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_i_h_p_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__i_h_p_s)8 int ostei_i_h_p_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__i_h_p_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__i_h_p_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__i_s_p_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__k_s_p_s = work + (SIMINT_NSHELL_SIMD * 84);
30 double * const INT__l_s_p_s = work + (SIMINT_NSHELL_SIMD * 192);
31 double * const INT__m_s_p_s = work + (SIMINT_NSHELL_SIMD * 327);
32 double * const INT__n_s_p_s = work + (SIMINT_NSHELL_SIMD * 492);
33 double * const INT__o_s_p_s = work + (SIMINT_NSHELL_SIMD * 690);
34 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*924);
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 + 13;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 49;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 115;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 215;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 350;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 518;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_p_s = primwork + 714;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_s_s = primwork + 798;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_p_s = primwork + 1014;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_s_s = primwork + 1122;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_p_s = primwork + 1347;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_s_s = primwork + 1482;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_p_s = primwork + 1702;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_s_s = primwork + 1867;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_p_s = primwork + 2065;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__o_s_s_s = primwork + 2263;
52 SIMINT_DBLTYPE * const restrict PRIM_INT__o_s_p_s = primwork + 2419;
53 double * const hrrwork = (double *)(primwork + 2653);
54 double * const HRR_INT__i_p_p_s = hrrwork + 0;
55 double * const HRR_INT__i_d_p_s = hrrwork + 252;
56 double * const HRR_INT__i_f_p_s = hrrwork + 756;
57 double * const HRR_INT__i_g_p_s = hrrwork + 1596;
58 double * const HRR_INT__k_p_p_s = hrrwork + 2856;
59 double * const HRR_INT__k_d_p_s = hrrwork + 3180;
60 double * const HRR_INT__k_f_p_s = hrrwork + 3828;
61 double * const HRR_INT__k_g_p_s = hrrwork + 4908;
62 double * const HRR_INT__l_p_p_s = hrrwork + 6528;
63 double * const HRR_INT__l_d_p_s = hrrwork + 6933;
64 double * const HRR_INT__l_f_p_s = hrrwork + 7743;
65 double * const HRR_INT__m_p_p_s = hrrwork + 9093;
66 double * const HRR_INT__m_d_p_s = hrrwork + 9588;
67 double * const HRR_INT__n_p_p_s = hrrwork + 10578;
68
69
70 // Create constants
71 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
72 const SIMINT_DBLTYPE const_10 = SIMINT_DBLSET1(10);
73 const SIMINT_DBLTYPE const_11 = SIMINT_DBLSET1(11);
74 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
75 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
76 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
77 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
78 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
79 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
80 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
81 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
82 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
83
84
85 ////////////////////////////////////////
86 // Loop over shells and primitives
87 ////////////////////////////////////////
88
89 real_abcd = 0;
90 istart = 0;
91 for(ab = 0; ab < P.nshell12_clip; ++ab)
92 {
93 const int iend = istart + P.nprim12[ab];
94
95 cd = 0;
96 jstart = 0;
97
98 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
99 {
100 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
101 int jend = jstart;
102 for(i = 0; i < nshellbatch; i++)
103 jend += Q.nprim12[cd+i];
104
