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_p_p_d_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_p_d_d)8 int ostei_p_p_d_d(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__p_p_d_d)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__p_p_d_d);
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__p_s_d_s = work + (SIMINT_NSHELL_SIMD * 0);
30 double * const INT__p_s_f_s = work + (SIMINT_NSHELL_SIMD * 18);
31 double * const INT__p_s_g_s = work + (SIMINT_NSHELL_SIMD * 48);
32 double * const INT__d_s_d_s = work + (SIMINT_NSHELL_SIMD * 93);
33 double * const INT__d_s_f_s = work + (SIMINT_NSHELL_SIMD * 129);
34 double * const INT__d_s_g_s = work + (SIMINT_NSHELL_SIMD * 189);
35 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*279);
36 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_p_s = primwork + 7;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_d_s = primwork + 25;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_f_s = primwork + 55;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_g_s = primwork + 95;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_p_s = primwork + 140;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_d_s = primwork + 158;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_f_s = primwork + 194;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_g_s = primwork + 254;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_d_s = primwork + 344;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_f_s = primwork + 380;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_g_s = primwork + 440;
48 double * const hrrwork = (double *)(primwork + 530);
49 double * const HRR_INT__p_p_d_s = hrrwork + 0;
50 double * const HRR_INT__p_p_d_p = hrrwork + 54;
51 double * const HRR_INT__p_p_f_s = hrrwork + 216;
52 double * const HRR_INT__p_p_f_p = hrrwork + 306;
53 double * const HRR_INT__p_p_g_s = hrrwork + 576;
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 one_half = SIMINT_DBLSET1(0.5);
62
63
64 ////////////////////////////////////////
65 // Loop over shells and primitives
66 ////////////////////////////////////////
67
68 real_abcd = 0;
69 istart = 0;
70 for(ab = 0; ab < P.nshell12_clip; ++ab)
71 {
72 const int iend = istart + P.nprim12[ab];
73
74 cd = 0;
75 jstart = 0;
76
77 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
78 {
79 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
80 int jend = jstart;
81 for(i = 0; i < nshellbatch; i++)
82 jend += Q.nprim12[cd+i];
83
84 // Clear the beginning of the workspace (where we are accumulating integrals)
85 memset(work, 0, SIMINT_NSHELL_SIMD * 279 * sizeof(double));
86 abcd = 0;
87
88
89 for(i = istart; i < iend; ++i)
90 {
91 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
92
93 if(check_screen)
94 {
95 // Skip this whole thing if always insignificant
96 if((P.screen[i] * Q.screen_max) < screen_tol)
97 continue;
98 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
99 }
100
101 icd = 0;
102 iprimcd = 0;
103 nprim_icd = Q.nprim12[cd];
104 double * restrict PRIM_PTR_INT__p_s_d_s = INT__p_s_d_s + abcd * 18;
105 double * restrict PRIM_PTR_INT__p_s_f_s = INT__p_s_f_s + abcd * 30;
106 double * restrict PRIM_PTR_INT__p_s_g_s = INT__p_s_g_s + abcd * 45;
107 double * restrict PRIM_PTR_INT__d_s_d_s = INT__d_s_d_s + abcd * 36;
108 double * restrict PRIM_PTR_INT__d_s_f_s = INT__d_s_f_s + abcd * 60;
109 double * restrict PRIM_PTR_INT__d_s_g_s = INT__d_s_g_s + abcd * 90;
110
111
112
113 // Load these one per loop over i
114 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
115 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
116 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
117
118 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
119
120 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
121 {
122 // calculate the shell offsets
123 // these are the offset from the shell pointed to by cd
124 // for each element
125 int shelloffsets[SIMINT_SIMD_LEN] = {0};
126 int lastoffset = 0;
127 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
128
129 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
130 {
131 // Handle if the first element of the vector is a new shell
132 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
133 {
134 nprim_icd += Q.nprim12[cd + (++icd)];
135 PRIM_PTR_INT__p_s_d_s += 18;
136 PRIM_PTR_INT__p_s_f_s += 30;
137 PRIM_PTR_INT__p_s_g_s += 45;
138 PRIM_PTR_INT__d_s_d_s += 36;
139 PRIM_PTR_INT__d_s_f_s += 60;
140 PRIM_PTR_INT__d_s_g_s += 90;
141 }
142 iprimcd++;
143 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
144 {
145 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
146 {
147 shelloffsets[n] = shelloffsets[n-1] + 1;
148 lastoffset++;
149 nprim_icd += Q.nprim12[cd + (++icd)];
150 }
151 else
152 shelloffsets[n] = shelloffsets[n-1];
153 iprimcd++;
154 }
155 }
156 else
157 iprimcd += SIMINT_SIMD_LEN;
158
159 // Do we have to compute this vector (or has it been screened out)?
