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_g_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__g_s_f_s)8 int ostei_g_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__g_s_f_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__g_s_f_s);
17 memset(INT__g_s_f_s, 0, P.nshell12_clip * Q.nshell12_clip * 150 * 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 + 8;
31 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 29;
32 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_p_s = primwork + 65;
33 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 119;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_p_s = primwork + 169;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_d_s = primwork + 259;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 379;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_p_s = primwork + 439;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_d_s = primwork + 574;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_f_s = primwork + 754;
40 double * const hrrwork = (double *)(primwork + 904);
41
42
43 // Create constants
44 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
45 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
46 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
47 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
48 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
49
50
51 ////////////////////////////////////////
52 // Loop over shells and primitives
53 ////////////////////////////////////////
54
55 abcd = 0;
56 istart = 0;
57 for(ab = 0; ab < P.nshell12_clip; ++ab)
58 {
59 const int iend = istart + P.nprim12[ab];
60
61 cd = 0;
62 jstart = 0;
63
64 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
65 {
66 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
67 int jend = jstart;
68 for(i = 0; i < nshellbatch; i++)
69 jend += Q.nprim12[cd+i];
70
71
72 for(i = istart; i < iend; ++i)
73 {
74 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
75
76 if(check_screen)
77 {
78 // Skip this whole thing if always insignificant
79 if((P.screen[i] * Q.screen_max) < screen_tol)
80 continue;
81 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
82 }
83
84 icd = 0;
85 iprimcd = 0;
86 nprim_icd = Q.nprim12[cd];
87 double * restrict PRIM_PTR_INT__g_s_f_s = INT__g_s_f_s + abcd * 150;
88
89
90
91 // Load these one per loop over i
92 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
93 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
94 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
95
96 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
97
98 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
99 {
100 // calculate the shell offsets
101 // these are the offset from the shell pointed to by cd
102 // for each element
103 int shelloffsets[SIMINT_SIMD_LEN] = {0};
104 int lastoffset = 0;
105 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
106
107 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
108 {
109 // Handle if the first element of the vector is a new shell
110 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
111 {
112 nprim_icd += Q.nprim12[cd + (++icd)];
113 PRIM_PTR_INT__g_s_f_s += 150;
114 }
115 iprimcd++;
116 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
117 {
118 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
119 {
120 shelloffsets[n] = shelloffsets[n-1] + 1;
121 lastoffset++;
122 nprim_icd += Q.nprim12[cd + (++icd)];
123 }
124 else
125 shelloffsets[n] = shelloffsets[n-1];
126 iprimcd++;
127 }
128 }
129 else
130 iprimcd += SIMINT_SIMD_LEN;
131
132 // Do we have to compute this vector (or has it been screened out)?
133 // (not_screened != 0 means we have to do this vector)
134 if(check_screen)
135 {
136 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
137 if(vmax < screen_tol)
138 {
139 PRIM_PTR_INT__g_s_f_s += lastoffset*150;
140 continue;
141 }
142 }
143
144 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
145 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
146 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
147 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
148
149
150 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
151 SIMINT_DBLTYPE PQ[3];
152 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
153 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
154 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
155 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
156 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
157 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
158
159 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
160 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
161 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
162 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
163 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
164 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
165 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
166
167 // NOTE: Minus sign!
168 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
169 SIMINT_DBLTYPE aop_PQ[3];
170 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
171 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
172 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
173
174 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
175 SIMINT_DBLTYPE aoq_PQ[3];
176 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
177 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
178 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
179 // Put a minus sign here so we don't have to in RR routines
180 a_over_q = SIMINT_NEG(a_over_q);
181
182
183 //////////////////////////////////////////////
184 // Fjt function section
185 // Maximum v value: 7
186 //////////////////////////////////////////////
187 // The parameter to the Fjt function
188 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
189
190
191 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
192
193
194 boys_F_split(PRIM_INT__s_s_s_s, F_x, 7);
195 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
196 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
