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_h_h_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__h_h_s_s)8 int ostei_h_h_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__h_h_s_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__h_h_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__h_s_s_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__i_s_s_s = work + (SIMINT_NSHELL_SIMD * 21);
30 double * const INT__k_s_s_s = work + (SIMINT_NSHELL_SIMD * 49);
31 double * const INT__l_s_s_s = work + (SIMINT_NSHELL_SIMD * 85);
32 double * const INT__m_s_s_s = work + (SIMINT_NSHELL_SIMD * 130);
33 double * const INT__n_s_s_s = work + (SIMINT_NSHELL_SIMD * 185);
34 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*251);
35 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_s_s = primwork + 11;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 41;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 95;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 175;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 280;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 406;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_s_s = primwork + 546;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_s_s = primwork + 690;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_s_s = primwork + 825;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_s_s = primwork + 935;
46 double * const hrrwork = (double *)(primwork + 1001);
47 double * const HRR_INT__h_p_s_s = hrrwork + 0;
48 double * const HRR_INT__h_d_s_s = hrrwork + 63;
49 double * const HRR_INT__h_f_s_s = hrrwork + 189;
50 double * const HRR_INT__h_g_s_s = hrrwork + 399;
51 double * const HRR_INT__i_p_s_s = hrrwork + 714;
52 double * const HRR_INT__i_d_s_s = hrrwork + 798;
53 double * const HRR_INT__i_f_s_s = hrrwork + 966;
54 double * const HRR_INT__i_g_s_s = hrrwork + 1246;
55 double * const HRR_INT__k_p_s_s = hrrwork + 1666;
56 double * const HRR_INT__k_d_s_s = hrrwork + 1774;
57 double * const HRR_INT__k_f_s_s = hrrwork + 1990;
58 double * const HRR_INT__l_p_s_s = hrrwork + 2350;
59 double * const HRR_INT__l_d_s_s = hrrwork + 2485;
60 double * const HRR_INT__m_p_s_s = hrrwork + 2755;
61
62
63 // Create constants
64 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
65 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
66 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
67 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
68 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
69 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
70 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
71 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
72 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
73 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
74
75
76 ////////////////////////////////////////
77 // Loop over shells and primitives
78 ////////////////////////////////////////
79
80 real_abcd = 0;
81 istart = 0;
82 for(ab = 0; ab < P.nshell12_clip; ++ab)
83 {
84 const int iend = istart + P.nprim12[ab];
85
86 cd = 0;
87 jstart = 0;
88
89 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
90 {
91 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
92 int jend = jstart;
93 for(i = 0; i < nshellbatch; i++)
94 jend += Q.nprim12[cd+i];
95
96 // Clear the beginning of the workspace (where we are accumulating integrals)
97 memset(work, 0, SIMINT_NSHELL_SIMD * 251 * sizeof(double));
98 abcd = 0;
99
100
101 for(i = istart; i < iend; ++i)
102 {
103 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
104
105 if(check_screen)
106 {
107 // Skip this whole thing if always insignificant
108 if((P.screen[i] * Q.screen_max) < screen_tol)
109 continue;
110 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
111 }
112
113 icd = 0;
114 iprimcd = 0;
115 nprim_icd = Q.nprim12[cd];
116 double * restrict PRIM_PTR_INT__h_s_s_s = INT__h_s_s_s + abcd * 21;
117 double * restrict PRIM_PTR_INT__i_s_s_s = INT__i_s_s_s + abcd * 28;
118 double * restrict PRIM_PTR_INT__k_s_s_s = INT__k_s_s_s + abcd * 36;