160 // (not_screened != 0 means we have to do this vector)
161 if(check_screen)
162 {
163 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
164 if(vmax < screen_tol)
165 {
166 PRIM_PTR_INT__p_s_d_s += lastoffset*18;
167 PRIM_PTR_INT__p_s_f_s += lastoffset*30;
168 PRIM_PTR_INT__p_s_g_s += lastoffset*45;
169 PRIM_PTR_INT__d_s_d_s += lastoffset*36;
170 PRIM_PTR_INT__d_s_f_s += lastoffset*60;
171 PRIM_PTR_INT__d_s_g_s += lastoffset*90;
172 continue;
173 }
174 }
175
176 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
177 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
178 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
179 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
180
181
182 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
183 SIMINT_DBLTYPE PQ[3];
184 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
185 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
186 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
187 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
188 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
189 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
190
191 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
192 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
193 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
194 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
195 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
196 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
197 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
198
199 // NOTE: Minus sign!
200 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
201 SIMINT_DBLTYPE aop_PQ[3];
202 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
203 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
204 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
205
206 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
207 SIMINT_DBLTYPE aoq_PQ[3];
208 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
209 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
210 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
211 // Put a minus sign here so we don't have to in RR routines
212 a_over_q = SIMINT_NEG(a_over_q);
213
214
215 //////////////////////////////////////////////
216 // Fjt function section
217 // Maximum v value: 6
218 //////////////////////////////////////////////
219 // The parameter to the Fjt function
220 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
221
222
223 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
224
225
226 boys_F_split(PRIM_INT__s_s_s_s, F_x, 6);
227 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
228 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
229 for(n = 0; n <= 6; n++)
230 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
231
232 //////////////////////////////////////////////
233 // Primitive integrals: Vertical recurrance
234 //////////////////////////////////////////////
235
236 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
237 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
238 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
239 const SIMINT_DBLTYPE vrr_const_3_over_2q = SIMINT_MUL(const_3, one_over_2q);
240 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
241 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
242 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
243 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
244
245
246
247 // Forming PRIM_INT__s_s_p_s[6 * 3];
248 for(n = 0; n < 6; ++n) // loop over orders of auxiliary function
249 {
250
251 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
