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_d_d_p_p(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__d_d_p_p)8 int ostei_d_d_p_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__d_d_p_p)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__d_d_p_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__d_s_p_s = work + (SIMINT_NSHELL_SIMD * 0);
30 double * const INT__d_s_d_s = work + (SIMINT_NSHELL_SIMD * 18);
31 double * const INT__f_s_p_s = work + (SIMINT_NSHELL_SIMD * 54);
32 double * const INT__f_s_d_s = work + (SIMINT_NSHELL_SIMD * 84);
33 double * const INT__g_s_p_s = work + (SIMINT_NSHELL_SIMD * 144);
34 double * const INT__g_s_d_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__p_s_s_s = primwork + 7;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_p_s = primwork + 25;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 43;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_p_s = primwork + 73;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_d_s = primwork + 109;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 145;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_p_s = primwork + 185;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_d_s = primwork + 245;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 305;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_p_s = primwork + 350;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_d_s = primwork + 440;
48 double * const hrrwork = (double *)(primwork + 530);
49 double * const HRR_INT__d_p_p_s = hrrwork + 0;
50 double * const HRR_INT__d_p_d_s = hrrwork + 54;
51 double * const HRR_INT__d_d_p_s = hrrwork + 162;
52 double * const HRR_INT__d_d_d_s = hrrwork + 270;
53 double * const HRR_INT__f_p_p_s = hrrwork + 486;
54 double * const HRR_INT__f_p_d_s = hrrwork + 576;
55
56
57 // Create constants
58 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
59 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
60 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
61 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
62 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
63
64
65 ////////////////////////////////////////
66 // Loop over shells and primitives
67 ////////////////////////////////////////
68
69 real_abcd = 0;
70 istart = 0;
71 for(ab = 0; ab < P.nshell12_clip; ++ab)
72 {
73 const int iend = istart + P.nprim12[ab];
74
75 cd = 0;
76 jstart = 0;
77
78 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
79 {
80 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
81 int jend = jstart;
82 for(i = 0; i < nshellbatch; i++)
83 jend += Q.nprim12[cd+i];
84
85 // Clear the beginning of the workspace (where we are accumulating integrals)
86 memset(work, 0, SIMINT_NSHELL_SIMD * 279 * sizeof(double));
87 abcd = 0;
88
89
90 for(i = istart; i < iend; ++i)
91 {
92 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
93
94 if(check_screen)
95 {
96 // Skip this whole thing if always insignificant
97 if((P.screen[i] * Q.screen_max) < screen_tol)
98 continue;
99 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
100 }
101
102 icd = 0;
103 iprimcd = 0;
104 nprim_icd = Q.nprim12[cd];
105 double * restrict PRIM_PTR_INT__d_s_p_s = INT__d_s_p_s + abcd * 18;
106 double * restrict PRIM_PTR_INT__d_s_d_s = INT__d_s_d_s + abcd * 36;
107 double * restrict PRIM_PTR_INT__f_s_p_s = INT__f_s_p_s + abcd * 30;
108 double * restrict PRIM_PTR_INT__f_s_d_s = INT__f_s_d_s + abcd * 60;
109 double * restrict PRIM_PTR_INT__g_s_p_s = INT__g_s_p_s + abcd * 45;
110 double * restrict PRIM_PTR_INT__g_s_d_s = INT__g_s_d_s + abcd * 90;
111
112
113
114 // Load these one per loop over i
115 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
116 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
117 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
118
119 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
120
121 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
122 {
123 // calculate the shell offsets
124 // these are the offset from the shell pointed to by cd
125 // for each element
126 int shelloffsets[SIMINT_SIMD_LEN] = {0};
127 int lastoffset = 0;
128 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
129
130 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
131 {
132 // Handle if the first element of the vector is a new shell
133 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
134 {
135 nprim_icd += Q.nprim12[cd + (++icd)];
136 PRIM_PTR_INT__d_s_p_s += 18;
137 PRIM_PTR_INT__d_s_d_s += 36;
138 PRIM_PTR_INT__f_s_p_s += 30;
139 PRIM_PTR_INT__f_s_d_s += 60;
140 PRIM_PTR_INT__g_s_p_s += 45;
141 PRIM_PTR_INT__g_s_d_s += 90;
142 }
143 iprimcd++;
144 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
145 {
146 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
147 {
148 shelloffsets[n] = shelloffsets[n-1] + 1;
149 lastoffset++;
150 nprim_icd += Q.nprim12[cd + (++icd)];
151 }
152 else
153 shelloffsets[n] = shelloffsets[n-1];
154 iprimcd++;
155 }
156 }
157 else
158 iprimcd += SIMINT_SIMD_LEN;
159
160 // Do we have to compute this vector (or has it been screened out)?