105 // Clear the beginning of the workspace (where we are accumulating integrals)
106 memset(work, 0, SIMINT_NSHELL_SIMD * 924 * sizeof(double));
107 abcd = 0;
108
109
110 for(i = istart; i < iend; ++i)
111 {
112 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
113
114 if(check_screen)
115 {
116 // Skip this whole thing if always insignificant
117 if((P.screen[i] * Q.screen_max) < screen_tol)
118 continue;
119 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
120 }
121
122 icd = 0;
123 iprimcd = 0;
124 nprim_icd = Q.nprim12[cd];
125 double * restrict PRIM_PTR_INT__i_s_p_s = INT__i_s_p_s + abcd * 84;
126 double * restrict PRIM_PTR_INT__k_s_p_s = INT__k_s_p_s + abcd * 108;
127 double * restrict PRIM_PTR_INT__l_s_p_s = INT__l_s_p_s + abcd * 135;
128 double * restrict PRIM_PTR_INT__m_s_p_s = INT__m_s_p_s + abcd * 165;
129 double * restrict PRIM_PTR_INT__n_s_p_s = INT__n_s_p_s + abcd * 198;
130 double * restrict PRIM_PTR_INT__o_s_p_s = INT__o_s_p_s + abcd * 234;
131
132
133
134 // Load these one per loop over i
135 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
136 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
137 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
138
139 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
140
141 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
142 {
143 // calculate the shell offsets
144 // these are the offset from the shell pointed to by cd
145 // for each element
146 int shelloffsets[SIMINT_SIMD_LEN] = {0};
147 int lastoffset = 0;
148 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
149
150 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
151 {
152 // Handle if the first element of the vector is a new shell
153 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
154 {
155 nprim_icd += Q.nprim12[cd + (++icd)];
156 PRIM_PTR_INT__i_s_p_s += 84;
157 PRIM_PTR_INT__k_s_p_s += 108;
158 PRIM_PTR_INT__l_s_p_s += 135;
159 PRIM_PTR_INT__m_s_p_s += 165;
160 PRIM_PTR_INT__n_s_p_s += 198;
161 PRIM_PTR_INT__o_s_p_s += 234;
162 }
163 iprimcd++;
164 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
165 {
166 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
167 {
168 shelloffsets[n] = shelloffsets[n-1] + 1;
169 lastoffset++;
170 nprim_icd += Q.nprim12[cd + (++icd)];
171 }
172 else
173 shelloffsets[n] = shelloffsets[n-1];
174 iprimcd++;
175 }
176 }
177 else
178 iprimcd += SIMINT_SIMD_LEN;
179
180 // Do we have to compute this vector (or has it been screened out)?
181 // (not_screened != 0 means we have to do this vector)
182 if(check_screen)
183 {
184 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
185 if(vmax < screen_tol)
186 {
187 PRIM_PTR_INT__i_s_p_s += lastoffset*84;
188 PRIM_PTR_INT__k_s_p_s += lastoffset*108;
189 PRIM_PTR_INT__l_s_p_s += lastoffset*135;
190 PRIM_PTR_INT__m_s_p_s += lastoffset*165;
191 PRIM_PTR_INT__n_s_p_s += lastoffset*198;
192 PRIM_PTR_INT__o_s_p_s += lastoffset*234;
193 continue;
194 }
195 }
196
197 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
198 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
199 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
200 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
201
202
203 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
204 SIMINT_DBLTYPE PQ[3];
205 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
206 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
207 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
208 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
209 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
210 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
211
212 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
213 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
214 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
215 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
216 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
217 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