252 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]);
253
254 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
255 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]);
256
257 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
258 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]);
259
260 }
261
262
263
264 // Forming PRIM_INT__s_s_d_s[5 * 6];
265 for(n = 0; n < 5; ++n) // loop over orders of auxiliary function
266 {
267
268 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_p_s[n * 3 + 0]);
269 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]);
270 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]);
271
272 PRIM_INT__s_s_d_s[n * 6 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 0]);
273 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]);
274
275 PRIM_INT__s_s_d_s[n * 6 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 0]);
276 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]);
277
278 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 1]);
279 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]);
280 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]);
281
282 PRIM_INT__s_s_d_s[n * 6 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 1]);
283 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]);
284
285 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 2]);
286 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]);
287 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]);
288
289 }
290
291
292
293 // Forming PRIM_INT__p_s_d_s[2 * 18];
294 for(n = 0; n < 2; ++n) // loop over orders of auxiliary function
295 {
296
297 PRIM_INT__p_s_d_s[n * 18 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
298 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]);
299 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]);
300
301 PRIM_INT__p_s_d_s[n * 18 + 1] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 1]);
302 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]);
303 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]);
304
305 PRIM_INT__p_s_d_s[n * 18 + 2] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 2]);
306 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]);
307 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]);
308
309 PRIM_INT__p_s_d_s[n * 18 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
310 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]);
311
312 PRIM_INT__p_s_d_s[n * 18 + 4] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 4]);
313 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]);
314
315 PRIM_INT__p_s_d_s[n * 18 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
316 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]);
317
318 PRIM_INT__p_s_d_s[n * 18 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
319 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]);
320
321 PRIM_INT__p_s_d_s[n * 18 + 7] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 1]);
322 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]);
323 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]);
324
325 PRIM_INT__p_s_d_s[n * 18 + 8] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 2]);
326 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]);
327
328 PRIM_INT__p_s_d_s[n * 18 + 9] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
329 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]);
330 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]);
331
332 PRIM_INT__p_s_d_s[n * 18 + 10] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 4]);
333 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]);