161 // (not_screened != 0 means we have to do this vector)
162 if(check_screen)
163 {
164 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
165 if(vmax < screen_tol)
166 {
167 PRIM_PTR_INT__d_s_p_s += lastoffset*18;
168 PRIM_PTR_INT__d_s_d_s += lastoffset*36;
169 PRIM_PTR_INT__f_s_p_s += lastoffset*30;
170 PRIM_PTR_INT__f_s_d_s += lastoffset*60;
171 PRIM_PTR_INT__g_s_p_s += lastoffset*45;
172 PRIM_PTR_INT__g_s_d_s += lastoffset*90;
173 continue;
174 }
175 }
176
177 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
178 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
179 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
180 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
181
182
183 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
184 SIMINT_DBLTYPE PQ[3];
185 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
186 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
187 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
188 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
189 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
190 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
191
192 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
193 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
194 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
195 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
196 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
197 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
198 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
199
200 // NOTE: Minus sign!
201 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
202 SIMINT_DBLTYPE aop_PQ[3];
203 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
204 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
205 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
206
207 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
208 SIMINT_DBLTYPE aoq_PQ[3];
209 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
210 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
211 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
212 // Put a minus sign here so we don't have to in RR routines
213 a_over_q = SIMINT_NEG(a_over_q);
214
215
216 //////////////////////////////////////////////
217 // Fjt function section
218 // Maximum v value: 6
219 //////////////////////////////////////////////
220 // The parameter to the Fjt function
221 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
222
223
224 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
225
226
227 boys_F_split(PRIM_INT__s_s_s_s, F_x, 6);
228 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
229 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
230 for(n = 0; n <= 6; n++)
231 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
232
233 //////////////////////////////////////////////
234 // Primitive integrals: Vertical recurrance
235 //////////////////////////////////////////////
236
237 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
238 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
239 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
240 const SIMINT_DBLTYPE vrr_const_1_over_2q = 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
246
247
248 // Forming PRIM_INT__p_s_s_s[6 * 3];
249 for(n = 0; n < 6; ++n) // loop over orders of auxiliary function
250 {
251
252 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
253 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]);
254
255 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
256 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]);
257
258 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