218 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
219
220 // NOTE: Minus sign!
221 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
222 SIMINT_DBLTYPE aop_PQ[3];
223 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
224 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
225 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
226
227 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
228 SIMINT_DBLTYPE aoq_PQ[3];
229 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
230 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
231 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
232 // Put a minus sign here so we don't have to in RR routines
233 a_over_q = SIMINT_NEG(a_over_q);
234
235
236 //////////////////////////////////////////////
237 // Fjt function section
238 // Maximum v value: 12
239 //////////////////////////////////////////////
240 // The parameter to the Fjt function
241 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
242
243
244 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
245
246
247 boys_F_split(PRIM_INT__s_s_s_s, F_x, 12);
248 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
249 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
250 for(n = 0; n <= 12; n++)
251 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
252
253 //////////////////////////////////////////////
254 // Primitive integrals: Vertical recurrance
255 //////////////////////////////////////////////
256
257 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
258 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
259 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
260 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
261 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
262 const SIMINT_DBLTYPE vrr_const_6_over_2p = SIMINT_MUL(const_6, one_over_2p);
263 const SIMINT_DBLTYPE vrr_const_7_over_2p = SIMINT_MUL(const_7, one_over_2p);
264 const SIMINT_DBLTYPE vrr_const_8_over_2p = SIMINT_MUL(const_8, one_over_2p);
265 const SIMINT_DBLTYPE vrr_const_9_over_2p = SIMINT_MUL(const_9, one_over_2p);
266 const SIMINT_DBLTYPE vrr_const_10_over_2p = SIMINT_MUL(const_10, one_over_2p);
267 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
268 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
269 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
270 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
271 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
272 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
273 const SIMINT_DBLTYPE vrr_const_7_over_2pq = SIMINT_MUL(const_7, one_over_2pq);
274 const SIMINT_DBLTYPE vrr_const_8_over_2pq = SIMINT_MUL(const_8, one_over_2pq);
275 const SIMINT_DBLTYPE vrr_const_9_over_2pq = SIMINT_MUL(const_9, one_over_2pq);
276 const SIMINT_DBLTYPE vrr_const_10_over_2pq = SIMINT_MUL(const_10, one_over_2pq);
277 const SIMINT_DBLTYPE vrr_const_11_over_2pq = SIMINT_MUL(const_11, one_over_2pq);
278
279
280
281 // Forming PRIM_INT__p_s_s_s[12 * 3];
282 for(n = 0; n < 12; ++n) // loop over orders of auxiliary function
283 {
284
285 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
286 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]);
287
288 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
289 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]);
290
291 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
292 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]);
293
294 }
295
296
297
298 // Forming PRIM_INT__d_s_s_s[11 * 6];
299 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
300 {
301
302 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
303 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]);
304 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]);
305
306 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
307 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]);
308 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]);
309
310 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
