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_k_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_k_s)8 int ostei_h_s_k_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_k_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__h_s_k_s);
17 memset(INT__h_s_k_s, 0, P.nshell12_clip * Q.nshell12_clip * 756 * 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__s_s_p_s = primwork + 13;
31 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_d_s = primwork + 49;
32 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_f_s = primwork + 115;
33 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_g_s = primwork + 215;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_h_s = primwork + 350;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_i_s = primwork + 518;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_k_s = primwork + 714;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_f_s = primwork + 930;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_g_s = primwork + 1080;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_h_s = primwork + 1305;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_i_s = primwork + 1620;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_k_s = primwork + 2040;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_g_s = primwork + 2580;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_h_s = primwork + 2940;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_i_s = primwork + 3444;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_k_s = primwork + 4116;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_h_s = primwork + 4980;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_i_s = primwork + 5610;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_k_s = primwork + 6450;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_i_s = primwork + 7530;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_k_s = primwork + 8370;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_k_s = primwork + 9450;
52 double * const hrrwork = (double *)(primwork + 10206);
53
54
55 // Create constants
56 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
57 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
58 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
59 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
60 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
61 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
62 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
63 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
64
65
66 ////////////////////////////////////////
67 // Loop over shells and primitives
68 ////////////////////////////////////////
69
70 abcd = 0;
71 istart = 0;
72 for(ab = 0; ab < P.nshell12_clip; ++ab)
73 {
74 const int iend = istart + P.nprim12[ab];
75
76 cd = 0;
77 jstart = 0;
78
79 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
80 {
81 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
82 int jend = jstart;
83 for(i = 0; i < nshellbatch; i++)
84 jend += Q.nprim12[cd+i];
85
86
87 for(i = istart; i < iend; ++i)
88 {
89 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
90
91 if(check_screen)
92 {
93 // Skip this whole thing if always insignificant
94 if((P.screen[i] * Q.screen_max) < screen_tol)
95 continue;
96 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
97 }
98
99 icd = 0;
100 iprimcd = 0;
101 nprim_icd = Q.nprim12[cd];
102 double * restrict PRIM_PTR_INT__h_s_k_s = INT__h_s_k_s + abcd * 756;
103
104
105
106 // Load these one per loop over i
107 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
108 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
109 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
110
111 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
112
113 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
114 {
115 // calculate the shell offsets
116 // these are the offset from the shell pointed to by cd
117 // for each element
118 int shelloffsets[SIMINT_SIMD_LEN] = {0};
119 int lastoffset = 0;
120 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
121
122 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
123 {
124 // Handle if the first element of the vector is a new shell
125 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
126 {
127 nprim_icd += Q.nprim12[cd + (++icd)];
128 PRIM_PTR_INT__h_s_k_s += 756;
129 }
130 iprimcd++;
131 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
132 {
133 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
134 {
135 shelloffsets[n] = shelloffsets[n-1] + 1;
136 lastoffset++;
137 nprim_icd += Q.nprim12[cd + (++icd)];
138 }
139 else
140 shelloffsets[n] = shelloffsets[n-1];
141 iprimcd++;
142 }
143 }
144 else
145 iprimcd += SIMINT_SIMD_LEN;
146
147 // Do we have to compute this vector (or has it been screened out)?
148 // (not_screened != 0 means we have to do this vector)
149 if(check_screen)
150 {
151 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
152 if(vmax < screen_tol)
153 {
154 PRIM_PTR_INT__h_s_k_s += lastoffset*756;
155 continue;
156 }
157 }
158
159 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
160 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
161 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
162 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
163
164
165 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
166 SIMINT_DBLTYPE PQ[3];
167 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
168 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
169 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
170 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
171 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
172 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
173
174 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
175 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
176 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
177 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
178 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
179 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
180 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
181
182 // NOTE: Minus sign!
