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_s_f_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_s_f_s)8 int ostei_h_s_f_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_s_f_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__h_s_f_s);
17 memset(INT__h_s_f_s, 0, P.nshell12_clip * Q.nshell12_clip * 210 * sizeof(double));
18
19 int ab, cd, abcd;
20 int istart, jstart;
21 int iprimcd, nprim_icd, icd;
22 const int check_screen = (screen_tol > 0.0);
23 int i, j;
24 int n;
25 int not_screened;
26
27 // partition workspace
28 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*0);
29 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
30 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_s_s = primwork + 9;
31 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 33;
32 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 75;
33 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_p_s = primwork + 135;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 225;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_p_s = primwork + 300;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_d_s = primwork + 435;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 615;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_p_s = primwork + 699;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_d_s = primwork + 888;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_f_s = primwork + 1140;
41 double * const hrrwork = (double *)(primwork + 1350);
42
43
44 // Create constants
45 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
46 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
47 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
48 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
49 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
50 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
51
52
53 ////////////////////////////////////////
54 // Loop over shells and primitives
55 ////////////////////////////////////////
56
57 abcd = 0;
58 istart = 0;
59 for(ab = 0; ab < P.nshell12_clip; ++ab)
60 {
61 const int iend = istart + P.nprim12[ab];
62
63 cd = 0;
64 jstart = 0;
65
66 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
67 {
68 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
69 int jend = jstart;
70 for(i = 0; i < nshellbatch; i++)
71 jend += Q.nprim12[cd+i];
72
73
74 for(i = istart; i < iend; ++i)
75 {
76 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
77
78 if(check_screen)
79 {
80 // Skip this whole thing if always insignificant
81 if((P.screen[i] * Q.screen_max) < screen_tol)
82 continue;
83 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
84 }
85
86 icd = 0;
87 iprimcd = 0;
88 nprim_icd = Q.nprim12[cd];
89 double * restrict PRIM_PTR_INT__h_s_f_s = INT__h_s_f_s + abcd * 210;
90
91
92
93 // Load these one per loop over i
94 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
95 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
96 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
97
98 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
99
100 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
101 {
102 // calculate the shell offsets
103 // these are the offset from the shell pointed to by cd
104 // for each element
105 int shelloffsets[SIMINT_SIMD_LEN] = {0};
106 int lastoffset = 0;
107 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
108
109 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
110 {
111 // Handle if the first element of the vector is a new shell
112 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
113 {
114 nprim_icd += Q.nprim12[cd + (++icd)];
115 PRIM_PTR_INT__h_s_f_s += 210;
116 }
117 iprimcd++;
118 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
119 {
120 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
121 {
122 shelloffsets[n] = shelloffsets[n-1] + 1;
123 lastoffset++;
124 nprim_icd += Q.nprim12[cd + (++icd)];
125 }
126 else
127 shelloffsets[n] = shelloffsets[n-1];
128 iprimcd++;
129 }
130 }
131 else
132 iprimcd += SIMINT_SIMD_LEN;
133
134 // Do we have to compute this vector (or has it been screened out)?
135 // (not_screened != 0 means we have to do this vector)
136 if(check_screen)
137 {
138 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
139 if(vmax < screen_tol)
140 {
141 PRIM_PTR_INT__h_s_f_s += lastoffset*210;
142 continue;
143 }
144 }
145
146 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
147 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
148 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
149 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
150
151
152 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
153 SIMINT_DBLTYPE PQ[3];
154 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
155 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
156 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
157 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
158 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
159 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
160
161 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
162 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
163 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
164 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
165 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
166 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
167 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
168
169 // NOTE: Minus sign!