311 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]);
312 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]);
313
314 }
315
316
317
318 // Forming PRIM_INT__f_s_s_s[10 * 10];
319 for(n = 0; n < 10; ++n) // loop over orders of auxiliary function
320 {
321
322 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
323 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]);
324 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]);
325
326 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
327 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]);
328
329 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
330 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]);
331
332 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
333 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]);
334 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]);
335
336 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
337 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]);
338
339 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
340 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]);
341 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]);
342
343 }
344
345
346 VRR_I_g_s_s_s(
347 PRIM_INT__g_s_s_s,
348 PRIM_INT__f_s_s_s,
349 PRIM_INT__d_s_s_s,
350 P_PA,
351 a_over_p,
352 aop_PQ,
353 one_over_2p,
354 9);
355
356
357 VRR_I_h_s_s_s(
358 PRIM_INT__h_s_s_s,
359 PRIM_INT__g_s_s_s,
360 PRIM_INT__f_s_s_s,
361 P_PA,
362 a_over_p,
363 aop_PQ,
364 one_over_2p,
365 8);
366
367
368 ostei_general_vrr1_I(6, 7,
369 one_over_2p, a_over_p, aop_PQ, P_PA,
370 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
371
372
373 ostei_general_vrr_K(6, 0, 1, 0, 1,
374 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
375 PRIM_INT__i_s_s_s, NULL, NULL, PRIM_INT__h_s_s_s, NULL, PRIM_INT__i_s_p_s);
376
377
378 ostei_general_vrr1_I(7, 6,
379 one_over_2p, a_over_p, aop_PQ, P_PA,
380 PRIM_INT__i_s_s_s, PRIM_INT__h_s_s_s, PRIM_INT__k_s_s_s);
381
382
383 ostei_general_vrr_K(7, 0, 1, 0, 1,
384 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
385 PRIM_INT__k_s_s_s, NULL, NULL, PRIM_INT__i_s_s_s, NULL, PRIM_INT__k_s_p_s);
386
387
388 ostei_general_vrr1_I(8, 5,
389 one_over_2p, a_over_p, aop_PQ, P_PA,
390 PRIM_INT__k_s_s_s, PRIM_INT__i_s_s_s, PRIM_INT__l_s_s_s);
391
392
393 ostei_general_vrr_K(8, 0, 1, 0, 1,
394 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
395 PRIM_INT__l_s_s_s, NULL, NULL, PRIM_INT__k_s_s_s, NULL, PRIM_INT__l_s_p_s);
396
397
398 ostei_general_vrr1_I(9, 4,
399 one_over_2p, a_over_p, aop_PQ, P_PA,
400 PRIM_INT__l_s_s_s, PRIM_INT__k_s_s_s, PRIM_INT__m_s_s_s);
401
402
403 ostei_general_vrr_K(9, 0, 1, 0, 1,
404 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
405 PRIM_INT__m_s_s_s, NULL, NULL, PRIM_INT__l_s_s_s, NULL, PRIM_INT__m_s_p_s);
406
407
408 ostei_general_vrr1_I(10, 3,
409 one_over_2p, a_over_p, aop_PQ, P_PA,
410 PRIM_INT__m_s_s_s, PRIM_INT__l_s_s_s, PRIM_INT__n_s_s_s);
411
412
413 ostei_general_vrr_K(10, 0, 1, 0, 1,
414 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
415 PRIM_INT__n_s_s_s, NULL, NULL, PRIM_INT__m_s_s_s, NULL, PRIM_INT__n_s_p_s);
416
417
418 ostei_general_vrr1_I(11, 2,
419 one_over_2p, a_over_p, aop_PQ, P_PA,
420 PRIM_INT__n_s_s_s, PRIM_INT__m_s_s_s, PRIM_INT__o_s_s_s);
421
422
423 ostei_general_vrr_K(11, 0, 1, 0, 1,
424 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
425 PRIM_INT__o_s_s_s, NULL, NULL, PRIM_INT__n_s_s_s, NULL, PRIM_INT__o_s_p_s);
426
427
428
429
430 ////////////////////////////////////
431 // Accumulate contracted integrals
432 ////////////////////////////////////
433 if(lastoffset == 0)
434 {
435 contract_all(84, PRIM_INT__i_s_p_s, PRIM_PTR_INT__i_s_p_s);
436 contract_all(108, PRIM_INT__k_s_p_s, PRIM_PTR_INT__k_s_p_s);
437 contract_all(135, PRIM_INT__l_s_p_s, PRIM_PTR_INT__l_s_p_s);
438 contract_all(165, PRIM_INT__m_s_p_s, PRIM_PTR_INT__m_s_p_s);
439 contract_all(198, PRIM_INT__n_s_p_s, PRIM_PTR_INT__n_s_p_s);
