6 + 2], PRIM_INT__d_s_p_s[n * 18 + 6]);
393 PRIM_INT__d_s_p_s[n * 18 + 6] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__p_s_s_s[(n+1) * 3 + 2], PRIM_INT__d_s_p_s[n * 18 + 6]);
394
395 PRIM_INT__d_s_p_s[n * 18 + 7] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 2]);
396 PRIM_INT__d_s_p_s[n * 18 + 7] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__d_s_s_s[(n+1) * 6 + 2], PRIM_INT__d_s_p_s[n * 18 + 7]);
397
398 PRIM_INT__d_s_p_s[n * 18 + 8] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 2]);
399 PRIM_INT__d_s_p_s[n * 18 + 8] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__d_s_s_s[(n+1) * 6 + 2], PRIM_INT__d_s_p_s[n * 18 + 8]);
400 PRIM_INT__d_s_p_s[n * 18 + 8] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__d_s_p_s[n * 18 + 8]);
401
402 PRIM_INT__d_s_p_s[n * 18 + 9] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
403 PRIM_INT__d_s_p_s[n * 18 + 9] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__d_s_s_s[(n+1) * 6 + 3], PRIM_INT__d_s_p_s[n * 18 + 9]);
404
405 PRIM_INT__d_s_p_s[n * 18 + 10] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
406 PRIM_INT__d_s_p_s[n * 18 + 10] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__d_s_s_s[(n+1) * 6 + 3], PRIM_INT__d_s_p_s[n * 18 + 10]);
407 PRIM_INT__d_s_p_s[n * 18 + 10] = SIMINT_FMADD( vrr_const_2_over_2pq, PRIM_INT__p_s_s_s[(n+1) * 3 + 1], PRIM_INT__d_s_p_s[n * 18 + 10]);
408
409 PRIM_INT__d_s_p_s[n * 18 + 11] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
410 PRIM_INT__d_s_p_s[n * 18 + 11] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__d_s_s_s[(n+1) * 6 + 3], PRIM_INT__d_s_p_s[n * 18 + 11]);
411
412 PRIM_INT__d_s_p_s[n * 18 + 12] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 4]);
413 PRIM_INT__d_s_p_s[n * 18 + 12] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__d_s_s_s[(n+1) * 6 + 4], PRIM_INT__d_s_p_s[n * 18 + 12]);
414
415 PRIM_INT__d_s_p_s[n * 18 + 13] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 4]);
416 PRIM_INT__d_s_p_s[n * 18 + 13] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__d_s_s_s[(n+1) * 6 + 4], PRIM_INT__d_s_p_s[n * 18 + 13]);
417 PRIM_INT__d_s_p_s[n * 18 + 13] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__p_s_s_s[(n+1) * 3 + 2], PRIM_INT__d_s_p_s[n * 18 + 13]);
418
419 PRIM_INT__d_s_p_s[n * 18 + 14] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 4]);
420 PRIM_INT__d_s_p_s[n * 18 + 14] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__d_s_s_s[(n+1) * 6 + 4], PRIM_INT__d_s_p_s[n * 18 + 14]);
421 PRIM_INT__d_s_p_s[n * 18 + 14] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__p_s_s_s[(n+1) * 3 + 1], PRIM_INT__d_s_p_s[n * 18 + 14]);
422
423 PRIM_INT__d_s_p_s[n * 18 + 15] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
424 PRIM_INT__d_s_p_s[n * 18 + 15] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__d_s_s_s[(n+1) * 6 + 5], PRIM_INT__d_s_p_s[n * 18 + 15]);
425
426 PRIM_INT__d_s_p_s[n * 18 + 16] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 5]);
427 PRIM_INT__d_s_p_s[n * 18 + 16] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__d_s_s_s[(n+1) * 6 + 5], PRIM_INT__d_s_p_s[n * 18 + 16]);
428
429 PRIM_INT__d_s_p_s[n * 18 + 17] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
430 PRIM_INT__d_s_p_s[n * 18 + 17] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__d_s_s_s[(n+1) * 6 + 5], PRIM_INT__d_s_p_s[n * 18 + 17]);
