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_d_s_i_f(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__d_s_i_f)8 int ostei_d_s_i_f(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__d_s_i_f)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__d_s_i_f);
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 ibra;
26
27 // partition workspace
28 double * const INT__d_s_i_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__d_s_k_s = work + (SIMINT_NSHELL_SIMD * 168);
30 double * const INT__d_s_l_s = work + (SIMINT_NSHELL_SIMD * 384);
31 double * const INT__d_s_m_s = work + (SIMINT_NSHELL_SIMD * 654);
32 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*984);
33 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_p_s = primwork + 12;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_d_s = primwork + 45;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_f_s = primwork + 105;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_g_s = primwork + 195;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_h_s = primwork + 315;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_i_s = primwork + 462;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_k_s = primwork + 630;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_l_s = primwork + 810;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_m_s = primwork + 990;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_h_s = primwork + 1155;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_i_s = primwork + 1281;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_k_s = primwork + 1449;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_l_s = primwork + 1665;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_m_s = primwork + 1935;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_i_s = primwork + 2265;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_k_s = primwork + 2433;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_l_s = primwork + 2649;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_m_s = primwork + 2919;
52 double * const hrrwork = (double *)(primwork + 3249);
53 double * const HRR_INT__d_s_i_p = hrrwork + 0;
54 double * const HRR_INT__d_s_i_d = hrrwork + 504;
55 double * const HRR_INT__d_s_k_p = hrrwork + 1512;
56 double * const HRR_INT__d_s_k_d = hrrwork + 2160;
57 double * const HRR_INT__d_s_l_p = hrrwork + 3456;
58
59
60 // Create constants
61 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
62 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
63 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
64 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
65 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
66 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
67 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
68 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
69 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
70 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
71
72
73 ////////////////////////////////////////
74 // Loop over shells and primitives
75 ////////////////////////////////////////
76
77 real_abcd = 0;
78 istart = 0;
79 for(ab = 0; ab < P.nshell12_clip; ++ab)
80 {
81 const int iend = istart + P.nprim12[ab];
82
83 cd = 0;
84 jstart = 0;
85
86 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
87 {
88 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
89 int jend = jstart;
90 for(i = 0; i < nshellbatch; i++)
91 jend += Q.nprim12[cd+i];
92
93 // Clear the beginning of the workspace (where we are accumulating integrals)
94 memset(work, 0, SIMINT_NSHELL_SIMD * 984 * sizeof(double));
95 abcd = 0;
96
97
98 for(i = istart; i < iend; ++i)
99 {
100 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
101
102 if(check_screen)
103 {
104 // Skip this whole thing if always insignificant
105 if((P.screen[i] * Q.screen_max) < screen_tol)
106 continue;
107 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
108 }
109
110 icd = 0;
111 iprimcd = 0;
112 nprim_icd = Q.nprim12[cd];
113 double * restrict PRIM_PTR_INT__d_s_i_s = INT__d_s_i_s + abcd * 168;
114 double * restrict PRIM_PTR_INT__d_s_k_s = INT__d_s_k_s + abcd * 216;
115 double * restrict PRIM_PTR_INT__d_s_l_s = INT__d_s_l_s + abcd * 270;
116 double * restrict PRIM_PTR_INT__d_s_m_s = INT__d_s_m_s + abcd * 330;
117
118
119
120 // Load these one per loop over i
121 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
122 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
123 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
124
125 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
126
127 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
128 {
129 // calculate the shell offsets
130 // these are the offset from the shell pointed to by cd
131 // for each element
132 int shelloffsets[SIMINT_SIMD_LEN] = {0};
133 int lastoffset = 0;
134 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
135
136 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
137 {
138 // Handle if the first element of the vector is a new shell
139 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
140 {
141 nprim_icd += Q.nprim12[cd + (++icd)];
142 PRIM_PTR_INT__d_s_i_s += 168;
143 PRIM_PTR_INT__d_s_k_s += 216;
144 PRIM_PTR_INT__d_s_l_s += 270;
145 PRIM_PTR_INT__d_s_m_s += 330;
146 }
147 iprimcd++;
148 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
149 {
150 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
151 {
152 shelloffsets[n] = shelloffsets[n-1] + 1;
153 lastoffset++;
154 nprim_icd += Q.nprim12[cd + (++icd)];
155 }
156 else
157 shelloffsets[n] = shelloffsets[n-1];
158 iprimcd++;
159 }
160 }
161 else
162 iprimcd += SIMINT_SIMD_LEN;
163
164 // Do we have to compute this vector (or has it been screened out)?
