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_s_f_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_f_s_f)8 int ostei_s_f_s_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__s_f_s_f)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__s_f_s_f);
17 memset(INT__s_f_s_f, 0, P.nshell12_clip * Q.nshell12_clip * 100 * 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_p_s_s = primwork + 7;
31 SIMINT_DBLTYPE * const restrict PRIM_INT__s_p_s_p = primwork + 25;
32 SIMINT_DBLTYPE * const restrict PRIM_INT__s_d_s_s = primwork + 52;
33 SIMINT_DBLTYPE * const restrict PRIM_INT__s_d_s_p = primwork + 82;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__s_d_s_d = primwork + 136;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__s_f_s_s = primwork + 208;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__s_f_s_p = primwork + 248;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__s_f_s_d = primwork + 338;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__s_f_s_f = primwork + 458;
39 double * const hrrwork = (double *)(primwork + 558);
40
41
42 // Create constants
43 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
44 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
45 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
46 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
47
48
49 ////////////////////////////////////////
50 // Loop over shells and primitives
51 ////////////////////////////////////////
52
53 abcd = 0;
54 istart = 0;
55 for(ab = 0; ab < P.nshell12_clip; ++ab)
56 {
57 const int iend = istart + P.nprim12[ab];
58
59 cd = 0;
60 jstart = 0;
61
62 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
63 {
64 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
65 int jend = jstart;
66 for(i = 0; i < nshellbatch; i++)
67 jend += Q.nprim12[cd+i];
68
69
70 for(i = istart; i < iend; ++i)
71 {
72 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
73
74 if(check_screen)
75 {
76 // Skip this whole thing if always insignificant
77 if((P.screen[i] * Q.screen_max) < screen_tol)
78 continue;
79 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
80 }
81
82 icd = 0;
83 iprimcd = 0;
84 nprim_icd = Q.nprim12[cd];
85 double * restrict PRIM_PTR_INT__s_f_s_f = INT__s_f_s_f + abcd * 100;
86
87
88
89 // Load these one per loop over i
90 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
91 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
92 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
93
94 const SIMINT_DBLTYPE P_PB[3] = { SIMINT_DBLSET1(P.PB_x[i]), SIMINT_DBLSET1(P.PB_y[i]), SIMINT_DBLSET1(P.PB_z[i]) };
95
96 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
97 {
98 // calculate the shell offsets
99 // these are the offset from the shell pointed to by cd
100 // for each element
101 int shelloffsets[SIMINT_SIMD_LEN] = {0};
102 int lastoffset = 0;
103 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
104
105 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
106 {
107 // Handle if the first element of the vector is a new shell
108 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
109 {
110 nprim_icd += Q.nprim12[cd + (++icd)];
111 PRIM_PTR_INT__s_f_s_f += 100;
112 }
113 iprimcd++;
114 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
115 {
116 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
117 {
118 shelloffsets[n] = shelloffsets[n-1] + 1;
119 lastoffset++;
120 nprim_icd += Q.nprim12[cd + (++icd)];
121 }
122 else
123 shelloffsets[n] = shelloffsets[n-1];
124 iprimcd++;
125 }
126 }
127 else
128 iprimcd += SIMINT_SIMD_LEN;
129
130 // Do we have to compute this vector (or has it been screened out)?
131 // (not_screened != 0 means we have to do this vector)
132 if(check_screen)
133 {
134 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
135 if(vmax < screen_tol)
136 {
137 PRIM_PTR_INT__s_f_s_f += lastoffset*100;
138 continue;
139 }
140 }
141
142 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
143 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
144 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
145 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
146
147
148 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
149 SIMINT_DBLTYPE PQ[3];
150 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
151 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
152 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
153 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
154 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
155 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
156
157 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
158 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
159 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
160 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
161 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
162 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
163 const SIMINT_DBLTYPE Q_PB[3] = { SIMINT_DBLLOAD(Q.PB_x, j), SIMINT_DBLLOAD(Q.PB_y, j), SIMINT_DBLLOAD(Q.PB_z, j) };
164
165 // NOTE: Minus sign!