431 PRIM_INT__d_s_p_s[n * 18 + 17] = SIMINT_FMADD( vrr_const_2_over_2pq, PRIM_INT__p_s_s_s[(n+1) * 3 + 2], PRIM_INT__d_s_p_s[n * 18 + 17]);
432
433 }
434
435
436 VRR_K_f_s_d_s(
437 PRIM_INT__f_s_d_s,
438 PRIM_INT__f_s_p_s,
439 PRIM_INT__f_s_s_s,
440 PRIM_INT__d_s_p_s,
441 Q_PA,
442 a_over_q,
443 aoq_PQ,
444 one_over_2pq,
445 one_over_2q,
446 2);
447
448
449 ostei_general_vrr_K(4, 0, 3, 0, 1,
450 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
451 PRIM_INT__g_s_d_s, PRIM_INT__g_s_p_s, NULL, PRIM_INT__f_s_d_s, NULL, PRIM_INT__g_s_f_s);
452
453
454 VRR_I_h_s_s_s(
455 PRIM_INT__h_s_s_s,
456 PRIM_INT__g_s_s_s,
457 PRIM_INT__f_s_s_s,
458 P_PA,
459 a_over_p,
460 aop_PQ,
461 one_over_2p,
462 4);
463
464
465 ostei_general_vrr_K(5, 0, 1, 0, 3,
466 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
467 PRIM_INT__h_s_s_s, NULL, NULL, PRIM_INT__g_s_s_s, NULL, PRIM_INT__h_s_p_s);
468
469
470 ostei_general_vrr_K(5, 0, 2, 0, 2,
471 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
472 PRIM_INT__h_s_p_s, PRIM_INT__h_s_s_s, NULL, PRIM_INT__g_s_p_s, NULL, PRIM_INT__h_s_d_s);
473
474
475 ostei_general_vrr_K(5, 0, 3, 0, 1,
476 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
477 PRIM_INT__h_s_d_s, PRIM_INT__h_s_p_s, NULL, PRIM_INT__g_s_d_s, NULL, PRIM_INT__h_s_f_s);
478
479
480
481
482 ////////////////////////////////////
483 // Accumulate contracted integrals
484 ////////////////////////////////////
485 if(lastoffset == 0)
486 {
487 contract_all(90, PRIM_INT__g_s_d_s, PRIM_PTR_INT__g_s_d_s);
488 contract_all(150, PRIM_INT__g_s_f_s, PRIM_PTR_INT__g_s_f_s);
489 contract_all(126, PRIM_INT__h_s_d_s, PRIM_PTR_INT__h_s_d_s);
490 contract_all(210, PRIM_INT__h_s_f_s, PRIM_PTR_INT__h_s_f_s);
491 }
492 else
493 {
494 contract(90, shelloffsets, PRIM_INT__g_s_d_s, PRIM_PTR_INT__g_s_d_s);
495 contract(150, shelloffsets, PRIM_INT__g_s_f_s, PRIM_PTR_INT__g_s_f_s);
496 contract(126, shelloffsets, PRIM_INT__h_s_d_s, PRIM_PTR_INT__h_s_d_s);
497 contract(210, shelloffsets, PRIM_INT__h_s_f_s, PRIM_PTR_INT__h_s_f_s);
498 PRIM_PTR_INT__g_s_d_s += lastoffset*90;
499 PRIM_PTR_INT__g_s_f_s += lastoffset*150;
500 PRIM_PTR_INT__h_s_d_s += lastoffset*126;
501 PRIM_PTR_INT__h_s_f_s += lastoffset*210;
502 }
503
504 } // close loop over j
505 } // close loop over i
506
507 //Advance to the next batch
508 jstart = SIMINT_SIMD_ROUND(jend);
509
510 //////////////////////////////////////////////
511 // Contracted integrals: Horizontal recurrance
512 //////////////////////////////////////////////
513
514
515 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
516
517
518 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
519 {
520 const double hCD[3] = { Q.AB_x[cd+abcd], Q.AB_y[cd+abcd], Q.AB_z[cd+abcd] };
521
522 // set up HRR pointers
523 double const * restrict HRR_INT__g_s_d_s = INT__g_s_d_s + abcd * 90;
524 double const * restrict HRR_INT__g_s_f_s = INT__g_s_f_s + abcd * 150;
525 double const * restrict HRR_INT__h_s_d_s = INT__h_s_d_s + abcd * 126;
526 double const * restrict HRR_INT__h_s_f_s = INT__h_s_f_s + abcd * 210;
527 double * restrict HRR_INT__g_p_d_p = INT__g_p_d_p + real_abcd * 810;
528
529 // form INT__g_p_d_s
530 HRR_J_g_p(
531 HRR_INT__g_p_d_s,
532 HRR_INT__g_s_d_s,
533 HRR_INT__h_s_d_s,
534 hAB, 6);
535