440 contract_all(234, PRIM_INT__o_s_p_s, PRIM_PTR_INT__o_s_p_s);
441 }
442 else
443 {
444 contract(84, shelloffsets, PRIM_INT__i_s_p_s, PRIM_PTR_INT__i_s_p_s);
445 contract(108, shelloffsets, PRIM_INT__k_s_p_s, PRIM_PTR_INT__k_s_p_s);
446 contract(135, shelloffsets, PRIM_INT__l_s_p_s, PRIM_PTR_INT__l_s_p_s);
447 contract(165, shelloffsets, PRIM_INT__m_s_p_s, PRIM_PTR_INT__m_s_p_s);
448 contract(198, shelloffsets, PRIM_INT__n_s_p_s, PRIM_PTR_INT__n_s_p_s);
449 contract(234, shelloffsets, PRIM_INT__o_s_p_s, PRIM_PTR_INT__o_s_p_s);
450 PRIM_PTR_INT__i_s_p_s += lastoffset*84;
451 PRIM_PTR_INT__k_s_p_s += lastoffset*108;
452 PRIM_PTR_INT__l_s_p_s += lastoffset*135;
453 PRIM_PTR_INT__m_s_p_s += lastoffset*165;
454 PRIM_PTR_INT__n_s_p_s += lastoffset*198;
455 PRIM_PTR_INT__o_s_p_s += lastoffset*234;
456 }
457
458 } // close loop over j
459 } // close loop over i
460
461 //Advance to the next batch
462 jstart = SIMINT_SIMD_ROUND(jend);
463
464 //////////////////////////////////////////////
465 // Contracted integrals: Horizontal recurrance
466 //////////////////////////////////////////////
467
468
469 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
470
471
472 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
473 {
474
475 // set up HRR pointers
476 double const * restrict HRR_INT__i_s_p_s = INT__i_s_p_s + abcd * 84;
477 double const * restrict HRR_INT__k_s_p_s = INT__k_s_p_s + abcd * 108;
478 double const * restrict HRR_INT__l_s_p_s = INT__l_s_p_s + abcd * 135;
479 double const * restrict HRR_INT__m_s_p_s = INT__m_s_p_s + abcd * 165;
480 double const * restrict HRR_INT__n_s_p_s = INT__n_s_p_s + abcd * 198;
481 double const * restrict HRR_INT__o_s_p_s = INT__o_s_p_s + abcd * 234;
482 double * restrict HRR_INT__i_h_p_s = INT__i_h_p_s + real_abcd * 1764;
483
484 // form INT__i_p_p_s
485 ostei_general_hrr_J(6, 1, 1, 0, hAB, HRR_INT__k_s_p_s, HRR_INT__i_s_p_s, HRR_INT__i_p_p_s);
486
487 // form INT__k_p_p_s
488 ostei_general_hrr_J(7, 1, 1, 0, hAB, HRR_INT__l_s_p_s, HRR_INT__k_s_p_s, HRR_INT__k_p_p_s);
489
490 // form INT__l_p_p_s
491 ostei_general_hrr_J(8, 1, 1, 0, hAB, HRR_INT__m_s_p_s, HRR_INT__l_s_p_s, HRR_INT__l_p_p_s);
492
493 // form INT__m_p_p_s
494 ostei_general_hrr_J(9, 1, 1, 0, hAB, HRR_INT__n_s_p_s, HRR_INT__m_s_p_s, HRR_INT__m_p_p_s);
495
496 // form INT__n_p_p_s
497 ostei_general_hrr_J(10, 1, 1, 0, hAB, HRR_INT__o_s_p_s, HRR_INT__n_s_p_s, HRR_INT__n_p_p_s);
498
499 // form INT__i_d_p_s
500 ostei_general_hrr_J(6, 2, 1, 0, hAB, HRR_INT__k_p_p_s, HRR_INT__i_p_p_s, HRR_INT__i_d_p_s);
501
502 // form INT__k_d_p_s
503 ostei_general_hrr_J(7, 2, 1, 0, hAB, HRR_INT__l_p_p_s, HRR_INT__k_p_p_s, HRR_INT__k_d_p_s);
504
505 // form INT__l_d_p_s
506 ostei_general_hrr_J(8, 2, 1, 0, hAB, HRR_INT__m_p_p_s, HRR_INT__l_p_p_s, HRR_INT__l_d_p_s);
507
508 // form INT__m_d_p_s
509 ostei_general_hrr_J(9, 2, 1, 0, hAB, HRR_INT__n_p_p_s, HRR_INT__m_p_p_s, HRR_INT__m_d_p_s);
510
511 // form INT__i_f_p_s
512 ostei_general_hrr_J(6, 3, 1, 0, hAB, HRR_INT__k_d_p_s, HRR_INT__i_d_p_s, HRR_INT__i_f_p_s);
513
514 // form INT__k_f_p_s
515 ostei_general_hrr_J(7, 3, 1, 0, hAB, HRR_INT__l_d_p_s, HRR_INT__k_d_p_s, HRR_INT__k_f_p_s);
516
517 // form INT__l_f_p_s
518 ostei_general_hrr_J(8, 3, 1, 0, hAB, HRR_INT__m_d_p_s, HRR_INT__l_d_p_s, HRR_INT__l_f_p_s);
519
520 // form INT__i_g_p_s
521 ostei_general_hrr_J(6, 4, 1, 0, hAB, HRR_INT__k_f_p_s, HRR_INT__i_f_p_s, HRR_INT__i_g_p_s);
522
523 // form INT__k_g_p_s
524 ostei_general_hrr_J(7, 4, 1, 0, hAB, HRR_INT__l_f_p_s, HRR_INT__k_f_p_s, HRR_INT__k_g_p_s);
525
526 // form INT__i_h_p_s
527 ostei_general_hrr_J(6, 5, 1, 0, hAB, HRR_INT__k_g_p_s, HRR_INT__i_g_p_s, HRR_INT__i_h_p_s);
528
529
530 } // close HRR loop
531
532
533 } // close loop cdbatch
534
535 istart = iend;
536 } // close loop over ab
537
538 return P.nshell12_clip * Q.nshell12_clip;
539 }
540
ostei_h_i_p_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_i_p_s)541 int ostei_h_i_p_s(struct simint_multi_shellpair const P,