334 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]);
335
336 PRIM_INT__p_s_d_s[n * 18 + 11] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
337 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]);
338
339 PRIM_INT__p_s_d_s[n * 18 + 12] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
340 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]);
341
342 PRIM_INT__p_s_d_s[n * 18 + 13] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
343 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]);
344
345 PRIM_INT__p_s_d_s[n * 18 + 14] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 2]);
346 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]);
347 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]);
348
349 PRIM_INT__p_s_d_s[n * 18 + 15] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
350 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]);
351
352 PRIM_INT__p_s_d_s[n * 18 + 16] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 4]);
353 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]);
354 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]);
355
356 PRIM_INT__p_s_d_s[n * 18 + 17] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
357 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]);
358 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]);
359
360 }
361
362
363
364 // Forming PRIM_INT__s_s_f_s[4 * 10];
365 for(n = 0; n < 4; ++n) // loop over orders of auxiliary function
366 {
367
368 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
369 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]);
370 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]);
371
372 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
373 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]);
374
375 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
376 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]);
377
378 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
379 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]);
380
381 PRIM_INT__s_s_f_s[n * 10 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
382 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]);
383
384 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
385 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]);
386
387 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
388 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]);
389 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]);
390
391 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
392 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]);
393
394 PRIM_INT__s_s_f_s[n * 10 + 8] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
395 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]);
396
397 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
398 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]);
399 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]);
400
401 }
402
403
404 VRR_I_p_s_f_s(
405 PRIM_INT__p_s_f_s,
406 PRIM_INT__s_s_f_s,
407 PRIM_INT__s_s_d_s,
408 P_PA,
409 aop_PQ,
410 one_over_2pq,
411 2);
412
413
414
415 // Forming PRIM_INT__p_s_p_s[2 * 9];
416 for(n = 0; n < 2; ++n) // loop over orders of auxiliary function
417 {
418
419 PRIM_INT__p_s_p_s[n * 9 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_p_s[n * 3 + 0]);
420 PRIM_INT__p_s_p_s[n * 9 + 0] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__p_s_p_s[n * 9 + 0]);
421 PRIM_INT__p_s_p_s[n * 9 + 0] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__p_s_p_s[n * 9 + 0]);
422
423 PRIM_INT__p_s_p_s[n * 9 + 1] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_p_s[n * 3 + 1]);