259 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]);
260
261 }
262
263
264
265 // Forming PRIM_INT__d_s_s_s[5 * 6];
266 for(n = 0; n < 5; ++n) // loop over orders of auxiliary function
267 {
268
269 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
270 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]);
271 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]);
272
273 PRIM_INT__d_s_s_s[n * 6 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 0]);
274 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]);
275
276 PRIM_INT__d_s_s_s[n * 6 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 0]);
277 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]);
278
279 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
280 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]);
281 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]);
282
283 PRIM_INT__d_s_s_s[n * 6 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 1]);
284 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]);
285
286 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
287 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]);
288 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]);
289
290 }
291
292
293
294 // Forming PRIM_INT__d_s_p_s[2 * 18];
295 for(n = 0; n < 2; ++n) // loop over orders of auxiliary function
296 {
297
298 PRIM_INT__d_s_p_s[n * 18 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
299 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]);
300 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]);
301
302 PRIM_INT__d_s_p_s[n * 18 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
303 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]);
304
305 PRIM_INT__d_s_p_s[n * 18 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
306 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]);
307
308 PRIM_INT__d_s_p_s[n * 18 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 1]);
309 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]);
310 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]);
311
312 PRIM_INT__d_s_p_s[n * 18 + 4] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 1]);
313 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]);
314 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]);
315
316 PRIM_INT__d_s_p_s[n * 18 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 1]);
317 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]);
318
319 PRIM_INT__d_s_p_s[n * 18 + 6] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 2]);
320 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]);
321 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]);
322
323 PRIM_INT__d_s_p_s[n * 18 + 7] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 2]);
324 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]);
325
326 PRIM_INT__d_s_p_s[n * 18 + 8] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 2]);
327 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]);
328 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]);
329
330 PRIM_INT__d_s_p_s[n * 18 + 9] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
331 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]);
332
333 PRIM_INT__d_s_p_s[n * 18 + 10] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
334 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]);
335 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]);
336
337 PRIM_INT__d_s_p_s[n * 18 + 11] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
338 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]);
339
340 PRIM_INT__d_s_p_s[n * 18 + 12] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 4]);
341 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]);