119 double * restrict PRIM_PTR_INT__l_s_s_s = INT__l_s_s_s + abcd * 45;
120 double * restrict PRIM_PTR_INT__m_s_s_s = INT__m_s_s_s + abcd * 55;
121 double * restrict PRIM_PTR_INT__n_s_s_s = INT__n_s_s_s + abcd * 66;
122
123
124
125 // Load these one per loop over i
126 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
127 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
128 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
129
130 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
131
132 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
133 {
134 // calculate the shell offsets
135 // these are the offset from the shell pointed to by cd
136 // for each element
137 int shelloffsets[SIMINT_SIMD_LEN] = {0};
138 int lastoffset = 0;
139 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
140
141 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
142 {
143 // Handle if the first element of the vector is a new shell
144 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
145 {
146 nprim_icd += Q.nprim12[cd + (++icd)];
147 PRIM_PTR_INT__h_s_s_s += 21;
148 PRIM_PTR_INT__i_s_s_s += 28;
149 PRIM_PTR_INT__k_s_s_s += 36;
150 PRIM_PTR_INT__l_s_s_s += 45;
151 PRIM_PTR_INT__m_s_s_s += 55;
152 PRIM_PTR_INT__n_s_s_s += 66;
153 }
154 iprimcd++;
155 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
156 {
157 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
158 {
159 shelloffsets[n] = shelloffsets[n-1] + 1;
160 lastoffset++;
161 nprim_icd += Q.nprim12[cd + (++icd)];
162 }
163 else
164 shelloffsets[n] = shelloffsets[n-1];
165 iprimcd++;
166 }
167 }
168 else
169 iprimcd += SIMINT_SIMD_LEN;
170
171 // Do we have to compute this vector (or has it been screened out)?
172 // (not_screened != 0 means we have to do this vector)
173 if(check_screen)
174 {
175 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
176 if(vmax < screen_tol)
177 {
178 PRIM_PTR_INT__h_s_s_s += lastoffset*21;
179 PRIM_PTR_INT__i_s_s_s += lastoffset*28;
180 PRIM_PTR_INT__k_s_s_s += lastoffset*36;
181 PRIM_PTR_INT__l_s_s_s += lastoffset*45;
182 PRIM_PTR_INT__m_s_s_s += lastoffset*55;
183 PRIM_PTR_INT__n_s_s_s += lastoffset*66;
184 continue;
185 }
186 }
187
188 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
189 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
190 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
191 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
192
193
194 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
195 SIMINT_DBLTYPE PQ[3];
196 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
197 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
198 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
199 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
200 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
201 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
202
203 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
204 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
205 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
206 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
207 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
208 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
209
210 // NOTE: Minus sign!
211 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
212 SIMINT_DBLTYPE aop_PQ[3];
213 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
214 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
215 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
216
217
218 //////////////////////////////////////////////
219 // Fjt function section
220 // Maximum v value: 10
221 //////////////////////////////////////////////
222 // The parameter to the Fjt function
223 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
224
225
226 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
227
228
229 boys_F_split(PRIM_INT__s_s_s_s, F_x, 10);
230 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
231 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
232 for(n = 0; n <= 10; n++)
233 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
234