183 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
184 SIMINT_DBLTYPE aop_PQ[3];
185 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
186 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
187 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
188
189 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
190 SIMINT_DBLTYPE aoq_PQ[3];
191 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
192 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
193 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
194 // Put a minus sign here so we don't have to in RR routines
195 a_over_q = SIMINT_NEG(a_over_q);
196
197
198 //////////////////////////////////////////////
199 // Fjt function section
200 // Maximum v value: 12
201 //////////////////////////////////////////////
202 // The parameter to the Fjt function
203 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
204
205
206 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
207
208
209 boys_F_split(PRIM_INT__s_s_s_s, F_x, 12);
210 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
211 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
212 for(n = 0; n <= 12; n++)
213 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
214
215 //////////////////////////////////////////////
216 // Primitive integrals: Vertical recurrance
217 //////////////////////////////////////////////
218
219 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
220 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
221 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
222 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
223 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
224 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
225 const SIMINT_DBLTYPE vrr_const_3_over_2q = SIMINT_MUL(const_3, one_over_2q);
226 const SIMINT_DBLTYPE vrr_const_4_over_2q = SIMINT_MUL(const_4, one_over_2q);
227 const SIMINT_DBLTYPE vrr_const_5_over_2q = SIMINT_MUL(const_5, one_over_2q);
228 const SIMINT_DBLTYPE vrr_const_6_over_2q = SIMINT_MUL(const_6, one_over_2q);
229 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
230 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
231 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
232 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
233 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
234 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
235 const SIMINT_DBLTYPE vrr_const_7_over_2pq = SIMINT_MUL(const_7, one_over_2pq);
236
237
238
239 // Forming PRIM_INT__s_s_p_s[12 * 3];
240 for(n = 0; n < 12; ++n) // loop over orders of auxiliary function
241 {
242
243 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
244 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_p_s[n * 3 + 0]);
245
246 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
247 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_p_s[n * 3 + 1]);
248
249 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
250 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_p_s[n * 3 + 2]);
251
252 }
253
254
255
256 // Forming PRIM_INT__s_s_d_s[11 * 6];
257 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
258 {
259
260 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_p_s[n * 3 + 0]);
261 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_s_d_s[n * 6 + 0]);
262 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_FMADD( vrr_const_1_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__s_s_d_s[n * 6 + 0]);
263
264 PRIM_INT__s_s_d_s[n * 6 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 0]);
265 PRIM_INT__s_s_d_s[n * 6 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_s_d_s[n * 6 + 1]);
266
267 PRIM_INT__s_s_d_s[n * 6 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 0]);
268 PRIM_INT__s_s_d_s[n * 6 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_s_d_s[n * 6 + 2]);
269
270 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 1]);
271 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_s_d_s[n * 6 + 3]);
272 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_FMADD( vrr_const_1_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__s_s_d_s[n * 6 + 3]);
273
274 PRIM_INT__s_s_d_s[n * 6 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 1]);
275 PRIM_INT__s_s_d_s[n * 6 + 4] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_s_d_s[n * 6 + 4]);
276
277 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 2]);
278 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__s_s_d_s[n * 6 + 5]);
279 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_FMADD( vrr_const_1_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__s_s_d_s[n * 6 + 5]);
280
281 }
282
283