536 // form INT__g_p_f_s
537 HRR_J_g_p(
538 HRR_INT__g_p_f_s,
539 HRR_INT__g_s_f_s,
540 HRR_INT__h_s_f_s,
541 hAB, 10);
542
543 // form INT__g_p_d_p
544 for(ibra = 0; ibra < 45; ++ibra)
545 {
546 HRR_INT__g_p_d_p[ibra * 18 + 0] = HRR_INT__g_p_f_s[ibra * 10 + 0] + ( hCD[0] * HRR_INT__g_p_d_s[ibra * 6 + 0] );
547
548 HRR_INT__g_p_d_p[ibra * 18 + 1] = HRR_INT__g_p_f_s[ibra * 10 + 1] + ( hCD[1] * HRR_INT__g_p_d_s[ibra * 6 + 0] );
549
550 HRR_INT__g_p_d_p[ibra * 18 + 2] = HRR_INT__g_p_f_s[ibra * 10 + 2] + ( hCD[2] * HRR_INT__g_p_d_s[ibra * 6 + 0] );
551
552 HRR_INT__g_p_d_p[ibra * 18 + 3] = HRR_INT__g_p_f_s[ibra * 10 + 1] + ( hCD[0] * HRR_INT__g_p_d_s[ibra * 6 + 1] );
553
554 HRR_INT__g_p_d_p[ibra * 18 + 4] = HRR_INT__g_p_f_s[ibra * 10 + 3] + ( hCD[1] * HRR_INT__g_p_d_s[ibra * 6 + 1] );
555
556 HRR_INT__g_p_d_p[ibra * 18 + 5] = HRR_INT__g_p_f_s[ibra * 10 + 4] + ( hCD[2] * HRR_INT__g_p_d_s[ibra * 6 + 1] );
557
558 HRR_INT__g_p_d_p[ibra * 18 + 6] = HRR_INT__g_p_f_s[ibra * 10 + 2] + ( hCD[0] * HRR_INT__g_p_d_s[ibra * 6 + 2] );
559
560 HRR_INT__g_p_d_p[ibra * 18 + 7] = HRR_INT__g_p_f_s[ibra * 10 + 4] + ( hCD[1] * HRR_INT__g_p_d_s[ibra * 6 + 2] );
561
562 HRR_INT__g_p_d_p[ibra * 18 + 8] = HRR_INT__g_p_f_s[ibra * 10 + 5] + ( hCD[2] * HRR_INT__g_p_d_s[ibra * 6 + 2] );
563
564 HRR_INT__g_p_d_p[ibra * 18 + 9] = HRR_INT__g_p_f_s[ibra * 10 + 3] + ( hCD[0] * HRR_INT__g_p_d_s[ibra * 6 + 3] );
565
566 HRR_INT__g_p_d_p[ibra * 18 + 10] = HRR_INT__g_p_f_s[ibra * 10 + 6] + ( hCD[1] * HRR_INT__g_p_d_s[ibra * 6 + 3] );
567
568 HRR_INT__g_p_d_p[ibra * 18 + 11] = HRR_INT__g_p_f_s[ibra * 10 + 7] + ( hCD[2] * HRR_INT__g_p_d_s[ibra * 6 + 3] );
569
570 HRR_INT__g_p_d_p[ibra * 18 + 12] = HRR_INT__g_p_f_s[ibra * 10 + 4] + ( hCD[0] * HRR_INT__g_p_d_s[ibra * 6 + 4] );
571
572 HRR_INT__g_p_d_p[ibra * 18 + 13] = HRR_INT__g_p_f_s[ibra * 10 + 7] + ( hCD[1] * HRR_INT__g_p_d_s[ibra * 6 + 4] );
573
574 HRR_INT__g_p_d_p[ibra * 18 + 14] = HRR_INT__g_p_f_s[ibra * 10 + 8] + ( hCD[2] * HRR_INT__g_p_d_s[ibra * 6 + 4] );
575
576 HRR_INT__g_p_d_p[ibra * 18 + 15] = HRR_INT__g_p_f_s[ibra * 10 + 5] + ( hCD[0] * HRR_INT__g_p_d_s[ibra * 6 + 5] );
577
578 HRR_INT__g_p_d_p[ibra * 18 + 16] = HRR_INT__g_p_f_s[ibra * 10 + 8] + ( hCD[1] * HRR_INT__g_p_d_s[ibra * 6 + 5] );
579
580 HRR_INT__g_p_d_p[ibra * 18 + 17] = HRR_INT__g_p_f_s[ibra * 10 + 9] + ( hCD[2] * HRR_INT__g_p_d_s[ibra * 6 + 5] );
581
582 }
583
584
585 } // close HRR loop
586
587
588 } // close loop cdbatch
589
590 istart = iend;
591 } // close loop over ab
592
593 return P.nshell12_clip * Q.nshell12_clip;
594 }
595
ostei_p_g_d_p(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__p_g_d_p)596 int ostei_p_g_d_p(struct simint_multi_shellpair const P,
597 struct simint_multi_shellpair const Q,
598 double screen_tol,
599 double * const restrict work,
600 double * const restrict INT__p_g_d_p)
601 {
602 double P_AB[3*P.nshell12];
603 struct simint_multi_shellpair P_tmp = P;
604 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
605 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
606 P_tmp.AB_x = P_AB;
607 P_tmp.AB_y = P_AB + P.nshell12;
608 P_tmp.AB_z = P_AB + 2*P.nshell12;
609
610 for(int i = 0; i < P.nshell12; i++)
611 {
612 P_tmp.AB_x[i] = -P.AB_x[i];
613 P_tmp.AB_y[i] = -P.AB_y[i];
614 P_tmp.AB_z[i] = -P.AB_z[i];
615 }
616