342
343 PRIM_INT__d_s_p_s[n * 18 + 13] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 4]);
344 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]);
345 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]);
346
347 PRIM_INT__d_s_p_s[n * 18 + 14] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 4]);
348 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]);
349 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]);
350
351 PRIM_INT__d_s_p_s[n * 18 + 15] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
352 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]);
353
354 PRIM_INT__d_s_p_s[n * 18 + 16] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 5]);
355 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]);
356
357 PRIM_INT__d_s_p_s[n * 18 + 17] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
358 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]);
359 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]);
360
361 }
362
363
364
365 // Forming PRIM_INT__p_s_p_s[2 * 9];
366 for(n = 0; n < 2; ++n) // loop over orders of auxiliary function
367 {
368
369 PRIM_INT__p_s_p_s[n * 9 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
370 PRIM_INT__p_s_p_s[n * 9 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__p_s_p_s[n * 9 + 0]);
371 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]);
372
373 PRIM_INT__p_s_p_s[n * 9 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__p_s_s_s[n * 3 + 0]);
374 PRIM_INT__p_s_p_s[n * 9 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__p_s_p_s[n * 9 + 1]);
375
376 PRIM_INT__p_s_p_s[n * 9 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__p_s_s_s[n * 3 + 0]);
377 PRIM_INT__p_s_p_s[n * 9 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__p_s_p_s[n * 9 + 2]);
378
379 PRIM_INT__p_s_p_s[n * 9 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__p_s_s_s[n * 3 + 1]);
380 PRIM_INT__p_s_p_s[n * 9 + 3] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__p_s_s_s[(n+1) * 3 + 1], PRIM_INT__p_s_p_s[n * 9 + 3]);
381
382 PRIM_INT__p_s_p_s[n * 9 + 4] = SIMINT_MUL(Q_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
383 PRIM_INT__p_s_p_s[n * 9 + 4] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__p_s_s_s[(n+1) * 3 + 1], PRIM_INT__p_s_p_s[n * 9 + 4]);
384 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]);
385
386 PRIM_INT__p_s_p_s[n * 9 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__p_s_s_s[n * 3 + 1]);
387 PRIM_INT__p_s_p_s[n * 9 + 5] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__p_s_s_s[(n+1) * 3 + 1], PRIM_INT__p_s_p_s[n * 9 + 5]);
388
389 PRIM_INT__p_s_p_s[n * 9 + 6] = SIMINT_MUL(Q_PA[0], PRIM_INT__p_s_s_s[n * 3 + 2]);
390 PRIM_INT__p_s_p_s[n * 9 + 6] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__p_s_s_s[(n+1) * 3 + 2], PRIM_INT__p_s_p_s[n * 9 + 6]);
391
392 PRIM_INT__p_s_p_s[n * 9 + 7] = SIMINT_MUL(Q_PA[1], PRIM_INT__p_s_s_s[n * 3 + 2]);
393 PRIM_INT__p_s_p_s[n * 9 + 7] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__p_s_s_s[(n+1) * 3 + 2], PRIM_INT__p_s_p_s[n * 9 + 7]);
394
395 PRIM_INT__p_s_p_s[n * 9 + 8] = SIMINT_MUL(Q_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
396 PRIM_INT__p_s_p_s[n * 9 + 8] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__p_s_s_s[(n+1) * 3 + 2], PRIM_INT__p_s_p_s[n * 9 + 8]);
397 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]);
398
399 }
400
401
402 VRR_K_d_s_d_s(
403 PRIM_INT__d_s_d_s,
404 PRIM_INT__d_s_p_s,
405 PRIM_INT__d_s_s_s,
406 PRIM_INT__p_s_p_s,
407 Q_PA,
408 a_over_q,
409 aoq_PQ,
410 one_over_2pq,
411 one_over_2q,
412 1);
413
414
415
416 // Forming PRIM_INT__f_s_s_s[4 * 10];