165 // (not_screened != 0 means we have to do this vector)
166 if(check_screen)
167 {
168 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
169 if(vmax < screen_tol)
170 {
171 PRIM_PTR_INT__d_s_i_s += lastoffset*168;
172 PRIM_PTR_INT__d_s_k_s += lastoffset*216;
173 PRIM_PTR_INT__d_s_l_s += lastoffset*270;
174 PRIM_PTR_INT__d_s_m_s += lastoffset*330;
175 continue;
176 }
177 }
178
179 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
180 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
181 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
182 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
183
184
185 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
186 SIMINT_DBLTYPE PQ[3];
187 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
188 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
189 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
190 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
191 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
192 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
193
194 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
195 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
196 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
197 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
198 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
199 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
200 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
201
202 // NOTE: Minus sign!
203 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
204 SIMINT_DBLTYPE aop_PQ[3];
205 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
206 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
207 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
208
209 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
210 SIMINT_DBLTYPE aoq_PQ[3];
211 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
212 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
213 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
214 // Put a minus sign here so we don't have to in RR routines
215 a_over_q = SIMINT_NEG(a_over_q);
216
217
218 //////////////////////////////////////////////
219 // Fjt function section
220 // Maximum v value: 11
221 //////////////////////////////////////////////
222 // The parameter to the Fjt function
223 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
224
225
226 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
227
228
229 boys_F_split(PRIM_INT__s_s_s_s, F_x, 11);
230 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
231 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
232 for(n = 0; n <= 11; n++)
233 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
234
235 //////////////////////////////////////////////
236 // Primitive integrals: Vertical recurrance
237 //////////////////////////////////////////////
238
239 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
240 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
241 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
242 const SIMINT_DBLTYPE vrr_const_3_over_2q = SIMINT_MUL(const_3, one_over_2q);
243 const SIMINT_DBLTYPE vrr_const_4_over_2q = SIMINT_MUL(const_4, one_over_2q);
244 const SIMINT_DBLTYPE vrr_const_5_over_2q = SIMINT_MUL(const_5, one_over_2q);
245 const SIMINT_DBLTYPE vrr_const_6_over_2q = SIMINT_MUL(const_6, one_over_2q);
246 const SIMINT_DBLTYPE vrr_const_7_over_2q = SIMINT_MUL(const_7, one_over_2q);
247 const SIMINT_DBLTYPE vrr_const_8_over_2q = SIMINT_MUL(const_8, one_over_2q);
248 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
249 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
250 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
251 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
252 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
253 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
254 const SIMINT_DBLTYPE vrr_const_7_over_2pq = SIMINT_MUL(const_7, one_over_2pq);