617 int ret = ostei_g_p_d_p(P_tmp, Q, screen_tol, work, INT__p_g_d_p);
618 double buffer[810] SIMINT_ALIGN_ARRAY_DBL;
619
620 for(int q = 0; q < ret; q++)
621 {
622 int idx = 0;
623 for(int a = 0; a < 3; ++a)
624 for(int b = 0; b < 15; ++b)
625 for(int c = 0; c < 6; ++c)
626 for(int d = 0; d < 3; ++d)
627 buffer[idx++] = INT__p_g_d_p[q*810+b*54+a*18+c*3+d];
628
629 memcpy(INT__p_g_d_p+q*810, buffer, 810*sizeof(double));
630 }
631
632 return ret;
633 }
634
ostei_g_p_p_d(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_p_d)635 int ostei_g_p_p_d(struct simint_multi_shellpair const P,
636 struct simint_multi_shellpair const Q,
637 double screen_tol,
638 double * const restrict work,
639 double * const restrict INT__g_p_p_d)
640 {
641 double Q_AB[3*Q.nshell12];
642 struct simint_multi_shellpair Q_tmp = Q;
643 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
644 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
645 Q_tmp.AB_x = Q_AB;
646 Q_tmp.AB_y = Q_AB + Q.nshell12;
647 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
648
649 for(int i = 0; i < Q.nshell12; i++)
650 {
651 Q_tmp.AB_x[i] = -Q.AB_x[i];
652 Q_tmp.AB_y[i] = -Q.AB_y[i];
653 Q_tmp.AB_z[i] = -Q.AB_z[i];
654 }
655
656 int ret = ostei_g_p_d_p(P, Q_tmp, screen_tol, work, INT__g_p_p_d);
657 double buffer[810] SIMINT_ALIGN_ARRAY_DBL;
658
659 for(int q = 0; q < ret; q++)
660 {
661 int idx = 0;
662 for(int a = 0; a < 15; ++a)
663 for(int b = 0; b < 3; ++b)
664 for(int c = 0; c < 3; ++c)
665 for(int d = 0; d < 6; ++d)
666 buffer[idx++] = INT__g_p_p_d[q*810+a*54+b*18+d*3+c];
667
668 memcpy(INT__g_p_p_d+q*810, buffer, 810*sizeof(double));
669 }
670
671 return ret;
672 }
673
ostei_p_g_p_d(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_p_d)674 int ostei_p_g_p_d(struct simint_multi_shellpair const P,
675 struct simint_multi_shellpair const Q,
676 double screen_tol,
677 double * const restrict work,
678 double * const restrict INT__p_g_p_d)
679 {
680 double P_AB[3*P.nshell12];
681 struct simint_multi_shellpair P_tmp = P;
682 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
683 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
684 P_tmp.AB_x = P_AB;
685 P_tmp.AB_y = P_AB + P.nshell12;
686 P_tmp.AB_z = P_AB + 2*P.nshell12;
687
688 for(int i = 0; i < P.nshell12; i++)
689 {
690 P_tmp.AB_x[i] = -P.AB_x[i];
691 P_tmp.AB_y[i] = -P.AB_y[i];
692 P_tmp.AB_z[i] = -P.AB_z[i];
693 }
694
695 double Q_AB[3*Q.nshell12];
696 struct simint_multi_shellpair Q_tmp = Q;
697 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
698 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
699 Q_tmp.AB_x = Q_AB;
700 Q_tmp.AB_y = Q_AB + Q.nshell12;
701 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
702
703 for(int i = 0; i < Q.nshell12; i++)
704 {
705 Q_tmp.AB_x[i] = -Q.AB_x[i];
706 Q_tmp.AB_y[i] = -Q.AB_y[i];
707 Q_tmp.AB_z[i] = -Q.AB_z[i];
708 }
709
710 int ret = ostei_g_p_d_p(P_tmp, Q_tmp, screen_tol, work, INT__p_g_p_d);
711 double buffer[810] SIMINT_ALIGN_ARRAY_DBL;
712
713 for(int q = 0; q < ret; q++)
714 {
715 int idx = 0;
716 for(int a = 0; a < 3; ++a)
717 for(int b = 0; b < 15; ++b)
718 for(int c = 0; c < 3; ++c)
719 for(int d = 0; d < 6; ++d)
720 buffer[idx++] = INT__p_g_p_d[q*810+b*54+a*18+d*3+c];
721
722 memcpy(INT__p_g_p_d+q*810, buffer, 810*sizeof(double));
723 }
724
725 return ret;
726 }
727
728