417 for(n = 0; n < 4; ++n) // loop over orders of auxiliary function
418 {
419
420 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
421 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]);
422 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]);
423
424 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
425 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]);
426
427 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
428 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]);
429
430 PRIM_INT__f_s_s_s[n * 10 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
431 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]);
432
433 PRIM_INT__f_s_s_s[n * 10 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 1]);
434 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]);
435
436 PRIM_INT__f_s_s_s[n * 10 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
437 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]);
438
439 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
440 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]);
441 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]);
442
443 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
444 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]);
445
446 PRIM_INT__f_s_s_s[n * 10 + 8] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 5]);
447 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]);
448
449 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
450 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]);
451 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]);
452
453 }
454
455
456 VRR_K_f_s_p_s(
457 PRIM_INT__f_s_p_s,
458 PRIM_INT__f_s_s_s,
459 PRIM_INT__d_s_s_s,
460 Q_PA,
461 aoq_PQ,
462 one_over_2pq,
463 2);
464
465
466 VRR_K_f_s_d_s(
467 PRIM_INT__f_s_d_s,
468 PRIM_INT__f_s_p_s,
469 PRIM_INT__f_s_s_s,
470 PRIM_INT__d_s_p_s,
471 Q_PA,
472 a_over_q,
473 aoq_PQ,
474 one_over_2pq,
475 one_over_2q,
476 1);
477
478
479 VRR_I_g_s_s_s(
480 PRIM_INT__g_s_s_s,
481 PRIM_INT__f_s_s_s,
482 PRIM_INT__d_s_s_s,
483 P_PA,
484 a_over_p,
485 aop_PQ,
486 one_over_2p,
487 3);
488
489
490 VRR_K_g_s_p_s(
491 PRIM_INT__g_s_p_s,
492 PRIM_INT__g_s_s_s,
493 PRIM_INT__f_s_s_s,
494 Q_PA,
495 aoq_PQ,
496 one_over_2pq,
497 2);
498
499
500 ostei_general_vrr_K(4, 0, 2, 0, 1,
501 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
502 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);
503
504
505
506
507 ////////////////////////////////////
508 // Accumulate contracted integrals
509 ////////////////////////////////////
510 if(lastoffset == 0)
511 {
512 contract_all(18, PRIM_INT__d_s_p_s, PRIM_PTR_INT__d_s_p_s);
513 contract_all(36, PRIM_INT__d_s_d_s, PRIM_PTR_INT__d_s_d_s);
514 contract_all(30, PRIM_INT__f_s_p_s, PRIM_PTR_INT__f_s_p_s);
515 contract_all(60, PRIM_INT__f_s_d_s, PRIM_PTR_INT__f_s_d_s);
516 contract_all(45, PRIM_INT__g_s_p_s, PRIM_PTR_INT__g_s_p_s);
517 contract_all(90, PRIM_INT__g_s_d_s, PRIM_PTR_INT__g_s_d_s);
518 }
519 else
520 {
521 contract(18, shelloffsets, PRIM_INT__d_s_p_s, PRIM_PTR_INT__d_s_p_s);
522 contract(36, shelloffsets, PRIM_INT__d_s_d_s, PRIM_PTR_INT__d_s_d_s);
523 contract(30, shelloffsets, PRIM_INT__f_s_p_s, PRIM_PTR_INT__f_s_p_s);
524 contract(60, shelloffsets, PRIM_INT__f_s_d_s, PRIM_PTR_INT__f_s_d_s);
525 contract(45, shelloffsets, PRIM_INT__g_s_p_s, PRIM_PTR_INT__g_s_p_s);
526 contract(90, shelloffsets, PRIM_INT__g_s_d_s, PRIM_PTR_INT__g_s_d_s);
527 PRIM_PTR_INT__d_s_p_s += lastoffset*18;
528 PRIM_PTR_INT__d_s_d_s += lastoffset*36;
529 PRIM_PTR_INT__f_s_p_s += lastoffset*30;
530 PRIM_PTR_INT__f_s_d_s += lastoffset*60;
531 PRIM_PTR_INT__g_s_p_s += lastoffset*45;
532 PRIM_PTR_INT__g_s_d_s += lastoffset*90;
