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_k_g_s_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__k_g_s_s)8 int ostei_k_g_s_s(struct simint_multi_shellpair const P,
9 struct simint_multi_shellpair const Q,
10 double screen_tol,
11 double * const restrict work,
12 double * const restrict INT__k_g_s_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__k_g_s_s);
17 int ab, cd, abcd;
18 int istart, jstart;
19 int iprimcd, nprim_icd, icd;
20 const int check_screen = (screen_tol > 0.0);
21 int i, j;
22 int n;
23 int not_screened;
24 int real_abcd;
25 int iket;
26
27 // partition workspace
28 double * const INT__k_s_s_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__l_s_s_s = work + (SIMINT_NSHELL_SIMD * 36);
30 double * const INT__m_s_s_s = work + (SIMINT_NSHELL_SIMD * 81);
31 double * const INT__n_s_s_s = work + (SIMINT_NSHELL_SIMD * 136);
32 double * const INT__o_s_s_s = work + (SIMINT_NSHELL_SIMD * 202);
33 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*280);
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 + 12;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 45;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 105;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 195;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 315;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 462;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_s_s = primwork + 630;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_s_s = primwork + 810;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_s_s = primwork + 990;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_s_s = primwork + 1155;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__o_s_s_s = primwork + 1287;
46 double * const hrrwork = (double *)(primwork + 1365);
47 double * const HRR_INT__k_p_s_s = hrrwork + 0;
48 double * const HRR_INT__k_d_s_s = hrrwork + 108;
49 double * const HRR_INT__k_f_s_s = hrrwork + 324;
50 double * const HRR_INT__l_p_s_s = hrrwork + 684;
51 double * const HRR_INT__l_d_s_s = hrrwork + 819;
52 double * const HRR_INT__l_f_s_s = hrrwork + 1089;
53 double * const HRR_INT__m_p_s_s = hrrwork + 1539;
54 double * const HRR_INT__m_d_s_s = hrrwork + 1704;
55 double * const HRR_INT__n_p_s_s = hrrwork + 2034;
56
57
58 // Create constants
59 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
60 const SIMINT_DBLTYPE const_10 = SIMINT_DBLSET1(10);
61 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
62 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
63 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
64 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
65 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
66 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
67 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
68 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
69 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
70
71
72 ////////////////////////////////////////
73 // Loop over shells and primitives
74 ////////////////////////////////////////
75
76 real_abcd = 0;
77 istart = 0;
78 for(ab = 0; ab < P.nshell12_clip; ++ab)
79 {
80 const int iend = istart + P.nprim12[ab];
81
82 cd = 0;
83 jstart = 0;
84
85 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
86 {
87 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
88 int jend = jstart;
89 for(i = 0; i < nshellbatch; i++)
90 jend += Q.nprim12[cd+i];
91
92 // Clear the beginning of the workspace (where we are accumulating integrals)
93 memset(work, 0, SIMINT_NSHELL_SIMD * 280 * sizeof(double));
94 abcd = 0;
95
96
97 for(i = istart; i < iend; ++i)
98 {
99 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
100
101 if(check_screen)
102 {
103 // Skip this whole thing if always insignificant
104 if((P.screen[i] * Q.screen_max) < screen_tol)
105 continue;
106 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
107 }
108
109 icd = 0;
110 iprimcd = 0;
111 nprim_icd = Q.nprim12[cd];
112 double * restrict PRIM_PTR_INT__k_s_s_s = INT__k_s_s_s + abcd * 36;
113 double * restrict PRIM_PTR_INT__l_s_s_s = INT__l_s_s_s + abcd * 45;
114 double * restrict PRIM_PTR_INT__m_s_s_s = INT__m_s_s_s + abcd * 55;
115 double * restrict PRIM_PTR_INT__n_s_s_s = INT__n_s_s_s + abcd * 66;
116 double * restrict PRIM_PTR_INT__o_s_s_s = INT__o_s_s_s + abcd * 78;
117
118
119
120 // Load these one per loop over i
121 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
122 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
123 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
124
125 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
126
127 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
128 {
129 // calculate the shell offsets
130 // these are the offset from the shell pointed to by cd
131 // for each element
132 int shelloffsets[SIMINT_SIMD_LEN] = {0};
133 int lastoffset = 0;
134 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
135
136 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
137 {
138 // Handle if the first element of the vector is a new shell
139 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
140 {
141 nprim_icd += Q.nprim12[cd + (++icd)];
142 PRIM_PTR_INT__k_s_s_s += 36;
143 PRIM_PTR_INT__l_s_s_s += 45;
144 PRIM_PTR_INT__m_s_s_s += 55;
145 PRIM_PTR_INT__n_s_s_s += 66;
146 PRIM_PTR_INT__o_s_s_s += 78;
147 }
148 iprimcd++;
149 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
150 {
151 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
152 {
153 shelloffsets[n] = shelloffsets[n-1] + 1;
154 lastoffset++;
155 nprim_icd += Q.nprim12[cd + (++icd)];
156 }
157 else
158 shelloffsets[n] = shelloffsets[n-1];
159 iprimcd++;
160 }
161 }
162 else
163 iprimcd += SIMINT_SIMD_LEN;
164
165 // Do we have to compute this vector (or has it been screened out)?