170 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
171 SIMINT_DBLTYPE aop_PQ[3];
172 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
173 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
174 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
175
176 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
177 SIMINT_DBLTYPE aoq_PQ[3];
178 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
179 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
180 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
181 // Put a minus sign here so we don't have to in RR routines
182 a_over_q = SIMINT_NEG(a_over_q);
183
184
185 //////////////////////////////////////////////
186 // Fjt function section
187 // Maximum v value: 8
188 //////////////////////////////////////////////
189 // The parameter to the Fjt function
190 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
191
192
193 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
194
195
196 boys_F_split(PRIM_INT__s_s_s_s, F_x, 8);
197 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
198 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
199 for(n = 0; n <= 8; n++)
200 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
201
202 //////////////////////////////////////////////
203 // Primitive integrals: Vertical recurrance
204 //////////////////////////////////////////////
205
206 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
207 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
208 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
209 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
210 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
211 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
212 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
213 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
214 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
215 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
216 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
217
218
219
220 // Forming PRIM_INT__p_s_s_s[8 * 3];
221 for(n = 0; n < 8; ++n) // loop over orders of auxiliary function
222 {
223
224 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
225 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]);
226
227 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
228 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]);
229
230 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
231 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]);
232
233 }
234
235
236
237 // Forming PRIM_INT__d_s_s_s[7 * 6];
238 for(n = 0; n < 7; ++n) // loop over orders of auxiliary function
239 {
240
241 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
242 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]);
243 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]);
244
245 PRIM_INT__d_s_s_s[n * 6 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 0]);
246 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]);
247
248 PRIM_INT__d_s_s_s[n * 6 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 0]);
249 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]);
250
251 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
252 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]);
253 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]);
254
255 PRIM_INT__d_s_s_s[n * 6 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 1]);
256 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]);
257
258 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
259 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]);
260 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]);
261
262 }
263
264
265
266 // Forming PRIM_INT__f_s_s_s[6 * 10];
267 for(n = 0; n < 6; ++n) // loop over orders of auxiliary function
268 {
269
270 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
271 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]);
272 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]);
273
274 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
275 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]);
276
277 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
278 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]);
279
280 PRIM_INT__f_s_s_s[n * 10 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
281 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]);
282
283 PRIM_INT__f_s_s_s[n * 10 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 1]);
284 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]);
285
286 PRIM_INT__f_s_s_s[n * 10 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
287 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]);
288
289 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
290 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]);
291 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]);
292
293 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
294 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]);
295
296 PRIM_INT__f_s_s_s[n * 10 + 8] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 5]);
297 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]);
298
299 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