255 const SIMINT_DBLTYPE vrr_const_8_over_2pq = SIMINT_MUL(const_8, one_over_2pq);
256 const SIMINT_DBLTYPE vrr_const_9_over_2pq = SIMINT_MUL(const_9, one_over_2pq);
257
258
259
260 // Forming PRIM_INT__s_s_p_s[11 * 3];
261 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
262 {
263
264 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
265 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]);
266
267 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
268 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]);
269
270 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
271 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]);
272
273 }
274
275
276
277 // Forming PRIM_INT__s_s_d_s[10 * 6];
278 for(n = 0; n < 10; ++n) // loop over orders of auxiliary function
279 {
280
281 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_p_s[n * 3 + 0]);
282 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]);
283 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]);
284
285 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 1]);
286 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]);
287 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]);
288
289 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 2]);
290 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]);
291 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]);
292
293 }
294
295
296
297 // Forming PRIM_INT__s_s_f_s[9 * 10];
298 for(n = 0; n < 9; ++n) // loop over orders of auxiliary function
299 {
300
301 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
302 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]);
303 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]);
304
305 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
306 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]);
307
308 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
309 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]);
310
311 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
312 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]);
313
314 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
315 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]);
316
317 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
318 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]);
319 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]);
320
321 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
322 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]);
323
324 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
325 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]);
326 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]);
327
328 }
329
330
331 VRR_K_s_s_g_s(
332 PRIM_INT__s_s_g_s,
333 PRIM_INT__s_s_f_s,
334 PRIM_INT__s_s_d_s,
335 Q_PA,
336 a_over_q,
337 aoq_PQ,
338 one_over_2q,
339 8);
340
341
342 VRR_K_s_s_h_s(
343 PRIM_INT__s_s_h_s,
344 PRIM_INT__s_s_g_s,
345 PRIM_INT__s_s_f_s,
346 Q_PA,
347 a_over_q,
348 aoq_PQ,
349 one_over_2q,
350 7);
351
352
353 ostei_general_vrr1_K(6, 6,
354 one_over_2q, a_over_q, aoq_PQ, Q_PA,
355 PRIM_INT__s_s_h_s, PRIM_INT__s_s_g_s, PRIM_INT__s_s_i_s);
356
357
358 ostei_general_vrr_I(1, 0, 6, 0, 2,
359 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
360 PRIM_INT__s_s_i_s, NULL, NULL, PRIM_INT__s_s_h_s, NULL, PRIM_INT__p_s_i_s);
361
362
363 ostei_general_vrr_I(1, 0, 5, 0, 2,
364 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
365 PRIM_INT__s_s_h_s, NULL, NULL, PRIM_INT__s_s_g_s, NULL, PRIM_INT__p_s_h_s);
366
367
368 ostei_general_vrr_I(2, 0, 6, 0, 1,
369 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
370 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);
371
372
373 ostei_general_vrr1_K(7, 5,
374 one_over_2q, a_over_q, aoq_PQ, Q_PA,
375 PRIM_INT__s_s_i_s, PRIM_INT__s_s_h_s, PRIM_INT__s_s_k_s);
376
377
378 ostei_general_vrr_I(1, 0, 7, 0, 2,
379 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
380 PRIM_INT__s_s_k_s, NULL, NULL, PRIM_INT__s_s_i_s, NULL, PRIM_INT__p_s_k_s);