197 for(n = 0; n <= 7; n++)
198 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
199
200 //////////////////////////////////////////////
201 // Primitive integrals: Vertical recurrance
202 //////////////////////////////////////////////
203
204 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
205 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
206 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
207 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
208 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
209 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
210 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
211 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
212 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
213
214
215
216 // Forming PRIM_INT__p_s_s_s[7 * 3];
217 for(n = 0; n < 7; ++n) // loop over orders of auxiliary function
218 {
219
220 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
221 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]);
222
223 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
224 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]);
225
226 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
227 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]);
228
229 }
230
231
232
233 // Forming PRIM_INT__d_s_s_s[6 * 6];
234 for(n = 0; n < 6; ++n) // loop over orders of auxiliary function
235 {
236
237 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
238 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]);
239 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]);
240
241 PRIM_INT__d_s_s_s[n * 6 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 0]);
242 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]);
243
244 PRIM_INT__d_s_s_s[n * 6 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 0]);
245 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]);
246
247 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
248 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]);
249 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]);
250
251 PRIM_INT__d_s_s_s[n * 6 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 1]);
252 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]);
253
254 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
255 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]);
256 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]);
257
258 }
259
260
261
262 // Forming PRIM_INT__f_s_s_s[5 * 10];
263 for(n = 0; n < 5; ++n) // loop over orders of auxiliary function
264 {
265
266 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
267 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]);
268 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]);
269
270 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
271 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]);
272
273 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
274 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]);
275
276 PRIM_INT__f_s_s_s[n * 10 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
277 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]);
278
279 PRIM_INT__f_s_s_s[n * 10 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 1]);
280 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]);
281
282 PRIM_INT__f_s_s_s[n * 10 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
283 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]);
284
285 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
286 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]);
287 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]);
288
289 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
290 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]);
291
292 PRIM_INT__f_s_s_s[n * 10 + 8] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 5]);
293 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]);
294
295 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
296 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]);
297 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]);
298
299 }
300
301
302 VRR_I_g_s_s_s(
303 PRIM_INT__g_s_s_s,
304 PRIM_INT__f_s_s_s,
305 PRIM_INT__d_s_s_s,
306 P_PA,
307 a_over_p,
308 aop_PQ,
309 one_over_2p,
310 4);
311
312
313 VRR_K_g_s_p_s(
314 PRIM_INT__g_s_p_s,
315 PRIM_INT__g_s_s_s,
316 PRIM_INT__f_s_s_s,
317 Q_PA,
318 aoq_PQ,
319 one_over_2pq,
320 3);
321
322
323 VRR_K_f_s_p_s(
324 PRIM_INT__f_s_p_s,
325 PRIM_INT__f_s_s_s,
326 PRIM_INT__d_s_s_s,
327 Q_PA,
328 aoq_PQ,
329 one_over_2pq,
330 3);
331
332
333 ostei_general_vrr_K(4, 0, 2, 0, 2,
334 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
335 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);
336
337
338
339 // Forming PRIM_INT__d_s_p_s[3 * 18];
340 for(n = 0; n < 3; ++n) // loop over orders of auxiliary function
341 {
342
343 PRIM_INT__d_s_p_s[n * 18 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
344 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]);
345 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]);
346
347 PRIM_INT__d_s_p_s[n * 18 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
348 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]);
349
350 PRIM_INT__d_s_p_s[n * 18 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
351 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]);
352
353 PRIM_INT__d_s_p_s[n * 18 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 1]);
354 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]);
355 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]);
356
357 PRIM_INT__d_s_p_s[n * 18 + 4] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 1]);
358 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]);
359 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]);
360
361 PRIM_INT__d_s_p_s[n * 18 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 1]);
362 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]);
363
364 PRIM_INT__d_s_p_s[n * 18 + 6] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 2]);
365 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]);
366 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]);