424 PRIM_INT__p_s_p_s[n * 9 + 1] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__p_s_p_s[n * 9 + 1]);
425
426 PRIM_INT__p_s_p_s[n * 9 + 2] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_p_s[n * 3 + 2]);
427 PRIM_INT__p_s_p_s[n * 9 + 2] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__p_s_p_s[n * 9 + 2]);
428
429 PRIM_INT__p_s_p_s[n * 9 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_p_s[n * 3 + 0]);
430 PRIM_INT__p_s_p_s[n * 9 + 3] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__p_s_p_s[n * 9 + 3]);
431
432 PRIM_INT__p_s_p_s[n * 9 + 4] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_p_s[n * 3 + 1]);
433 PRIM_INT__p_s_p_s[n * 9 + 4] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__p_s_p_s[n * 9 + 4]);
434 PRIM_INT__p_s_p_s[n * 9 + 4] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__p_s_p_s[n * 9 + 4]);
435
436 PRIM_INT__p_s_p_s[n * 9 + 5] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_p_s[n * 3 + 2]);
437 PRIM_INT__p_s_p_s[n * 9 + 5] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__p_s_p_s[n * 9 + 5]);
438
439 PRIM_INT__p_s_p_s[n * 9 + 6] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_p_s[n * 3 + 0]);
440 PRIM_INT__p_s_p_s[n * 9 + 6] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__p_s_p_s[n * 9 + 6]);
441
442 PRIM_INT__p_s_p_s[n * 9 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_p_s[n * 3 + 1]);
443 PRIM_INT__p_s_p_s[n * 9 + 7] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__p_s_p_s[n * 9 + 7]);
444
445 PRIM_INT__p_s_p_s[n * 9 + 8] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_p_s[n * 3 + 2]);
446 PRIM_INT__p_s_p_s[n * 9 + 8] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__p_s_p_s[n * 9 + 8]);
447 PRIM_INT__p_s_p_s[n * 9 + 8] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__p_s_p_s[n * 9 + 8]);
448
449 }
450
451
452 VRR_I_d_s_d_s(
453 PRIM_INT__d_s_d_s,
454 PRIM_INT__p_s_d_s,
455 PRIM_INT__s_s_d_s,
456 PRIM_INT__p_s_p_s,
457 P_PA,
458 a_over_p,
459 aop_PQ,
460 one_over_2p,
461 one_over_2pq,
462 1);
463
464
465 VRR_K_s_s_g_s(
466 PRIM_INT__s_s_g_s,
467 PRIM_INT__s_s_f_s,
468 PRIM_INT__s_s_d_s,
469 Q_PA,
470 a_over_q,
471 aoq_PQ,
472 one_over_2q,
473 3);
474
475
476 VRR_I_p_s_g_s(
477 PRIM_INT__p_s_g_s,
478 PRIM_INT__s_s_g_s,
479 PRIM_INT__s_s_f_s,
480 P_PA,
481 aop_PQ,
482 one_over_2pq,
483 2);
484
485
486 VRR_I_d_s_f_s(
487 PRIM_INT__d_s_f_s,
488 PRIM_INT__p_s_f_s,
489 PRIM_INT__s_s_f_s,
490 PRIM_INT__p_s_d_s,
491 P_PA,
492 a_over_p,
493 aop_PQ,
494 one_over_2p,
495 one_over_2pq,
496 1);
497
498
499 ostei_general_vrr_I(2, 0, 4, 0, 1,
500 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
501 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);
502
503
504
505
506 ////////////////////////////////////
507 // Accumulate contracted integrals
508 ////////////////////////////////////
509 if(lastoffset == 0)
510 {
511 contract_all(18, PRIM_INT__p_s_d_s, PRIM_PTR_INT__p_s_d_s);
512 contract_all(30, PRIM_INT__p_s_f_s, PRIM_PTR_INT__p_s_f_s);
513 contract_all(45, PRIM_INT__p_s_g_s, PRIM_PTR_INT__p_s_g_s);
514 contract_all(36, PRIM_INT__d_s_d_s, PRIM_PTR_INT__d_s_d_s);
515 contract_all(60, PRIM_INT__d_s_f_s, PRIM_PTR_INT__d_s_f_s);
516 contract_all(90, PRIM_INT__d_s_g_s, PRIM_PTR_INT__d_s_g_s);
517 }
518 else
519 {
520 contract(18, shelloffsets, PRIM_INT__p_s_d_s, PRIM_PTR_INT__p_s_d_s);
521 contract(30, shelloffsets, PRIM_INT__p_s_f_s, PRIM_PTR_INT__p_s_f_s);
522 contract(45, shelloffsets, PRIM_INT__p_s_g_s, PRIM_PTR_INT__p_s_g_s);
523 contract(36, shelloffsets, PRIM_INT__d_s_d_s, PRIM_PTR_INT__d_s_d_s);
524 contract(60, shelloffsets, PRIM_INT__d_s_f_s, PRIM_PTR_INT__d_s_f_s);