166 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
167 SIMINT_DBLTYPE aop_PQ[3];
168 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
169 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
170 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
171
172 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
173 SIMINT_DBLTYPE aoq_PQ[3];
174 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
175 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
176 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
177 // Put a minus sign here so we don't have to in RR routines
178 a_over_q = SIMINT_NEG(a_over_q);
179
180
181 //////////////////////////////////////////////
182 // Fjt function section
183 // Maximum v value: 6
184 //////////////////////////////////////////////
185 // The parameter to the Fjt function
186 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
187
188
189 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
190
191
192 boys_F_split(PRIM_INT__s_s_s_s, F_x, 6);
193 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
194 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
195 for(n = 0; n <= 6; n++)
196 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
197
198 //////////////////////////////////////////////
199 // Primitive integrals: Vertical recurrance
200 //////////////////////////////////////////////
201
202 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
203 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
204 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
205 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
206 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
207 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
208 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
209
210
211
212 // Forming PRIM_INT__s_p_s_s[6 * 3];
213 for(n = 0; n < 6; ++n) // loop over orders of auxiliary function
214 {
215
216 PRIM_INT__s_p_s_s[n * 3 + 0] = SIMINT_MUL(P_PB[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
217 PRIM_INT__s_p_s_s[n * 3 + 0] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_p_s_s[n * 3 + 0]);
218
219 PRIM_INT__s_p_s_s[n * 3 + 1] = SIMINT_MUL(P_PB[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
220 PRIM_INT__s_p_s_s[n * 3 + 1] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_p_s_s[n * 3 + 1]);
221
222 PRIM_INT__s_p_s_s[n * 3 + 2] = SIMINT_MUL(P_PB[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
223 PRIM_INT__s_p_s_s[n * 3 + 2] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_p_s_s[n * 3 + 2]);
224
225 }
226
227
228
229 // Forming PRIM_INT__s_d_s_s[5 * 6];
230 for(n = 0; n < 5; ++n) // loop over orders of auxiliary function
231 {
232
233 PRIM_INT__s_d_s_s[n * 6 + 0] = SIMINT_MUL(P_PB[0], PRIM_INT__s_p_s_s[n * 3 + 0]);
234 PRIM_INT__s_d_s_s[n * 6 + 0] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_p_s_s[(n+1) * 3 + 0], PRIM_INT__s_d_s_s[n * 6 + 0]);
235 PRIM_INT__s_d_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__s_d_s_s[n * 6 + 0]);
236
237 PRIM_INT__s_d_s_s[n * 6 + 1] = SIMINT_MUL(P_PB[1], PRIM_INT__s_p_s_s[n * 3 + 0]);