381
382
383 ostei_general_vrr_I(2, 0, 7, 0, 1,
384 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
385 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);
386
387
388 ostei_general_vrr1_K(8, 4,
389 one_over_2q, a_over_q, aoq_PQ, Q_PA,
390 PRIM_INT__s_s_k_s, PRIM_INT__s_s_i_s, PRIM_INT__s_s_l_s);
391
392
393 ostei_general_vrr_I(1, 0, 8, 0, 2,
394 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
395 PRIM_INT__s_s_l_s, NULL, NULL, PRIM_INT__s_s_k_s, NULL, PRIM_INT__p_s_l_s);
396
397
398 ostei_general_vrr_I(2, 0, 8, 0, 1,
399 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
400 PRIM_INT__p_s_l_s, PRIM_INT__s_s_l_s, NULL, PRIM_INT__p_s_k_s, NULL, PRIM_INT__d_s_l_s);
401
402
403 ostei_general_vrr1_K(9, 3,
404 one_over_2q, a_over_q, aoq_PQ, Q_PA,
405 PRIM_INT__s_s_l_s, PRIM_INT__s_s_k_s, PRIM_INT__s_s_m_s);
406
407
408 ostei_general_vrr_I(1, 0, 9, 0, 2,
409 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
410 PRIM_INT__s_s_m_s, NULL, NULL, PRIM_INT__s_s_l_s, NULL, PRIM_INT__p_s_m_s);
411
412
413 ostei_general_vrr_I(2, 0, 9, 0, 1,
414 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
415 PRIM_INT__p_s_m_s, PRIM_INT__s_s_m_s, NULL, PRIM_INT__p_s_l_s, NULL, PRIM_INT__d_s_m_s);
416
417
418
419
420 ////////////////////////////////////
421 // Accumulate contracted integrals
422 ////////////////////////////////////
423 if(lastoffset == 0)
424 {
425 contract_all(168, PRIM_INT__d_s_i_s, PRIM_PTR_INT__d_s_i_s);
426 contract_all(216, PRIM_INT__d_s_k_s, PRIM_PTR_INT__d_s_k_s);
427 contract_all(270, PRIM_INT__d_s_l_s, PRIM_PTR_INT__d_s_l_s);
428 contract_all(330, PRIM_INT__d_s_m_s, PRIM_PTR_INT__d_s_m_s);
429 }
430 else
431 {
432 contract(168, shelloffsets, PRIM_INT__d_s_i_s, PRIM_PTR_INT__d_s_i_s);
433 contract(216, shelloffsets, PRIM_INT__d_s_k_s, PRIM_PTR_INT__d_s_k_s);
434 contract(270, shelloffsets, PRIM_INT__d_s_l_s, PRIM_PTR_INT__d_s_l_s);
435 contract(330, shelloffsets, PRIM_INT__d_s_m_s, PRIM_PTR_INT__d_s_m_s);
436 PRIM_PTR_INT__d_s_i_s += lastoffset*168;
437 PRIM_PTR_INT__d_s_k_s += lastoffset*216;
438 PRIM_PTR_INT__d_s_l_s += lastoffset*270;
439 PRIM_PTR_INT__d_s_m_s += lastoffset*330;
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
448 //////////////////////////////////////////////
449 // Contracted integrals: Horizontal recurrance
450 //////////////////////////////////////////////
451
452
453
454
455 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
456 {
457 const double hCD[3] = { Q.AB_x[cd+abcd], Q.AB_y[cd+abcd], Q.AB_z[cd+abcd] };
458
459 // set up HRR pointers
460 double const * restrict HRR_INT__d_s_i_s = INT__d_s_i_s + abcd * 168;
461 double const * restrict HRR_INT__d_s_k_s = INT__d_s_k_s + abcd * 216;
462 double const * restrict HRR_INT__d_s_l_s = INT__d_s_l_s + abcd * 270;
463 double const * restrict HRR_INT__d_s_m_s = INT__d_s_m_s + abcd * 330;
464 double * restrict HRR_INT__d_s_i_f = INT__d_s_i_f + real_abcd * 1680;
465
466 // form INT__d_s_i_p
467 ostei_general_hrr_L(2, 0, 6, 1, hCD, HRR_INT__d_s_k_s, HRR_INT__d_s_i_s, HRR_INT__d_s_i_p);
468
469 // form INT__d_s_k_p
470 ostei_general_hrr_L(2, 0, 7, 1, hCD, HRR_INT__d_s_l_s, HRR_INT__d_s_k_s, HRR_INT__d_s_k_p);
471
472 // form INT__d_s_l_p
473 ostei_general_hrr_L(2, 0, 8, 1, hCD, HRR_INT__d_s_m_s, HRR_INT__d_s_l_s, HRR_INT__d_s_l_p);
474
475 // form INT__d_s_i_d
476 ostei_general_hrr_L(2, 0, 6, 2, hCD, HRR_INT__d_s_k_p, HRR_INT__d_s_i_p, HRR_INT__d_s_i_d);
477
478 // form INT__d_s_k_d
479 ostei_general_hrr_L(2, 0, 7, 2, hCD, HRR_INT__d_s_l_p, HRR_INT__d_s_k_p, HRR_INT__d_s_k_d);
480
481 // form INT__d_s_i_f
482 ostei_general_hrr_L(2, 0, 6, 3, hCD, HRR_INT__d_s_k_d, HRR_INT__d_s_i_d, HRR_INT__d_s_i_f);
483
484
485 } // close HRR loop
486
487
488 } // close loop cdbatch
489
490 istart = iend;
491 } // close loop over ab
492
493 return P.nshell12_clip * Q.nshell12_clip;
494 }
495
ostei_s_d_i_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_d_i_f)496 int ostei_s_d_i_f(struct simint_multi_shellpair const P,