284
285 // Forming PRIM_INT__s_s_f_s[10 * 10];
286 for(n = 0; n < 10; ++n) // loop over orders of auxiliary function
287 {
288
289 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
290 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__s_s_f_s[n * 10 + 0]);
291 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_FMADD( vrr_const_2_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_s_p_s[n * 3 + 0]), PRIM_INT__s_s_f_s[n * 10 + 0]);
292
293 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
294 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__s_s_f_s[n * 10 + 1]);
295
296 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
297 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__s_s_f_s[n * 10 + 2]);
298
299 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
300 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__s_s_f_s[n * 10 + 3]);
301
302 PRIM_INT__s_s_f_s[n * 10 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
303 PRIM_INT__s_s_f_s[n * 10 + 4] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 1], PRIM_INT__s_s_f_s[n * 10 + 4]);
304
305 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
306 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__s_s_f_s[n * 10 + 5]);
307
308 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
309 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__s_s_f_s[n * 10 + 6]);
310 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_FMADD( vrr_const_2_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_s_p_s[n * 3 + 1]), PRIM_INT__s_s_f_s[n * 10 + 6]);
311
312 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
313 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__s_s_f_s[n * 10 + 7]);
314
315 PRIM_INT__s_s_f_s[n * 10 + 8] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
316 PRIM_INT__s_s_f_s[n * 10 + 8] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__s_s_f_s[n * 10 + 8]);
317
318 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
319 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__s_s_f_s[n * 10 + 9]);
320 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_FMADD( vrr_const_2_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__s_s_p_s[n * 3 + 2]), PRIM_INT__s_s_f_s[n * 10 + 9]);
321
322 }
323
324
325 VRR_K_s_s_g_s(
326 PRIM_INT__s_s_g_s,
327 PRIM_INT__s_s_f_s,
328 PRIM_INT__s_s_d_s,
329 Q_PA,
330 a_over_q,
331 aoq_PQ,
332 one_over_2q,
333 9);
334
335
336 VRR_K_s_s_h_s(
337 PRIM_INT__s_s_h_s,
338 PRIM_INT__s_s_g_s,
339 PRIM_INT__s_s_f_s,
340 Q_PA,
341 a_over_q,
342 aoq_PQ,
343 one_over_2q,
344 8);
345
346
347 ostei_general_vrr1_K(6, 7,
348 one_over_2q, a_over_q, aoq_PQ, Q_PA,
349 PRIM_INT__s_s_h_s, PRIM_INT__s_s_g_s, PRIM_INT__s_s_i_s);
350
351
352 ostei_general_vrr1_K(7, 6,
353 one_over_2q, a_over_q, aoq_PQ, Q_PA,
354 PRIM_INT__s_s_i_s, PRIM_INT__s_s_h_s, PRIM_INT__s_s_k_s);
355
356
357 ostei_general_vrr_I(1, 0, 7, 0, 5,
358 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
359 PRIM_INT__s_s_k_s, NULL, NULL, PRIM_INT__s_s_i_s, NULL, PRIM_INT__p_s_k_s);
360
361
362 ostei_general_vrr_I(1, 0, 6, 0, 5,
363 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
364 PRIM_INT__s_s_i_s, NULL, NULL, PRIM_INT__s_s_h_s, NULL, PRIM_INT__p_s_i_s);
365
366
367 ostei_general_vrr_I(2, 0, 7, 0, 4,
368 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
369 PRIM_INT__p_s_k_s, PRIM_INT__s_s_k_s, NULL, PRIM_INT__p_s_i_s, NULL, PRIM_INT__d_s_k_s);
370
371
372 ostei_general_vrr_I(1, 0, 5, 0, 5,
373 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
374 PRIM_INT__s_s_h_s, NULL, NULL, PRIM_INT__s_s_g_s, NULL, PRIM_INT__p_s_h_s);
375
376
377 ostei_general_vrr_I(2, 0, 6, 0, 4,
378 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
379 PRIM_INT__p_s_i_s, PRIM_INT__s_s_i_s, NULL, PRIM_INT__p_s_h_s, NULL, PRIM_INT__d_s_i_s);
380
381
382 ostei_general_vrr_I(3, 0, 7, 0, 3,
383 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
384 PRIM_INT__d_s_k_s, PRIM_INT__p_s_k_s, NULL, PRIM_INT__d_s_i_s, NULL, PRIM_INT__f_s_k_s);
385
386
387 VRR_I_p_s_g_s(
388 PRIM_INT__p_s_g_s,
389 PRIM_INT__s_s_g_s,
390 PRIM_INT__s_s_f_s,
391 P_PA,
392 aop_PQ,
393 one_over_2pq,
394 5);
395
396
397 ostei_general_vrr_I(2, 0, 5, 0, 4,
398 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
399 PRIM_INT__p_s_h_s, PRIM_INT__s_s_h_s, NULL, PRIM_INT__p_s_g_s, NULL, PRIM_INT__d_s_h_s);
400
401
402 ostei_general_vrr_I(3, 0, 6, 0, 3,
403 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
404 PRIM_INT__d_s_i_s, PRIM_INT__p_s_i_s, NULL, PRIM_INT__d_s_h_s, NULL, PRIM_INT__f_s_i_s);
405
406
407 ostei_general_vrr_I(4, 0, 7, 0, 2,
408 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