238 PRIM_INT__s_d_s_s[n * 6 + 1] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_p_s_s[(n+1) * 3 + 0], PRIM_INT__s_d_s_s[n * 6 + 1]);
239
240 PRIM_INT__s_d_s_s[n * 6 + 2] = SIMINT_MUL(P_PB[2], PRIM_INT__s_p_s_s[n * 3 + 0]);
241 PRIM_INT__s_d_s_s[n * 6 + 2] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_p_s_s[(n+1) * 3 + 0], PRIM_INT__s_d_s_s[n * 6 + 2]);
242
243 PRIM_INT__s_d_s_s[n * 6 + 3] = SIMINT_MUL(P_PB[1], PRIM_INT__s_p_s_s[n * 3 + 1]);
244 PRIM_INT__s_d_s_s[n * 6 + 3] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_p_s_s[(n+1) * 3 + 1], PRIM_INT__s_d_s_s[n * 6 + 3]);
245 PRIM_INT__s_d_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__s_d_s_s[n * 6 + 3]);
246
247 PRIM_INT__s_d_s_s[n * 6 + 4] = SIMINT_MUL(P_PB[2], PRIM_INT__s_p_s_s[n * 3 + 1]);
248 PRIM_INT__s_d_s_s[n * 6 + 4] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_p_s_s[(n+1) * 3 + 1], PRIM_INT__s_d_s_s[n * 6 + 4]);
249
250 PRIM_INT__s_d_s_s[n * 6 + 5] = SIMINT_MUL(P_PB[2], PRIM_INT__s_p_s_s[n * 3 + 2]);
251 PRIM_INT__s_d_s_s[n * 6 + 5] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_p_s_s[(n+1) * 3 + 2], PRIM_INT__s_d_s_s[n * 6 + 5]);
252 PRIM_INT__s_d_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__s_d_s_s[n * 6 + 5]);
253
254 }
255
256
257
258 // Forming PRIM_INT__s_f_s_s[4 * 10];
259 for(n = 0; n < 4; ++n) // loop over orders of auxiliary function
260 {
261
262 PRIM_INT__s_f_s_s[n * 10 + 0] = SIMINT_MUL(P_PB[0], PRIM_INT__s_d_s_s[n * 6 + 0]);
263 PRIM_INT__s_f_s_s[n * 10 + 0] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_d_s_s[(n+1) * 6 + 0], PRIM_INT__s_f_s_s[n * 10 + 0]);
264 PRIM_INT__s_f_s_s[n * 10 + 0] = SIMINT_FMADD( vrr_const_2_over_2p, SIMINT_FMADD(a_over_p, PRIM_INT__s_p_s_s[(n+1) * 3 + 0], PRIM_INT__s_p_s_s[n * 3 + 0]), PRIM_INT__s_f_s_s[n * 10 + 0]);
265
266 PRIM_INT__s_f_s_s[n * 10 + 1] = SIMINT_MUL(P_PB[1], PRIM_INT__s_d_s_s[n * 6 + 0]);
267 PRIM_INT__s_f_s_s[n * 10 + 1] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_d_s_s[(n+1) * 6 + 0], PRIM_INT__s_f_s_s[n * 10 + 1]);
268
269 PRIM_INT__s_f_s_s[n * 10 + 2] = SIMINT_MUL(P_PB[2], PRIM_INT__s_d_s_s[n * 6 + 0]);
270 PRIM_INT__s_f_s_s[n * 10 + 2] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_d_s_s[(n+1) * 6 + 0], PRIM_INT__s_f_s_s[n * 10 + 2]);
271
272 PRIM_INT__s_f_s_s[n * 10 + 3] = SIMINT_MUL(P_PB[0], PRIM_INT__s_d_s_s[n * 6 + 3]);
273 PRIM_INT__s_f_s_s[n * 10 + 3] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_d_s_s[(n+1) * 6 + 3], PRIM_INT__s_f_s_s[n * 10 + 3]);
274
275 PRIM_INT__s_f_s_s[n * 10 + 4] = SIMINT_MUL(P_PB[2], PRIM_INT__s_d_s_s[n * 6 + 1]);
276 PRIM_INT__s_f_s_s[n * 10 + 4] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_d_s_s[(n+1) * 6 + 1], PRIM_INT__s_f_s_s[n * 10 + 4]);
277
278 PRIM_INT__s_f_s_s[n * 10 + 5] = SIMINT_MUL(P_PB[0], PRIM_INT__s_d_s_s[n * 6 + 5]);
279 PRIM_INT__s_f_s_s[n * 10 + 5] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__s_d_s_s[(n+1) * 6 + 5], PRIM_INT__s_f_s_s[n * 10 + 5]);
280
281 PRIM_INT__s_f_s_s[n * 10 + 6] = SIMINT_MUL(P_PB[1], PRIM_INT__s_d_s_s[n * 6 + 3]);