497 struct simint_multi_shellpair const Q,
498 double screen_tol,
499 double * const restrict work,
500 double * const restrict INT__s_d_i_f)
501 {
502 double P_AB[3*P.nshell12];
503 struct simint_multi_shellpair P_tmp = P;
504 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
505 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
506 P_tmp.AB_x = P_AB;
507 P_tmp.AB_y = P_AB + P.nshell12;
508 P_tmp.AB_z = P_AB + 2*P.nshell12;
509
510 for(int i = 0; i < P.nshell12; i++)
511 {
512 P_tmp.AB_x[i] = -P.AB_x[i];
513 P_tmp.AB_y[i] = -P.AB_y[i];
514 P_tmp.AB_z[i] = -P.AB_z[i];
515 }
516
517 int ret = ostei_d_s_i_f(P_tmp, Q, screen_tol, work, INT__s_d_i_f);
518 double buffer[1680] SIMINT_ALIGN_ARRAY_DBL;
519
520 for(int q = 0; q < ret; q++)
521 {
522 int idx = 0;
523 for(int a = 0; a < 1; ++a)
524 for(int b = 0; b < 6; ++b)
525 for(int c = 0; c < 28; ++c)
526 for(int d = 0; d < 10; ++d)
527 buffer[idx++] = INT__s_d_i_f[q*1680+b*280+a*280+c*10+d];
528
529 memcpy(INT__s_d_i_f+q*1680, buffer, 1680*sizeof(double));
530 }
531
532 return ret;
533 }
534
ostei_d_s_f_i(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__d_s_f_i)535 int ostei_d_s_f_i(struct simint_multi_shellpair const P,
536 struct simint_multi_shellpair const Q,
537 double screen_tol,
538 double * const restrict work,
539 double * const restrict INT__d_s_f_i)
540 {
541 double Q_AB[3*Q.nshell12];
542 struct simint_multi_shellpair Q_tmp = Q;
543 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
544 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
545 Q_tmp.AB_x = Q_AB;
546 Q_tmp.AB_y = Q_AB + Q.nshell12;
547 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
548
549 for(int i = 0; i < Q.nshell12; i++)
550 {
551 Q_tmp.AB_x[i] = -Q.AB_x[i];
552 Q_tmp.AB_y[i] = -Q.AB_y[i];
553 Q_tmp.AB_z[i] = -Q.AB_z[i];
554 }
555
556 int ret = ostei_d_s_i_f(P, Q_tmp, screen_tol, work, INT__d_s_f_i);
557 double buffer[1680] SIMINT_ALIGN_ARRAY_DBL;
558
559 for(int q = 0; q < ret; q++)
560 {
561 int idx = 0;
562 for(int a = 0; a < 6; ++a)
563 for(int b = 0; b < 1; ++b)
564 for(int c = 0; c < 10; ++c)
565 for(int d = 0; d < 28; ++d)
566 buffer[idx++] = INT__d_s_f_i[q*1680+a*280+b*280+d*10+c];
567
568 memcpy(INT__d_s_f_i+q*1680, buffer, 1680*sizeof(double));
569 }
570
571 return ret;
572 }
573
ostei_s_d_f_i(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__s_d_f_i)574 int ostei_s_d_f_i(struct simint_multi_shellpair const P,
575 struct simint_multi_shellpair const Q,
576 double screen_tol,
577 double * const restrict work,
578 double * const restrict INT__s_d_f_i)
579 {
580 double P_AB[3*P.nshell12];
581 struct simint_multi_shellpair P_tmp = P;
582 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
583 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
584 P_tmp.AB_x = P_AB;
585 P_tmp.AB_y = P_AB + P.nshell12;
586 P_tmp.AB_z = P_AB + 2*P.nshell12;
587
588 for(int i = 0; i < P.nshell12; i++)
589 {
590 P_tmp.AB_x[i] = -P.AB_x[i];
591 P_tmp.AB_y[i] = -P.AB_y[i];
592 P_tmp.AB_z[i] = -P.AB_z[i];
593 }
594
595 double Q_AB[3*Q.nshell12];
596 struct simint_multi_shellpair Q_tmp = Q;
597 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
598 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
599 Q_tmp.AB_x = Q_AB;
600 Q_tmp.AB_y = Q_AB + Q.nshell12;
601 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
602
603 for(int i = 0; i < Q.nshell12; i++)
604 {
605 Q_tmp.AB_x[i] = -Q.AB_x[i];
606 Q_tmp.AB_y[i] = -Q.AB_y[i];
607 Q_tmp.AB_z[i] = -Q.AB_z[i];
608 }
609
610 int ret = ostei_d_s_i_f(P_tmp, Q_tmp, screen_tol, work, INT__s_d_f_i);
611 double buffer[1680] SIMINT_ALIGN_ARRAY_DBL;
612
613 for(int q = 0; q < ret; q++)
614 {
615 int idx = 0;
616 for(int a = 0; a < 1; ++a)
617 for(int b = 0; b < 6; ++b)
618 for(int c = 0; c < 10; ++c)
619 for(int d = 0; d < 28; ++d)
620 buffer[idx++] = INT__s_d_f_i[q*1680+b*280+a*280+d*10+c];
621
622 memcpy(INT__s_d_f_i+q*1680, buffer, 1680*sizeof(double));
623 }
624
625 return ret;
626 }
627
628