367
368 PRIM_INT__d_s_p_s[n * 18 + 7] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 2]);
369 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]);
370
371 PRIM_INT__d_s_p_s[n * 18 + 8] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 2]);
372 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]);
373 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]);
374
375 PRIM_INT__d_s_p_s[n * 18 + 9] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
376 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]);
377
378 PRIM_INT__d_s_p_s[n * 18 + 10] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
379 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]);
380 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]);
381
382 PRIM_INT__d_s_p_s[n * 18 + 11] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
383 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]);
384
385 PRIM_INT__d_s_p_s[n * 18 + 12] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 4]);
386 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]);
387
388 PRIM_INT__d_s_p_s[n * 18 + 13] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 4]);
389 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]);
390 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]);
391
392 PRIM_INT__d_s_p_s[n * 18 + 14] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 4]);
393 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]);
394 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]);
395
396 PRIM_INT__d_s_p_s[n * 18 + 15] = SIMINT_MUL(Q_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
397 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]);
398
399 PRIM_INT__d_s_p_s[n * 18 + 16] = SIMINT_MUL(Q_PA[1], PRIM_INT__d_s_s_s[n * 6 + 5]);
400 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]);
401
402 PRIM_INT__d_s_p_s[n * 18 + 17] = SIMINT_MUL(Q_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
403 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]);
404 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]);
405
406 }
407
408
409 VRR_K_f_s_d_s(
410 PRIM_INT__f_s_d_s,
411 PRIM_INT__f_s_p_s,
412 PRIM_INT__f_s_s_s,
413 PRIM_INT__d_s_p_s,
414 Q_PA,
415 a_over_q,
416 aoq_PQ,
417 one_over_2pq,
418 one_over_2q,
419 2);
420
421
422 ostei_general_vrr_K(4, 0, 3, 0, 1,
423 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
424 PRIM_INT__g_s_d_s, PRIM_INT__g_s_p_s, NULL, PRIM_INT__f_s_d_s, NULL, PRIM_INT__g_s_f_s);
425
426
427
428
429 ////////////////////////////////////
430 // Accumulate contracted integrals
431 ////////////////////////////////////
432 if(lastoffset == 0)
433 {
434 contract_all(150, PRIM_INT__g_s_f_s, PRIM_PTR_INT__g_s_f_s);
435 }
436 else
437 {
438 contract(150, shelloffsets, PRIM_INT__g_s_f_s, PRIM_PTR_INT__g_s_f_s);
439 PRIM_PTR_INT__g_s_f_s += lastoffset*150;
440 }
441
442 } // close loop over j
443 } // close loop over i
444
445 //Advance to the next batch
446 jstart = SIMINT_SIMD_ROUND(jend);
447 abcd += nshellbatch;
448
449 } // close loop cdbatch
450
451 istart = iend;
452 } // close loop over ab
453
454 return P.nshell12_clip * Q.nshell12_clip;
455 }
456
ostei_s_g_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_g_f_s)457 int ostei_s_g_f_s(struct simint_multi_shellpair const P,
458 struct simint_multi_shellpair const Q,
459 double screen_tol,
460 double * const restrict work,
461 double * const restrict INT__s_g_f_s)
462 {
463 double P_AB[3*P.nshell12];
464 struct simint_multi_shellpair P_tmp = P;
465 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
466 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
467 P_tmp.AB_x = P_AB;
468 P_tmp.AB_y = P_AB + P.nshell12;
469 P_tmp.AB_z = P_AB + 2*P.nshell12;
470
471 for(int i = 0; i < P.nshell12; i++)
472 {
473 P_tmp.AB_x[i] = -P.AB_x[i];
474 P_tmp.AB_y[i] = -P.AB_y[i];
475 P_tmp.AB_z[i] = -P.AB_z[i];
476 }
477
478 int ret = ostei_g_s_f_s(P_tmp, Q, screen_tol, work, INT__s_g_f_s);
479
480 return ret;
481 }
482
ostei_g_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__g_s_s_f)483 int ostei_g_s_s_f(struct simint_multi_shellpair const P,
484 struct simint_multi_shellpair const Q,
485 double screen_tol,
486 double * const restrict work,
487 double * const restrict INT__g_s_s_f)
488 {
489 double Q_AB[3*Q.nshell12];
490 struct simint_multi_shellpair Q_tmp = Q;
491 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
492 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
493 Q_tmp.AB_x = Q_AB;
494 Q_tmp.AB_y = Q_AB + Q.nshell12;
495 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
496
497 for(int i = 0; i < Q.nshell12; i++)
498 {
499 Q_tmp.AB_x[i] = -Q.AB_x[i];
500 Q_tmp.AB_y[i] = -Q.AB_y[i];
501 Q_tmp.AB_z[i] = -Q.AB_z[i];
502 }
503
504 int ret = ostei_g_s_f_s(P, Q_tmp, screen_tol, work, INT__g_s_s_f);
505
506 return ret;
507 }
508
ostei_s_g_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_g_s_f)509 int ostei_s_g_s_f(struct simint_multi_shellpair const P,
510 struct simint_multi_shellpair const Q,
511 double screen_tol,
512 double * const restrict work,
513 double * const restrict INT__s_g_s_f)
514 {
515 double P_AB[3*P.nshell12];
516 struct simint_multi_shellpair P_tmp = P;
517 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
518 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
519 P_tmp.AB_x = P_AB;
520 P_tmp.AB_y = P_AB + P.nshell12;
521 P_tmp.AB_z = P_AB + 2*P.nshell12;
522
523 for(int i = 0; i < P.nshell12; i++)
524 {
525 P_tmp.AB_x[i] = -P.AB_x[i];
526 P_tmp.AB_y[i] = -P.AB_y[i];
527 P_tmp.AB_z[i] = -P.AB_z[i];
528 }
529
530 double Q_AB[3*Q.nshell12];
531 struct simint_multi_shellpair Q_tmp = Q;
532 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
533 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
534 Q_tmp.AB_x = Q_AB;
535 Q_tmp.AB_y = Q_AB + Q.nshell12;
536 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
537
538 for(int i = 0; i < Q.nshell12; i++)
539 {
540 Q_tmp.AB_x[i] = -Q.AB_x[i];
541 Q_tmp.AB_y[i] = -Q.AB_y[i];
542 Q_tmp.AB_z[i] = -Q.AB_z[i];
543 }
544
545 int ret = ostei_g_s_f_s(P_tmp, Q_tmp, screen_tol, work, INT__s_g_s_f);
546
547 return ret;
548 }
549
550