235 //////////////////////////////////////////////
236 // Primitive integrals: Vertical recurrance
237 //////////////////////////////////////////////
238
239 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
240 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
241 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
242 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
243 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
244 const SIMINT_DBLTYPE vrr_const_6_over_2p = SIMINT_MUL(const_6, one_over_2p);
245 const SIMINT_DBLTYPE vrr_const_7_over_2p = SIMINT_MUL(const_7, one_over_2p);
246 const SIMINT_DBLTYPE vrr_const_8_over_2p = SIMINT_MUL(const_8, one_over_2p);
247 const SIMINT_DBLTYPE vrr_const_9_over_2p = SIMINT_MUL(const_9, one_over_2p);
248
249
250
251 // Forming PRIM_INT__p_s_s_s[10 * 3];
252 for(n = 0; n < 10; ++n) // loop over orders of auxiliary function
253 {
254
255 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
256 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]);
257
258 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
259 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]);
260
261 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
262 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]);
263
264 }
265
266
267
268 // Forming PRIM_INT__d_s_s_s[9 * 6];
269 for(n = 0; n < 9; ++n) // loop over orders of auxiliary function
270 {
271
272 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
273 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]);
274 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]);
275
276 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
277 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]);
278 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]);
279
280 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
281 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]);
282 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]);
283
284 }
285
286
287
288 // Forming PRIM_INT__f_s_s_s[8 * 10];
289 for(n = 0; n < 8; ++n) // loop over orders of auxiliary function
290 {
291
292 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
293 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]);
294 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]);
295
296 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
297 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]);
298
299 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
300 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]);
301
302 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
303 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]);
304 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]);
305
306 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
307 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]);
308
309 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
310 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]);
311 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]);
312
313 }
314
315
316 VRR_I_g_s_s_s(
317 PRIM_INT__g_s_s_s,
318 PRIM_INT__f_s_s_s,
319 PRIM_INT__d_s_s_s,
320 P_PA,
321 a_over_p,
322 aop_PQ,
323 one_over_2p,
324 7);
325
326
327 VRR_I_h_s_s_s(
328 PRIM_INT__h_s_s_s,
329 PRIM_INT__g_s_s_s,
330 PRIM_INT__f_s_s_s,
331 P_PA,
332 a_over_p,
333 aop_PQ,
334 one_over_2p,
335 6);
336
337
338 ostei_general_vrr1_I(6, 5,
339 one_over_2p, a_over_p, aop_PQ, P_PA,
340 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
341
342
343 ostei_general_vrr1_I(7, 4,
344 one_over_2p, a_over_p, aop_PQ, P_PA,
345 PRIM_INT__i_s_s_s, PRIM_INT__h_s_s_s, PRIM_INT__k_s_s_s);
346
347
348 ostei_general_vrr1_I(8, 3,
349 one_over_2p, a_over_p, aop_PQ, P_PA,
350 PRIM_INT__k_s_s_s, PRIM_INT__i_s_s_s, PRIM_INT__l_s_s_s);
351
352
353 ostei_general_vrr1_I(9, 2,
354 one_over_2p, a_over_p, aop_PQ, P_PA,
355 PRIM_INT__l_s_s_s, PRIM_INT__k_s_s_s, PRIM_INT__m_s_s_s);
356
357
358 ostei_general_vrr1_I(10, 1,
359 one_over_2p, a_over_p, aop_PQ, P_PA,
360 PRIM_INT__m_s_s_s, PRIM_INT__l_s_s_s, PRIM_INT__n_s_s_s);
361
362
363
364