166 // (not_screened != 0 means we have to do this vector)
167 if(check_screen)
168 {
169 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
170 if(vmax < screen_tol)
171 {
172 PRIM_PTR_INT__k_s_s_s += lastoffset*36;
173 PRIM_PTR_INT__l_s_s_s += lastoffset*45;
174 PRIM_PTR_INT__m_s_s_s += lastoffset*55;
175 PRIM_PTR_INT__n_s_s_s += lastoffset*66;
176 PRIM_PTR_INT__o_s_s_s += lastoffset*78;
177 continue;
178 }
179 }
180
181 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
182 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
183 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
184 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
185
186
187 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
188 SIMINT_DBLTYPE PQ[3];
189 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
190 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
191 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
192 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
193 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
194 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
195
196 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
197 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
198 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
199 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
200 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
201 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
202
203 // NOTE: Minus sign!
204 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
205 SIMINT_DBLTYPE aop_PQ[3];
206 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
207 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
208 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
209
210
211 //////////////////////////////////////////////
212 // Fjt function section
213 // Maximum v value: 11
214 //////////////////////////////////////////////
215 // The parameter to the Fjt function
216 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
217
218
219 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
220
221
222 boys_F_split(PRIM_INT__s_s_s_s, F_x, 11);
223 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
224 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
225 for(n = 0; n <= 11; n++)
226 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
227
228 //////////////////////////////////////////////
229 // Primitive integrals: Vertical recurrance
230 //////////////////////////////////////////////
231
232 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
233 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
234 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
235 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
236 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
237 const SIMINT_DBLTYPE vrr_const_6_over_2p = SIMINT_MUL(const_6, one_over_2p);
238 const SIMINT_DBLTYPE vrr_const_7_over_2p = SIMINT_MUL(const_7, one_over_2p);
239 const SIMINT_DBLTYPE vrr_const_8_over_2p = SIMINT_MUL(const_8, one_over_2p);
240 const SIMINT_DBLTYPE vrr_const_9_over_2p = SIMINT_MUL(const_9, one_over_2p);
241 const SIMINT_DBLTYPE vrr_const_10_over_2p = SIMINT_MUL(const_10, one_over_2p);
242
243
244
245 // Forming PRIM_INT__p_s_s_s[11 * 3];
246 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
247 {
248
249 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
250 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]);
251
252 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
253 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]);
254
255 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