409 PRIM_INT__f_s_k_s, PRIM_INT__d_s_k_s, NULL, PRIM_INT__f_s_i_s, NULL, PRIM_INT__g_s_k_s);
410
411
412 VRR_I_p_s_f_s(
413 PRIM_INT__p_s_f_s,
414 PRIM_INT__s_s_f_s,
415 PRIM_INT__s_s_d_s,
416 P_PA,
417 aop_PQ,
418 one_over_2pq,
419 5);
420
421
422 ostei_general_vrr_I(2, 0, 4, 0, 4,
423 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
424 PRIM_INT__p_s_g_s, PRIM_INT__s_s_g_s, NULL, PRIM_INT__p_s_f_s, NULL, PRIM_INT__d_s_g_s);
425
426
427 ostei_general_vrr_I(3, 0, 5, 0, 3,
428 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
429 PRIM_INT__d_s_h_s, PRIM_INT__p_s_h_s, NULL, PRIM_INT__d_s_g_s, NULL, PRIM_INT__f_s_h_s);
430
431
432 ostei_general_vrr_I(4, 0, 6, 0, 2,
433 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
434 PRIM_INT__f_s_i_s, PRIM_INT__d_s_i_s, NULL, PRIM_INT__f_s_h_s, NULL, PRIM_INT__g_s_i_s);
435
436
437 ostei_general_vrr_I(5, 0, 7, 0, 1,
438 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
439 PRIM_INT__g_s_k_s, PRIM_INT__f_s_k_s, NULL, PRIM_INT__g_s_i_s, NULL, PRIM_INT__h_s_k_s);
440
441
442
443
444 ////////////////////////////////////
445 // Accumulate contracted integrals
446 ////////////////////////////////////
447 if(lastoffset == 0)
448 {
449 contract_all(756, PRIM_INT__h_s_k_s, PRIM_PTR_INT__h_s_k_s);
450 }
451 else
452 {
453 contract(756, shelloffsets, PRIM_INT__h_s_k_s, PRIM_PTR_INT__h_s_k_s);
454 PRIM_PTR_INT__h_s_k_s += lastoffset*756;
455 }
456
457 } // close loop over j
458 } // close loop over i
459
460 //Advance to the next batch
461 jstart = SIMINT_SIMD_ROUND(jend);
462 abcd += nshellbatch;
463
464 } // close loop cdbatch
465
466 istart = iend;
467 } // close loop over ab
468
469 return P.nshell12_clip * Q.nshell12_clip;
470 }
471
ostei_s_h_k_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_k_s)472 int ostei_s_h_k_s(struct simint_multi_shellpair const P,
473 struct simint_multi_shellpair const Q,
474 double screen_tol,
475 double * const restrict work,
476 double * const restrict INT__s_h_k_s)
477 {
478 double P_AB[3*P.nshell12];
479 struct simint_multi_shellpair P_tmp = P;
480 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
481 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
482 P_tmp.AB_x = P_AB;
483 P_tmp.AB_y = P_AB + P.nshell12;
484 P_tmp.AB_z = P_AB + 2*P.nshell12;
485
486 for(int i = 0; i < P.nshell12; i++)
487 {
488 P_tmp.AB_x[i] = -P.AB_x[i];
489 P_tmp.AB_y[i] = -P.AB_y[i];
490 P_tmp.AB_z[i] = -P.AB_z[i];
491 }
492
493 int ret = ostei_h_s_k_s(P_tmp, Q, screen_tol, work, INT__s_h_k_s);
494
495 return ret;
496 }
497
ostei_h_s_s_k(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_k)498 int ostei_h_s_s_k(struct simint_multi_shellpair const P,
499 struct simint_multi_shellpair const Q,
500 double screen_tol,
501 double * const restrict work,
502 double * const restrict INT__h_s_s_k)
503 {
504 double Q_AB[3*Q.nshell12];
505 struct simint_multi_shellpair Q_tmp = Q;
506 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
507 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
508 Q_tmp.AB_x = Q_AB;
509 Q_tmp.AB_y = Q_AB + Q.nshell12;
510 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
511
512 for(int i = 0; i < Q.nshell12; i++)
513 {
514 Q_tmp.AB_x[i] = -Q.AB_x[i];
515 Q_tmp.AB_y[i] = -Q.AB_y[i];
516 Q_tmp.AB_z[i] = -Q.AB_z[i];
517 }
518
519 int ret = ostei_h_s_k_s(P, Q_tmp, screen_tol, work, INT__h_s_s_k);
520
521 return ret;
522 }
523
ostei_s_h_s_k(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_k)524 int ostei_s_h_s_k(struct simint_multi_shellpair const P,
525 struct simint_multi_shellpair const Q,
526 double screen_tol,
527 double * const restrict work,
528 double * const restrict INT__s_h_s_k)
529 {
530 double P_AB[3*P.nshell12];
531 struct simint_multi_shellpair P_tmp = P;
532 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
533 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
534 P_tmp.AB_x = P_AB;
535 P_tmp.AB_y = P_AB + P.nshell12;
536 P_tmp.AB_z = P_AB + 2*P.nshell12;
537
538 for(int i = 0; i < P.nshell12; i++)
539 {
540 P_tmp.AB_x[i] = -P.AB_x[i];
541 P_tmp.AB_y[i] = -P.AB_y[i];
542 P_tmp.AB_z[i] = -P.AB_z[i];
543 }
544
545 double Q_AB[3*Q.nshell12];
546 struct simint_multi_shellpair Q_tmp = Q;
547 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
548 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
549 Q_tmp.AB_x = Q_AB;
550 Q_tmp.AB_y = Q_AB + Q.nshell12;
551 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
552
553 for(int i = 0; i < Q.nshell12; i++)
554 {
555 Q_tmp.AB_x[i] = -Q.AB_x[i];
556 Q_tmp.AB_y[i] = -Q.AB_y[i];
557 Q_tmp.AB_z[i] = -Q.AB_z[i];
558 }
559
560 int ret = ostei_h_s_k_s(P_tmp, Q_tmp, screen_tol, work, INT__s_h_s_k);
561
562 return ret;
563 }
564
565