542 struct simint_multi_shellpair const Q,
543 double screen_tol,
544 double * const restrict work,
545 double * const restrict INT__h_i_p_s)
546 {
547 double P_AB[3*P.nshell12];
548 struct simint_multi_shellpair P_tmp = P;
549 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
550 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
551 P_tmp.AB_x = P_AB;
552 P_tmp.AB_y = P_AB + P.nshell12;
553 P_tmp.AB_z = P_AB + 2*P.nshell12;
554
555 for(int i = 0; i < P.nshell12; i++)
556 {
557 P_tmp.AB_x[i] = -P.AB_x[i];
558 P_tmp.AB_y[i] = -P.AB_y[i];
559 P_tmp.AB_z[i] = -P.AB_z[i];
560 }
561
562 int ret = ostei_i_h_p_s(P_tmp, Q, screen_tol, work, INT__h_i_p_s);
563 double buffer[1764] SIMINT_ALIGN_ARRAY_DBL;
564
565 for(int q = 0; q < ret; q++)
566 {
567 int idx = 0;
568 for(int a = 0; a < 21; ++a)
569 for(int b = 0; b < 28; ++b)
570 for(int c = 0; c < 3; ++c)
571 for(int d = 0; d < 1; ++d)
572 buffer[idx++] = INT__h_i_p_s[q*1764+b*63+a*3+c*1+d];
573
574 memcpy(INT__h_i_p_s+q*1764, buffer, 1764*sizeof(double));
575 }
576
577 return ret;
578 }
579
ostei_i_h_s_p(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__i_h_s_p)580 int ostei_i_h_s_p(struct simint_multi_shellpair const P,
581 struct simint_multi_shellpair const Q,
582 double screen_tol,
583 double * const restrict work,
584 double * const restrict INT__i_h_s_p)
585 {
586 double Q_AB[3*Q.nshell12];
587 struct simint_multi_shellpair Q_tmp = Q;
588 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
589 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
590 Q_tmp.AB_x = Q_AB;
591 Q_tmp.AB_y = Q_AB + Q.nshell12;
592 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
593
594 for(int i = 0; i < Q.nshell12; i++)
595 {
596 Q_tmp.AB_x[i] = -Q.AB_x[i];
597 Q_tmp.AB_y[i] = -Q.AB_y[i];
598 Q_tmp.AB_z[i] = -Q.AB_z[i];
599 }
600
601 int ret = ostei_i_h_p_s(P, Q_tmp, screen_tol, work, INT__i_h_s_p);
602 double buffer[1764] SIMINT_ALIGN_ARRAY_DBL;
603
604 for(int q = 0; q < ret; q++)
605 {
606 int idx = 0;
607 for(int a = 0; a < 28; ++a)
608 for(int b = 0; b < 21; ++b)
609 for(int c = 0; c < 1; ++c)
610 for(int d = 0; d < 3; ++d)
611 buffer[idx++] = INT__i_h_s_p[q*1764+a*63+b*3+d*1+c];
612
613 memcpy(INT__i_h_s_p+q*1764, buffer, 1764*sizeof(double));
614 }
615
616 return ret;
617 }
618
ostei_h_i_s_p(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__h_i_s_p)619 int ostei_h_i_s_p(struct simint_multi_shellpair const P,
620 struct simint_multi_shellpair const Q,
621 double screen_tol,
622 double * const restrict work,
623 double * const restrict INT__h_i_s_p)
624 {
625 double P_AB[3*P.nshell12];
626 struct simint_multi_shellpair P_tmp = P;
627 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
628 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
629 P_tmp.AB_x = P_AB;
630 P_tmp.AB_y = P_AB + P.nshell12;
631 P_tmp.AB_z = P_AB + 2*P.nshell12;
632
633 for(int i = 0; i < P.nshell12; i++)
634 {
635 P_tmp.AB_x[i] = -P.AB_x[i];
636 P_tmp.AB_y[i] = -P.AB_y[i];
637 P_tmp.AB_z[i] = -P.AB_z[i];
638 }
639
640 double Q_AB[3*Q.nshell12];
641 struct simint_multi_shellpair Q_tmp = Q;
642 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
643 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
644 Q_tmp.AB_x = Q_AB;
645 Q_tmp.AB_y = Q_AB + Q.nshell12;
646 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
647
648 for(int i = 0; i < Q.nshell12; i++)
649 {
650 Q_tmp.AB_x[i] = -Q.AB_x[i];
651 Q_tmp.AB_y[i] = -Q.AB_y[i];
652 Q_tmp.AB_z[i] = -Q.AB_z[i];
653 }
654
655 int ret = ostei_i_h_p_s(P_tmp, Q_tmp, screen_tol, work, INT__h_i_s_p);
656 double buffer[1764] SIMINT_ALIGN_ARRAY_DBL;
657
658 for(int q = 0; q < ret; q++)
659 {
660 int idx = 0;
661 for(int a = 0; a < 21; ++a)
662 for(int b = 0; b < 28; ++b)
663 for(int c = 0; c < 1; ++c)
664 for(int d = 0; d < 3; ++d)
665 buffer[idx++] = INT__h_i_s_p[q*1764+b*63+a*3+d*1+c];
666
667 memcpy(INT__h_i_s_p+q*1764, buffer, 1764*sizeof(double));
668 }
669
670 return ret;
671 }
672
673