525 contract(90, shelloffsets, PRIM_INT__d_s_g_s, PRIM_PTR_INT__d_s_g_s);
526 PRIM_PTR_INT__p_s_d_s += lastoffset*18;
527 PRIM_PTR_INT__p_s_f_s += lastoffset*30;
528 PRIM_PTR_INT__p_s_g_s += lastoffset*45;
529 PRIM_PTR_INT__d_s_d_s += lastoffset*36;
530 PRIM_PTR_INT__d_s_f_s += lastoffset*60;
531 PRIM_PTR_INT__d_s_g_s += lastoffset*90;
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 const double hCD[3] = { Q.AB_x[cd+abcd], Q.AB_y[cd+abcd], Q.AB_z[cd+abcd] };
551
552 // set up HRR pointers
553 double const * restrict HRR_INT__p_s_d_s = INT__p_s_d_s + abcd * 18;
554 double const * restrict HRR_INT__p_s_f_s = INT__p_s_f_s + abcd * 30;
555 double const * restrict HRR_INT__p_s_g_s = INT__p_s_g_s + abcd * 45;
556 double const * restrict HRR_INT__d_s_d_s = INT__d_s_d_s + abcd * 36;
557 double const * restrict HRR_INT__d_s_f_s = INT__d_s_f_s + abcd * 60;
558 double const * restrict HRR_INT__d_s_g_s = INT__d_s_g_s + abcd * 90;
559 double * restrict HRR_INT__p_p_d_d = INT__p_p_d_d + real_abcd * 324;
560
561 // form INT__p_p_d_s
562 for(iket = 0; iket < 6; ++iket)
563 {
564 HRR_INT__p_p_d_s[0 * 6 + iket] = HRR_INT__d_s_d_s[0 * 6 + iket] + ( hAB[0] * HRR_INT__p_s_d_s[0 * 6 + iket] );
565
566 HRR_INT__p_p_d_s[1 * 6 + iket] = HRR_INT__d_s_d_s[1 * 6 + iket] + ( hAB[1] * HRR_INT__p_s_d_s[0 * 6 + iket] );
567
568 HRR_INT__p_p_d_s[2 * 6 + iket] = HRR_INT__d_s_d_s[2 * 6 + iket] + ( hAB[2] * HRR_INT__p_s_d_s[0 * 6 + iket] );
569
570 HRR_INT__p_p_d_s[3 * 6 + iket] = HRR_INT__d_s_d_s[1 * 6 + iket] + ( hAB[0] * HRR_INT__p_s_d_s[1 * 6 + iket] );
571
572 HRR_INT__p_p_d_s[4 * 6 + iket] = HRR_INT__d_s_d_s[3 * 6 + iket] + ( hAB[1] * HRR_INT__p_s_d_s[1 * 6 + iket] );
573
574 HRR_INT__p_p_d_s[5 * 6 + iket] = HRR_INT__d_s_d_s[4 * 6 + iket] + ( hAB[2] * HRR_INT__p_s_d_s[1 * 6 + iket] );
575
576 HRR_INT__p_p_d_s[6 * 6 + iket] = HRR_INT__d_s_d_s[2 * 6 + iket] + ( hAB[0] * HRR_INT__p_s_d_s[2 * 6 + iket] );
577
578 HRR_INT__p_p_d_s[7 * 6 + iket] = HRR_INT__d_s_d_s[4 * 6 + iket] + ( hAB[1] * HRR_INT__p_s_d_s[2 * 6 + iket] );
579
580 HRR_INT__p_p_d_s[8 * 6 + iket] = HRR_INT__d_s_d_s[5 * 6 + iket] + ( hAB[2] * HRR_INT__p_s_d_s[2 * 6 + iket] );
581
582 }
583
584
585 // form INT__p_p_f_s
586 for(iket = 0; iket < 10; ++iket)
587 {
588 HRR_INT__p_p_f_s[0 * 10 + iket] = HRR_INT__d_s_f_s[0 * 10 + iket] + ( hAB[0] * HRR_INT__p_s_f_s[0 * 10 + iket] );
589
590 HRR_INT__p_p_f_s[1 * 10 + iket] = HRR_INT__d_s_f_s[1 * 10 + iket] + ( hAB[1] * HRR_INT__p_s_f_s[0 * 10 + iket] );
591
592 HRR_INT__p_p_f_s[2 * 10 + iket] = HRR_INT__d_s_f_s[2 * 10 + iket] + ( hAB[2] * HRR_INT__p_s_f_s[0 * 10 + iket] );
593
594 HRR_INT__p_p_f_s[3 * 10 + iket] = HRR_INT__d_s_f_s[1 * 10 + iket] + ( hAB[0] * HRR_INT__p_s_f_s[1 * 10 + iket] );
595
596 HRR_INT__p_p_f_s[4 * 10 + iket] = HRR_INT__d_s_f_s[3 * 10 + iket] + ( hAB[1] * HRR_INT__p_s_f_s[1 * 10 + iket] );
597
598 HRR_INT__p_p_f_s[5 * 10 + iket] = HRR_INT__d_s_f_s[4 * 10 + iket] + ( hAB[2] * HRR_INT__p_s_f_s[1 * 10 + iket] );
599
600 HRR_INT__p_p_f_s[6 * 10 + iket] = HRR_INT__d_s_f_s[2 * 10 + iket] + ( hAB[0] * HRR_INT__p_s_f_s[2 * 10 + iket] );
601
602 HRR_INT__p_p_f_s[7 * 10 + iket] = HRR_INT__d_s_f_s[4 * 10 + iket] + ( hAB[1] * HRR_INT__p_s_f_s[2 * 10 + iket] );
603
604 HRR_INT__p_p_f_s[8 * 10 + iket] = HRR_INT__d_s_f_s[5 * 10 + iket] + ( hAB[2] * HRR_INT__p_s_f_s[2 * 10 + iket] );
605
606 }
607
608
609 // form INT__p_p_g_s
610 for(iket = 0; iket < 15; ++iket)
611 {
612 HRR_INT__p_p_g_s[0 * 15 + iket] = HRR_INT__d_s_g_s[0 * 15 + iket] + ( hAB[0] * HRR_INT__p_s_g_s[0 * 15 + iket] );
613