533 }
534
535 } // close loop over j
536 } // close loop over i
537
538 //Advance to the next batch
539 jstart = SIMINT_SIMD_ROUND(jend);
540
541 //////////////////////////////////////////////
542 // Contracted integrals: Horizontal recurrance
543 //////////////////////////////////////////////
544
545
546 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
547
548
549 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
550 {
551 const double hCD[3] = { Q.AB_x[cd+abcd], Q.AB_y[cd+abcd], Q.AB_z[cd+abcd] };
552
553 // set up HRR pointers
554 double const * restrict HRR_INT__d_s_p_s = INT__d_s_p_s + abcd * 18;
555 double const * restrict HRR_INT__d_s_d_s = INT__d_s_d_s + abcd * 36;
556 double const * restrict HRR_INT__f_s_p_s = INT__f_s_p_s + abcd * 30;
557 double const * restrict HRR_INT__f_s_d_s = INT__f_s_d_s + abcd * 60;
558 double const * restrict HRR_INT__g_s_p_s = INT__g_s_p_s + abcd * 45;
559 double const * restrict HRR_INT__g_s_d_s = INT__g_s_d_s + abcd * 90;
560 double * restrict HRR_INT__d_d_p_p = INT__d_d_p_p + real_abcd * 324;
561
562 // form INT__d_p_p_s
563 for(iket = 0; iket < 3; ++iket)
564 {
565 HRR_INT__d_p_p_s[0 * 3 + iket] = HRR_INT__f_s_p_s[0 * 3 + iket] + ( hAB[0] * HRR_INT__d_s_p_s[0 * 3 + iket] );
566
567 HRR_INT__d_p_p_s[1 * 3 + iket] = HRR_INT__f_s_p_s[1 * 3 + iket] + ( hAB[1] * HRR_INT__d_s_p_s[0 * 3 + iket] );
568
569 HRR_INT__d_p_p_s[2 * 3 + iket] = HRR_INT__f_s_p_s[2 * 3 + iket] + ( hAB[2] * HRR_INT__d_s_p_s[0 * 3 + iket] );
570
571 HRR_INT__d_p_p_s[3 * 3 + iket] = HRR_INT__f_s_p_s[1 * 3 + iket] + ( hAB[0] * HRR_INT__d_s_p_s[1 * 3 + iket] );
572
573 HRR_INT__d_p_p_s[4 * 3 + iket] = HRR_INT__f_s_p_s[3 * 3 + iket] + ( hAB[1] * HRR_INT__d_s_p_s[1 * 3 + iket] );
574
575 HRR_INT__d_p_p_s[5 * 3 + iket] = HRR_INT__f_s_p_s[4 * 3 + iket] + ( hAB[2] * HRR_INT__d_s_p_s[1 * 3 + iket] );
576
577 HRR_INT__d_p_p_s[6 * 3 + iket] = HRR_INT__f_s_p_s[2 * 3 + iket] + ( hAB[0] * HRR_INT__d_s_p_s[2 * 3 + iket] );
578
579 HRR_INT__d_p_p_s[7 * 3 + iket] = HRR_INT__f_s_p_s[4 * 3 + iket] + ( hAB[1] * HRR_INT__d_s_p_s[2 * 3 + iket] );
580
581 HRR_INT__d_p_p_s[8 * 3 + iket] = HRR_INT__f_s_p_s[5 * 3 + iket] + ( hAB[2] * HRR_INT__d_s_p_s[2 * 3 + iket] );
582
583 HRR_INT__d_p_p_s[9 * 3 + iket] = HRR_INT__f_s_p_s[3 * 3 + iket] + ( hAB[0] * HRR_INT__d_s_p_s[3 * 3 + iket] );
584
585 HRR_INT__d_p_p_s[10 * 3 + iket] = HRR_INT__f_s_p_s[6 * 3 + iket] + ( hAB[1] * HRR_INT__d_s_p_s[3 * 3 + iket] );
586
587 HRR_INT__d_p_p_s[11 * 3 + iket] = HRR_INT__f_s_p_s[7 * 3 + iket] + ( hAB[2] * HRR_INT__d_s_p_s[3 * 3 + iket] );
588
589 HRR_INT__d_p_p_s[12 * 3 + iket] = HRR_INT__f_s_p_s[4 * 3 + iket] + ( hAB[0] * HRR_INT__d_s_p_s[4 * 3 + iket] );
590
591 HRR_INT__d_p_p_s[13 * 3 + iket] = HRR_INT__f_s_p_s[7 * 3 + iket] + ( hAB[1] * HRR_INT__d_s_p_s[4 * 3 + iket] );
592
593 HRR_INT__d_p_p_s[14 * 3 + iket] = HRR_INT__f_s_p_s[8 * 3 + iket] + ( hAB[2] * HRR_INT__d_s_p_s[4 * 3 + iket] );
594
595 HRR_INT__d_p_p_s[15 * 3 + iket] = HRR_INT__f_s_p_s[5 * 3 + iket] + ( hAB[0] * HRR_INT__d_s_p_s[5 * 3 + iket] );
596
597 HRR_INT__d_p_p_s[16 * 3 + iket] = HRR_INT__f_s_p_s[8 * 3 + iket] + ( hAB[1] * HRR_INT__d_s_p_s[5 * 3 + iket] );
598
599 HRR_INT__d_p_p_s[17 * 3 + iket] = HRR_INT__f_s_p_s[9 * 3 + iket] + ( hAB[2] * HRR_INT__d_s_p_s[5 * 3 + iket] );
600
601 }
602
603
604 // form INT__d_p_d_s
605 for(iket = 0; iket < 6; ++iket)
606 {
607 HRR_INT__d_p_d_s[0 * 6 + iket] = HRR_INT__f_s_d_s[0 * 6 + iket] + ( hAB[0] * HRR_INT__d_s_d_s[0 * 6 + iket] );
608
609 HRR_INT__d_p_d_s[1 * 6 + iket] = HRR_INT__f_s_d_s[1 * 6 + iket] + ( hAB[1] * HRR_INT__d_s_d_s[0 * 6 + iket] );
610
611 HRR_INT__d_p_d_s[2 * 6 + iket] = HRR_INT__f_s_d_s[2 * 6 + iket] + ( hAB[2] * HRR_INT__d_s_d_s[0 * 6 + iket] );