256 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]);
257
258 }
259
260
261
262 // Forming PRIM_INT__d_s_s_s[10 * 6];
263 for(n = 0; n < 10; ++n) // loop over orders of auxiliary function
264 {
265
266 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
267 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]);
268 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]);
269
270 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
271 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]);
272 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]);
273
274 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
275 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]);
276 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]);
277
278 }
279
280
281
282 // Forming PRIM_INT__f_s_s_s[9 * 10];
283 for(n = 0; n < 9; ++n) // loop over orders of auxiliary function
284 {
285
286 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
287 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]);
288 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]);
289
290 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
291 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]);
292 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]);
293
294 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
295 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]);
296 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]);
297
298 }
299
300
301 VRR_I_g_s_s_s(
302 PRIM_INT__g_s_s_s,
303 PRIM_INT__f_s_s_s,
304 PRIM_INT__d_s_s_s,
305 P_PA,
306 a_over_p,
307 aop_PQ,
308 one_over_2p,
309 8);
310
311
312 VRR_I_h_s_s_s(
313 PRIM_INT__h_s_s_s,
314 PRIM_INT__g_s_s_s,
315 PRIM_INT__f_s_s_s,
316 P_PA,
317 a_over_p,
318 aop_PQ,
319 one_over_2p,
320 7);
321
322
323 ostei_general_vrr1_I(6, 6,
324 one_over_2p, a_over_p, aop_PQ, P_PA,
325 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
326
327
328 ostei_general_vrr1_I(7, 5,
329 one_over_2p, a_over_p, aop_PQ, P_PA,
330 PRIM_INT__i_s_s_s, PRIM_INT__h_s_s_s, PRIM_INT__k_s_s_s);
331
332
333 ostei_general_vrr1_I(8, 4,
334 one_over_2p, a_over_p, aop_PQ, P_PA,
335 PRIM_INT__k_s_s_s, PRIM_INT__i_s_s_s, PRIM_INT__l_s_s_s);
336
337
338 ostei_general_vrr1_I(9, 3,
339 one_over_2p, a_over_p, aop_PQ, P_PA,
340 PRIM_INT__l_s_s_s, PRIM_INT__k_s_s_s, PRIM_INT__m_s_s_s);
341
342
343 ostei_general_vrr1_I(10, 2,
344 one_over_2p, a_over_p, aop_PQ, P_PA,
345 PRIM_INT__m_s_s_s, PRIM_INT__l_s_s_s, PRIM_INT__n_s_s_s);
346
347
348 ostei_general_vrr1_I(11, 1,
349 one_over_2p, a_over_p, aop_PQ, P_PA,
350 PRIM_INT__n_s_s_s, PRIM_INT__m_s_s_s, PRIM_INT__o_s_s_s);
351
352
353
354
355 ////////////////////////////////////
356 // Accumulate contracted integrals
357 ////////////////////////////////////
358 if(lastoffset == 0)
359 {
360 contract_all(36, PRIM_INT__k_s_s_s, PRIM_PTR_INT__k_s_s_s);
361 contract_all(45, PRIM_INT__l_s_s_s, PRIM_PTR_INT__l_s_s_s);
362 contract_all(55, PRIM_INT__m_s_s_s, PRIM_PTR_INT__m_s_s_s);
363 contract_all(66, PRIM_INT__n_s_s_s, PRIM_PTR_INT__n_s_s_s);
364 contract_all(78, PRIM_INT__o_s_s_s, PRIM_PTR_INT__o_s_s_s);
365 }
366 else
367 {
368 contract(36, shelloffsets, PRIM_INT__k_s_s_s, PRIM_PTR_INT__k_s_s_s);
369 contract(45, shelloffsets, PRIM_INT__l_s_s_s, PRIM_PTR_INT__l_s_s_s);
370 contract(55, shelloffsets, PRIM_INT__m_s_s_s, PRIM_PTR_INT__m_s_s_s);
371 contract(66, shelloffsets, PRIM_INT__n_s_s_s, PRIM_PTR_INT__n_s_s_s);
372 contract(78, shelloffsets, PRIM_INT__o_s_s_s, PRIM_PTR_INT__o_s_s_s);
373 PRIM_PTR_INT__k_s_s_s += lastoffset*36;
374 PRIM_PTR_INT__l_s_s_s += lastoffset*45;
375 PRIM_PTR_INT__m_s_s_s += lastoffset*55;