614 HRR_INT__p_p_g_s[1 * 15 + iket] = HRR_INT__d_s_g_s[1 * 15 + iket] + ( hAB[1] * HRR_INT__p_s_g_s[0 * 15 + iket] );
615
616 HRR_INT__p_p_g_s[2 * 15 + iket] = HRR_INT__d_s_g_s[2 * 15 + iket] + ( hAB[2] * HRR_INT__p_s_g_s[0 * 15 + iket] );
617
618 HRR_INT__p_p_g_s[3 * 15 + iket] = HRR_INT__d_s_g_s[1 * 15 + iket] + ( hAB[0] * HRR_INT__p_s_g_s[1 * 15 + iket] );
619
620 HRR_INT__p_p_g_s[4 * 15 + iket] = HRR_INT__d_s_g_s[3 * 15 + iket] + ( hAB[1] * HRR_INT__p_s_g_s[1 * 15 + iket] );
621
622 HRR_INT__p_p_g_s[5 * 15 + iket] = HRR_INT__d_s_g_s[4 * 15 + iket] + ( hAB[2] * HRR_INT__p_s_g_s[1 * 15 + iket] );
623
624 HRR_INT__p_p_g_s[6 * 15 + iket] = HRR_INT__d_s_g_s[2 * 15 + iket] + ( hAB[0] * HRR_INT__p_s_g_s[2 * 15 + iket] );
625
626 HRR_INT__p_p_g_s[7 * 15 + iket] = HRR_INT__d_s_g_s[4 * 15 + iket] + ( hAB[1] * HRR_INT__p_s_g_s[2 * 15 + iket] );
627
628 HRR_INT__p_p_g_s[8 * 15 + iket] = HRR_INT__d_s_g_s[5 * 15 + iket] + ( hAB[2] * HRR_INT__p_s_g_s[2 * 15 + iket] );
629
630 }
631
632
633 // form INT__p_p_d_p
634 for(ibra = 0; ibra < 9; ++ibra)
635 {
636 HRR_INT__p_p_d_p[ibra * 18 + 0] = HRR_INT__p_p_f_s[ibra * 10 + 0] + ( hCD[0] * HRR_INT__p_p_d_s[ibra * 6 + 0] );
637
638 HRR_INT__p_p_d_p[ibra * 18 + 1] = HRR_INT__p_p_f_s[ibra * 10 + 1] + ( hCD[1] * HRR_INT__p_p_d_s[ibra * 6 + 0] );
639
640 HRR_INT__p_p_d_p[ibra * 18 + 2] = HRR_INT__p_p_f_s[ibra * 10 + 2] + ( hCD[2] * HRR_INT__p_p_d_s[ibra * 6 + 0] );
641
642 HRR_INT__p_p_d_p[ibra * 18 + 3] = HRR_INT__p_p_f_s[ibra * 10 + 1] + ( hCD[0] * HRR_INT__p_p_d_s[ibra * 6 + 1] );
643
644 HRR_INT__p_p_d_p[ibra * 18 + 4] = HRR_INT__p_p_f_s[ibra * 10 + 3] + ( hCD[1] * HRR_INT__p_p_d_s[ibra * 6 + 1] );
645
646 HRR_INT__p_p_d_p[ibra * 18 + 5] = HRR_INT__p_p_f_s[ibra * 10 + 4] + ( hCD[2] * HRR_INT__p_p_d_s[ibra * 6 + 1] );
647
648 HRR_INT__p_p_d_p[ibra * 18 + 6] = HRR_INT__p_p_f_s[ibra * 10 + 2] + ( hCD[0] * HRR_INT__p_p_d_s[ibra * 6 + 2] );
649
650 HRR_INT__p_p_d_p[ibra * 18 + 7] = HRR_INT__p_p_f_s[ibra * 10 + 4] + ( hCD[1] * HRR_INT__p_p_d_s[ibra * 6 + 2] );
651
652 HRR_INT__p_p_d_p[ibra * 18 + 8] = HRR_INT__p_p_f_s[ibra * 10 + 5] + ( hCD[2] * HRR_INT__p_p_d_s[ibra * 6 + 2] );
653
654 HRR_INT__p_p_d_p[ibra * 18 + 9] = HRR_INT__p_p_f_s[ibra * 10 + 3] + ( hCD[0] * HRR_INT__p_p_d_s[ibra * 6 + 3] );
655
656 HRR_INT__p_p_d_p[ibra * 18 + 10] = HRR_INT__p_p_f_s[ibra * 10 + 6] + ( hCD[1] * HRR_INT__p_p_d_s[ibra * 6 + 3] );
657
658 HRR_INT__p_p_d_p[ibra * 18 + 11] = HRR_INT__p_p_f_s[ibra * 10 + 7] + ( hCD[2] * HRR_INT__p_p_d_s[ibra * 6 + 3] );
659
660 HRR_INT__p_p_d_p[ibra * 18 + 12] = HRR_INT__p_p_f_s[ibra * 10 + 4] + ( hCD[0] * HRR_INT__p_p_d_s[ibra * 6 + 4] );
661
662 HRR_INT__p_p_d_p[ibra * 18 + 13] = HRR_INT__p_p_f_s[ibra * 10 + 7] + ( hCD[1] * HRR_INT__p_p_d_s[ibra * 6 + 4] );
663
664 HRR_INT__p_p_d_p[ibra * 18 + 14] = HRR_INT__p_p_f_s[ibra * 10 + 8] + ( hCD[2] * HRR_INT__p_p_d_s[ibra * 6 + 4] );
665
666 HRR_INT__p_p_d_p[ibra * 18 + 15] = HRR_INT__p_p_f_s[ibra * 10 + 5] + ( hCD[0] * HRR_INT__p_p_d_s[ibra * 6 + 5] );
667
668 HRR_INT__p_p_d_p[ibra * 18 + 16] = HRR_INT__p_p_f_s[ibra * 10 + 8] + ( hCD[1] * HRR_INT__p_p_d_s[ibra * 6 + 5] );
669
670 HRR_INT__p_p_d_p[ibra * 18 + 17] = HRR_INT__p_p_f_s[ibra * 10 + 9] + ( hCD[2] * HRR_INT__p_p_d_s[ibra * 6 + 5] );
671
672 }
673
674 // form INT__p_p_f_p
675 HRR_L_f_p(
676 HRR_INT__p_p_f_p,
677 HRR_INT__p_p_f_s,
678 HRR_INT__p_p_g_s,
679 hCD, 9);
680
681 // form INT__p_p_d_d
682 HRR_L_d_d(
683 HRR_INT__p_p_d_d,
684 HRR_INT__p_p_d_p,
685 HRR_INT__p_p_f_p,
686 hCD, 9);
687
688
689 } // close HRR loop
690
691
692 } // close loop cdbatch
693
694 istart = iend;
695 } // close loop over ab
696
697 return P.nshell12_clip * Q.nshell12_clip;
698 }
699
700