612
613 HRR_INT__d_p_d_s[3 * 6 + iket] = HRR_INT__f_s_d_s[1 * 6 + iket] + ( hAB[0] * HRR_INT__d_s_d_s[1 * 6 + iket] );
614
615 HRR_INT__d_p_d_s[4 * 6 + iket] = HRR_INT__f_s_d_s[3 * 6 + iket] + ( hAB[1] * HRR_INT__d_s_d_s[1 * 6 + iket] );
616
617 HRR_INT__d_p_d_s[5 * 6 + iket] = HRR_INT__f_s_d_s[4 * 6 + iket] + ( hAB[2] * HRR_INT__d_s_d_s[1 * 6 + iket] );
618
619 HRR_INT__d_p_d_s[6 * 6 + iket] = HRR_INT__f_s_d_s[2 * 6 + iket] + ( hAB[0] * HRR_INT__d_s_d_s[2 * 6 + iket] );
620
621 HRR_INT__d_p_d_s[7 * 6 + iket] = HRR_INT__f_s_d_s[4 * 6 + iket] + ( hAB[1] * HRR_INT__d_s_d_s[2 * 6 + iket] );
622
623 HRR_INT__d_p_d_s[8 * 6 + iket] = HRR_INT__f_s_d_s[5 * 6 + iket] + ( hAB[2] * HRR_INT__d_s_d_s[2 * 6 + iket] );
624
625 HRR_INT__d_p_d_s[9 * 6 + iket] = HRR_INT__f_s_d_s[3 * 6 + iket] + ( hAB[0] * HRR_INT__d_s_d_s[3 * 6 + iket] );
626
627 HRR_INT__d_p_d_s[10 * 6 + iket] = HRR_INT__f_s_d_s[6 * 6 + iket] + ( hAB[1] * HRR_INT__d_s_d_s[3 * 6 + iket] );
628
629 HRR_INT__d_p_d_s[11 * 6 + iket] = HRR_INT__f_s_d_s[7 * 6 + iket] + ( hAB[2] * HRR_INT__d_s_d_s[3 * 6 + iket] );
630
631 HRR_INT__d_p_d_s[12 * 6 + iket] = HRR_INT__f_s_d_s[4 * 6 + iket] + ( hAB[0] * HRR_INT__d_s_d_s[4 * 6 + iket] );
632
633 HRR_INT__d_p_d_s[13 * 6 + iket] = HRR_INT__f_s_d_s[7 * 6 + iket] + ( hAB[1] * HRR_INT__d_s_d_s[4 * 6 + iket] );
634
635 HRR_INT__d_p_d_s[14 * 6 + iket] = HRR_INT__f_s_d_s[8 * 6 + iket] + ( hAB[2] * HRR_INT__d_s_d_s[4 * 6 + iket] );
636
637 HRR_INT__d_p_d_s[15 * 6 + iket] = HRR_INT__f_s_d_s[5 * 6 + iket] + ( hAB[0] * HRR_INT__d_s_d_s[5 * 6 + iket] );
638
639 HRR_INT__d_p_d_s[16 * 6 + iket] = HRR_INT__f_s_d_s[8 * 6 + iket] + ( hAB[1] * HRR_INT__d_s_d_s[5 * 6 + iket] );
640
641 HRR_INT__d_p_d_s[17 * 6 + iket] = HRR_INT__f_s_d_s[9 * 6 + iket] + ( hAB[2] * HRR_INT__d_s_d_s[5 * 6 + iket] );
642
643 }
644
645
646 // form INT__f_p_p_s
647 HRR_J_f_p(
648 HRR_INT__f_p_p_s,
649 HRR_INT__f_s_p_s,
650 HRR_INT__g_s_p_s,
651 hAB, 3);
652
653 // form INT__f_p_d_s
654 HRR_J_f_p(
655 HRR_INT__f_p_d_s,
656 HRR_INT__f_s_d_s,
657 HRR_INT__g_s_d_s,
658 hAB, 6);
659
660 // form INT__d_d_p_s
661 HRR_J_d_d(
662 HRR_INT__d_d_p_s,
663 HRR_INT__d_p_p_s,
664 HRR_INT__f_p_p_s,
665 hAB, 3);
666
667 // form INT__d_d_d_s
668 HRR_J_d_d(
669 HRR_INT__d_d_d_s,
670 HRR_INT__d_p_d_s,
671 HRR_INT__f_p_d_s,
672 hAB, 6);
673
674 // form INT__d_d_p_p
675 for(ibra = 0; ibra < 36; ++ibra)
676 {
677 HRR_INT__d_d_p_p[ibra * 9 + 0] = HRR_INT__d_d_d_s[ibra * 6 + 0] + ( hCD[0] * HRR_INT__d_d_p_s[ibra * 3 + 0] );
678
679 HRR_INT__d_d_p_p[ibra * 9 + 1] = HRR_INT__d_d_d_s[ibra * 6 + 1] + ( hCD[1] * HRR_INT__d_d_p_s[ibra * 3 + 0] );
680
681 HRR_INT__d_d_p_p[ibra * 9 + 2] = HRR_INT__d_d_d_s[ibra * 6 + 2] + ( hCD[2] * HRR_INT__d_d_p_s[ibra * 3 + 0] );
682
683 HRR_INT__d_d_p_p[ibra * 9 + 3] = HRR_INT__d_d_d_s[ibra * 6 + 1] + ( hCD[0] * HRR_INT__d_d_p_s[ibra * 3 + 1] );
684
685 HRR_INT__d_d_p_p[ibra * 9 + 4] = HRR_INT__d_d_d_s[ibra * 6 + 3] + ( hCD[1] * HRR_INT__d_d_p_s[ibra * 3 + 1] );
686
687 HRR_INT__d_d_p_p[ibra * 9 + 5] = HRR_INT__d_d_d_s[ibra * 6 + 4] + ( hCD[2] * HRR_INT__d_d_p_s[ibra * 3 + 1] );
688
689 HRR_INT__d_d_p_p[ibra * 9 + 6] = HRR_INT__d_d_d_s[ibra * 6 + 2] + ( hCD[0] * HRR_INT__d_d_p_s[ibra * 3 + 2] );
690
691 HRR_INT__d_d_p_p[ibra * 9 + 7] = HRR_INT__d_d_d_s[ibra * 6 + 4] + ( hCD[1] * HRR_INT__d_d_p_s[ibra * 3 + 2] );
692
693 HRR_INT__d_d_p_p[ibra * 9 + 8] = HRR_INT__d_d_d_s[ibra * 6 + 5] + ( hCD[2] * HRR_INT__d_d_p_s[ibra * 3 + 2] );
694
695 }
696
697
698 } // close HRR loop
699
700
701 } // close loop cdbatch
702
703 istart = iend;
704 } // close loop over ab
705
706 return P.nshell12_clip * Q.nshell12_clip;
707 }
708
709