376 PRIM_PTR_INT__n_s_s_s += lastoffset*66;
377 PRIM_PTR_INT__o_s_s_s += lastoffset*78;
378 }
379
380 } // close loop over j
381 } // close loop over i
382
383 //Advance to the next batch
384 jstart = SIMINT_SIMD_ROUND(jend);
385
386 //////////////////////////////////////////////
387 // Contracted integrals: Horizontal recurrance
388 //////////////////////////////////////////////
389
390
391 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
392
393
394 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
395 {
396
397 // set up HRR pointers
398 double const * restrict HRR_INT__k_s_s_s = INT__k_s_s_s + abcd * 36;
399 double const * restrict HRR_INT__l_s_s_s = INT__l_s_s_s + abcd * 45;
400 double const * restrict HRR_INT__m_s_s_s = INT__m_s_s_s + abcd * 55;
401 double const * restrict HRR_INT__n_s_s_s = INT__n_s_s_s + abcd * 66;
402 double const * restrict HRR_INT__o_s_s_s = INT__o_s_s_s + abcd * 78;
403 double * restrict HRR_INT__k_g_s_s = INT__k_g_s_s + real_abcd * 540;
404
405 // form INT__k_p_s_s
406 ostei_general_hrr_J(7, 1, 0, 0, hAB, HRR_INT__l_s_s_s, HRR_INT__k_s_s_s, HRR_INT__k_p_s_s);
407
408 // form INT__l_p_s_s
409 ostei_general_hrr_J(8, 1, 0, 0, hAB, HRR_INT__m_s_s_s, HRR_INT__l_s_s_s, HRR_INT__l_p_s_s);
410
411 // form INT__m_p_s_s
412 ostei_general_hrr_J(9, 1, 0, 0, hAB, HRR_INT__n_s_s_s, HRR_INT__m_s_s_s, HRR_INT__m_p_s_s);
413
414 // form INT__n_p_s_s
415 ostei_general_hrr_J(10, 1, 0, 0, hAB, HRR_INT__o_s_s_s, HRR_INT__n_s_s_s, HRR_INT__n_p_s_s);
416
417 // form INT__k_d_s_s
418 ostei_general_hrr_J(7, 2, 0, 0, hAB, HRR_INT__l_p_s_s, HRR_INT__k_p_s_s, HRR_INT__k_d_s_s);
419
420 // form INT__l_d_s_s
421 ostei_general_hrr_J(8, 2, 0, 0, hAB, HRR_INT__m_p_s_s, HRR_INT__l_p_s_s, HRR_INT__l_d_s_s);
422
423 // form INT__m_d_s_s
424 ostei_general_hrr_J(9, 2, 0, 0, hAB, HRR_INT__n_p_s_s, HRR_INT__m_p_s_s, HRR_INT__m_d_s_s);
425
426 // form INT__k_f_s_s
427 ostei_general_hrr_J(7, 3, 0, 0, hAB, HRR_INT__l_d_s_s, HRR_INT__k_d_s_s, HRR_INT__k_f_s_s);
428
429 // form INT__l_f_s_s
430 ostei_general_hrr_J(8, 3, 0, 0, hAB, HRR_INT__m_d_s_s, HRR_INT__l_d_s_s, HRR_INT__l_f_s_s);
431
432 // form INT__k_g_s_s
433 ostei_general_hrr_J(7, 4, 0, 0, hAB, HRR_INT__l_f_s_s, HRR_INT__k_f_s_s, HRR_INT__k_g_s_s);
434
435
436 } // close HRR loop
437
438
439 } // close loop cdbatch
440
441 istart = iend;
442 } // close loop over ab
443
444 return P.nshell12_clip * Q.nshell12_clip;
445 }
446
ostei_g_k_s_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__g_k_s_s)447 int ostei_g_k_s_s(struct simint_multi_shellpair const P,
448 struct simint_multi_shellpair const Q,
449 double screen_tol,
450 double * const restrict work,
451 double * const restrict INT__g_k_s_s)
452 {
453 double P_AB[3*P.nshell12];
454 struct simint_multi_shellpair P_tmp = P;
455 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
456 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
457 P_tmp.AB_x = P_AB;
458 P_tmp.AB_y = P_AB + P.nshell12;
459 P_tmp.AB_z = P_AB + 2*P.nshell12;
460
461 for(int i = 0; i < P.nshell12; i++)
462 {
463 P_tmp.AB_x[i] = -P.AB_x[i];
464 P_tmp.AB_y[i] = -P.AB_y[i];
465 P_tmp.AB_z[i] = -P.AB_z[i];
466 }
467
468 int ret = ostei_k_g_s_s(P_tmp, Q, screen_tol, work, INT__g_k_s_s);
469 double buffer[540] SIMINT_ALIGN_ARRAY_DBL;
470
471 for(int q = 0; q < ret; q++)
472 {
473 int idx = 0;
474 for(int a = 0; a < 15; ++a)
475 for(int b = 0; b < 36; ++b)
476 for(int c = 0; c < 1; ++c)
477 for(int d = 0; d < 1; ++d)
478 buffer[idx++] = INT__g_k_s_s[q*540+b*15+a*1+c*1+d];
479
480 memcpy(INT__g_k_s_s+q*540, buffer, 540*sizeof(double));
481 }
482
483 return ret;
484 }
485
486