365 ////////////////////////////////////
366 // Accumulate contracted integrals
367 ////////////////////////////////////
368 if(lastoffset == 0)
369 {
370 contract_all(21, PRIM_INT__h_s_s_s, PRIM_PTR_INT__h_s_s_s);
371 contract_all(28, PRIM_INT__i_s_s_s, PRIM_PTR_INT__i_s_s_s);
372 contract_all(36, PRIM_INT__k_s_s_s, PRIM_PTR_INT__k_s_s_s);
373 contract_all(45, PRIM_INT__l_s_s_s, PRIM_PTR_INT__l_s_s_s);
374 contract_all(55, PRIM_INT__m_s_s_s, PRIM_PTR_INT__m_s_s_s);
375 contract_all(66, PRIM_INT__n_s_s_s, PRIM_PTR_INT__n_s_s_s);
376 }
377 else
378 {
379 contract(21, shelloffsets, PRIM_INT__h_s_s_s, PRIM_PTR_INT__h_s_s_s);
380 contract(28, shelloffsets, PRIM_INT__i_s_s_s, PRIM_PTR_INT__i_s_s_s);
381 contract(36, shelloffsets, PRIM_INT__k_s_s_s, PRIM_PTR_INT__k_s_s_s);
382 contract(45, shelloffsets, PRIM_INT__l_s_s_s, PRIM_PTR_INT__l_s_s_s);
383 contract(55, shelloffsets, PRIM_INT__m_s_s_s, PRIM_PTR_INT__m_s_s_s);
384 contract(66, shelloffsets, PRIM_INT__n_s_s_s, PRIM_PTR_INT__n_s_s_s);
385 PRIM_PTR_INT__h_s_s_s += lastoffset*21;
386 PRIM_PTR_INT__i_s_s_s += lastoffset*28;
387 PRIM_PTR_INT__k_s_s_s += lastoffset*36;
388 PRIM_PTR_INT__l_s_s_s += lastoffset*45;
389 PRIM_PTR_INT__m_s_s_s += lastoffset*55;
390 PRIM_PTR_INT__n_s_s_s += lastoffset*66;
391 }
392
393 } // close loop over j
394 } // close loop over i
395
396 //Advance to the next batch
397 jstart = SIMINT_SIMD_ROUND(jend);
398
399 //////////////////////////////////////////////
400 // Contracted integrals: Horizontal recurrance
401 //////////////////////////////////////////////
402
403
404 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
405
406
407 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
408 {
409
410 // set up HRR pointers
411 double const * restrict HRR_INT__h_s_s_s = INT__h_s_s_s + abcd * 21;
412 double const * restrict HRR_INT__i_s_s_s = INT__i_s_s_s + abcd * 28;
413 double const * restrict HRR_INT__k_s_s_s = INT__k_s_s_s + abcd * 36;
414 double const * restrict HRR_INT__l_s_s_s = INT__l_s_s_s + abcd * 45;
415 double const * restrict HRR_INT__m_s_s_s = INT__m_s_s_s + abcd * 55;
416 double const * restrict HRR_INT__n_s_s_s = INT__n_s_s_s + abcd * 66;
417 double * restrict HRR_INT__h_h_s_s = INT__h_h_s_s + real_abcd * 441;
418
419 // form INT__h_p_s_s
420 ostei_general_hrr_J(5, 1, 0, 0, hAB, HRR_INT__i_s_s_s, HRR_INT__h_s_s_s, HRR_INT__h_p_s_s);
421
422 // form INT__i_p_s_s
423 ostei_general_hrr_J(6, 1, 0, 0, hAB, HRR_INT__k_s_s_s, HRR_INT__i_s_s_s, HRR_INT__i_p_s_s);
424
425 // form INT__k_p_s_s
426 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);
427
428 // form INT__l_p_s_s
429 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);
430
431 // form INT__m_p_s_s
432 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);
433
434 // form INT__h_d_s_s
435 ostei_general_hrr_J(5, 2, 0, 0, hAB, HRR_INT__i_p_s_s, HRR_INT__h_p_s_s, HRR_INT__h_d_s_s);
436
437 // form INT__i_d_s_s
438 ostei_general_hrr_J(6, 2, 0, 0, hAB, HRR_INT__k_p_s_s, HRR_INT__i_p_s_s, HRR_INT__i_d_s_s);
439
440 // form INT__k_d_s_s
441 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);
442
443 // form INT__l_d_s_s
444 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);
445
446 // form INT__h_f_s_s
447 ostei_general_hrr_J(5, 3, 0, 0, hAB, HRR_INT__i_d_s_s, HRR_INT__h_d_s_s, HRR_INT__h_f_s_s);
448
449 // form INT__i_f_s_s
450 ostei_general_hrr_J(6, 3, 0, 0, hAB, HRR_INT__k_d_s_s, HRR_INT__i_d_s_s, HRR_INT__i_f_s_s);
451
452 // form INT__k_f_s_s
453 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);
454
455 // form INT__h_g_s_s
456 ostei_general_hrr_J(5, 4, 0, 0, hAB, HRR_INT__i_f_s_s, HRR_INT__h_f_s_s, HRR_INT__h_g_s_s);
457
458 // form INT__i_g_s_s
459 ostei_general_hrr_J(6, 4, 0, 0, hAB, HRR_INT__k_f_s_s, HRR_INT__i_f_s_s, HRR_INT__i_g_s_s);
460
461 // form INT__h_h_s_s
462 ostei_general_hrr_J(5, 5, 0, 0, hAB, HRR_INT__i_g_s_s, HRR_INT__h_g_s_s, HRR_INT__h_h_s_s);
463
464
465 } // close HRR loop
466
467
468 } // close loop cdbatch
469
470 istart = iend;
471 } // close loop over ab
472
473 return P.nshell12_clip * Q.nshell12_clip;
474 }
475
476