300 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]);
301 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]);
302
303 }
304
305
306 VRR_I_g_s_s_s(
307 PRIM_INT__g_s_s_s,
308 PRIM_INT__f_s_s_s,
309 PRIM_INT__d_s_s_s,
310 P_PA,
311 a_over_p,
312 aop_PQ,
313 one_over_2p,
314 5);
315
316
317 VRR_I_h_s_s_s(
318 PRIM_INT__h_s_s_s,
319 PRIM_INT__g_s_s_s,
320 PRIM_INT__f_s_s_s,
321 P_PA,
322 a_over_p,
323 aop_PQ,
324 one_over_2p,
325 4);
326
327
328 ostei_general_vrr_K(5, 0, 1, 0, 3,
329 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
330 PRIM_INT__h_s_s_s, NULL, NULL, PRIM_INT__g_s_s_s, NULL, PRIM_INT__h_s_p_s);
331
332
333 VRR_K_g_s_p_s(
334 PRIM_INT__g_s_p_s,
335 PRIM_INT__g_s_s_s,
336 PRIM_INT__f_s_s_s,
337 Q_PA,
338 aoq_PQ,
339 one_over_2pq,
340 3);
341
342
343 ostei_general_vrr_K(5, 0, 2, 0, 2,
344 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
345 PRIM_INT__h_s_p_s, PRIM_INT__h_s_s_s, NULL, PRIM_INT__g_s_p_s, NULL, PRIM_INT__h_s_d_s);
346
347
348 VRR_K_f_s_p_s(
349 PRIM_INT__f_s_p_s,
350 PRIM_INT__f_s_s_s,
351 PRIM_INT__d_s_s_s,
352 Q_PA,
353 aoq_PQ,
354 one_over_2pq,
355 3);
356
357
358 ostei_general_vrr_K(4, 0, 2, 0, 2,
359 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
360 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);
361
362
363 ostei_general_vrr_K(5, 0, 3, 0, 1,
364 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
365 PRIM_INT__h_s_d_s, PRIM_INT__h_s_p_s, NULL, PRIM_INT__g_s_d_s, NULL, PRIM_INT__h_s_f_s);
366
367
368
369
370 ////////////////////////////////////
371 // Accumulate contracted integrals
372 ////////////////////////////////////
373 if(lastoffset == 0)
374 {
375 contract_all(210, PRIM_INT__h_s_f_s, PRIM_PTR_INT__h_s_f_s);
376 }
377 else
378 {
379 contract(210, shelloffsets, PRIM_INT__h_s_f_s, PRIM_PTR_INT__h_s_f_s);
380 PRIM_PTR_INT__h_s_f_s += lastoffset*210;
381 }
382
383 } // close loop over j
384 } // close loop over i
385
386 //Advance to the next batch
387 jstart = SIMINT_SIMD_ROUND(jend);
388 abcd += nshellbatch;
389
390 } // close loop cdbatch
391
392 istart = iend;
393 } // close loop over ab
394
395 return P.nshell12_clip * Q.nshell12_clip;
396 }
397
ostei_s_h_f_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__s_h_f_s)398 int ostei_s_h_f_s(struct simint_multi_shellpair const P,
399 struct simint_multi_shellpair const Q,
400 double screen_tol,
401 double * const restrict work,
402 double * const restrict INT__s_h_f_s)
403 {
404 double P_AB[3*P.nshell12];
405 struct simint_multi_shellpair P_tmp = P;
406 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
407 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
408 P_tmp.AB_x = P_AB;
409 P_tmp.AB_y = P_AB + P.nshell12;
410 P_tmp.AB_z = P_AB + 2*P.nshell12;
411
412 for(int i = 0; i < P.nshell12; i++)
413 {
414 P_tmp.AB_x[i] = -P.AB_x[i];
415 P_tmp.AB_y[i] = -P.AB_y[i];
416 P_tmp.AB_z[i] = -P.AB_z[i];
417 }
418
419 int ret = ostei_h_s_f_s(P_tmp, Q, screen_tol, work, INT__s_h_f_s);
420
421 return ret;
422 }
423
ostei_h_s_s_f(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__h_s_s_f)424 int ostei_h_s_s_f(struct simint_multi_shellpair const P,
425 struct simint_multi_shellpair const Q,
426 double screen_tol,
427 double * const restrict work,
428 double * const restrict INT__h_s_s_f)
429 {
430 double Q_AB[3*Q.nshell12];
431 struct simint_multi_shellpair Q_tmp = Q;
432 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
433 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
434 Q_tmp.AB_x = Q_AB;
435 Q_tmp.AB_y = Q_AB + Q.nshell12;
436 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
437
438 for(int i = 0; i < Q.nshell12; i++)
439 {
440 Q_tmp.AB_x[i] = -Q.AB_x[i];
441 Q_tmp.AB_y[i] = -Q.AB_y[i];
442 Q_tmp.AB_z[i] = -Q.AB_z[i];
443 }
444
445 int ret = ostei_h_s_f_s(P, Q_tmp, screen_tol, work, INT__h_s_s_f);
446
447 return ret;
448 }
449
ostei_s_h_s_f(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__s_h_s_f)450 int ostei_s_h_s_f(struct simint_multi_shellpair const P,
451 struct simint_multi_shellpair const Q,
452 double screen_tol,
453 double * const restrict work,
454 double * const restrict INT__s_h_s_f)
455 {
456 double P_AB[3*P.nshell12];
457 struct simint_multi_shellpair P_tmp = P;
458 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
459 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
460 P_tmp.AB_x = P_AB;
461 P_tmp.AB_y = P_AB + P.nshell12;
462 P_tmp.AB_z = P_AB + 2*P.nshell12;
463
464 for(int i = 0; i < P.nshell12; i++)
465 {
466 P_tmp.AB_x[i] = -P.AB_x[i];
467 P_tmp.AB_y[i] = -P.AB_y[i];
468 P_tmp.AB_z[i] = -P.AB_z[i];
469 }
470
471 double Q_AB[3*Q.nshell12];
472 struct simint_multi_shellpair Q_tmp = Q;
473 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
474 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
475 Q_tmp.AB_x = Q_AB;
476 Q_tmp.AB_y = Q_AB + Q.nshell12;
477 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
478
479 for(int i = 0; i < Q.nshell12; i++)
480 {
481 Q_tmp.AB_x[i] = -Q.AB_x[i];
482 Q_tmp.AB_y[i] = -Q.AB_y[i];
483 Q_tmp.AB_z[i] = -Q.AB_z[i];
484 }
485
486 int ret = ostei_h_s_f_s(P_tmp, Q_tmp, screen_tol, work, INT__s_h_s_f);
487
488 return ret;
489 }
490
491