282 PRIM_INT__s_f_s_s[n * 10 + 6] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_d_s_s[(n+1) * 6 + 3], PRIM_INT__s_f_s_s[n * 10 + 6]);
283 PRIM_INT__s_f_s_s[n * 10 + 6] = SIMINT_FMADD( vrr_const_2_over_2p, SIMINT_FMADD(a_over_p, PRIM_INT__s_p_s_s[(n+1) * 3 + 1], PRIM_INT__s_p_s_s[n * 3 + 1]), PRIM_INT__s_f_s_s[n * 10 + 6]);
284
285 PRIM_INT__s_f_s_s[n * 10 + 7] = SIMINT_MUL(P_PB[2], PRIM_INT__s_d_s_s[n * 6 + 3]);
286 PRIM_INT__s_f_s_s[n * 10 + 7] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_d_s_s[(n+1) * 6 + 3], PRIM_INT__s_f_s_s[n * 10 + 7]);
287
288 PRIM_INT__s_f_s_s[n * 10 + 8] = SIMINT_MUL(P_PB[1], PRIM_INT__s_d_s_s[n * 6 + 5]);
289 PRIM_INT__s_f_s_s[n * 10 + 8] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__s_d_s_s[(n+1) * 6 + 5], PRIM_INT__s_f_s_s[n * 10 + 8]);
290
291 PRIM_INT__s_f_s_s[n * 10 + 9] = SIMINT_MUL(P_PB[2], PRIM_INT__s_d_s_s[n * 6 + 5]);
292 PRIM_INT__s_f_s_s[n * 10 + 9] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__s_d_s_s[(n+1) * 6 + 5], PRIM_INT__s_f_s_s[n * 10 + 9]);
293 PRIM_INT__s_f_s_s[n * 10 + 9] = SIMINT_FMADD( vrr_const_2_over_2p, SIMINT_FMADD(a_over_p, PRIM_INT__s_p_s_s[(n+1) * 3 + 2], PRIM_INT__s_p_s_s[n * 3 + 2]), PRIM_INT__s_f_s_s[n * 10 + 9]);
294
295 }
296
297
298 VRR_L_s_f_s_p(
299 PRIM_INT__s_f_s_p,
300 PRIM_INT__s_f_s_s,
301 PRIM_INT__s_d_s_s,
302 Q_PB,
303 aoq_PQ,
304 one_over_2pq,
305 3);
306
307
308
309 // Forming PRIM_INT__s_d_s_p[3 * 18];
310 for(n = 0; n < 3; ++n) // loop over orders of auxiliary function
311 {
312
313 PRIM_INT__s_d_s_p[n * 18 + 0] = SIMINT_MUL(Q_PB[0], PRIM_INT__s_d_s_s[n * 6 + 0]);
314 PRIM_INT__s_d_s_p[n * 18 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_d_s_s[(n+1) * 6 + 0], PRIM_INT__s_d_s_p[n * 18 + 0]);
315 PRIM_INT__s_d_s_p[n * 18 + 0] = SIMINT_FMADD( vrr_const_2_over_2pq, PRIM_INT__s_p_s_s[(n+1) * 3 + 0], PRIM_INT__s_d_s_p[n * 18 + 0]);
316
317 PRIM_INT__s_d_s_p[n * 18 + 1] = SIMINT_MUL(Q_PB[1], PRIM_INT__s_d_s_s[n * 6 + 0]);
318 PRIM_INT__s_d_s_p[n * 18 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_d_s_s[(n+1) * 6 + 0], PRIM_INT__s_d_s_p[n * 18 + 1]);
319
320 PRIM_INT__s_d_s_p[n * 18 + 2] = SIMINT_MUL(Q_PB[2], PRIM_INT__s_d_s_s[n * 6 + 0]);
321 PRIM_INT__s_d_s_p[n * 18 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_d_s_s[(n+1) * 6 + 0], PRIM_INT__s_d_s_p[n * 18 + 2]);
322
323 PRIM_INT__s_d_s_p[n * 18 + 3] = SIMINT_MUL(Q_PB[0], PRIM_INT__s_d_s_s[n * 6 + 1]);
324 PRIM_INT__s_d_s_p[n * 18 + 3] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_d_s_s[(n+1) * 6 + 1], PRIM_INT__s_d_s_p[n * 18 + 3]);
325 PRIM_INT__s_d_s_p[n * 18 + 3] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_p_s_s[(n+1) * 3 + 1], PRIM_INT__s_d_s_p[n * 18 + 3]);
326
327 PRIM_INT__s_d_s_p[n * 18 + 4] = SIMINT_MUL(Q_PB[1], PRIM_INT__s_d_s_s[n * 6 + 1]);
328 PRIM_INT__s_d_s_p[n * 18 + 4] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_d_s_s[(n+1) * 6 + 1], PRIM_INT__s_d_s_p[n * 18 + 4]);
329 PRIM_INT__s_d_s_p[n * 18 + 4] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_p_s_s[(n+1) * 3 + 0], PRIM_INT__s_d_s_p[n * 18 + 4]);
330
331 PRIM_INT__s_d_s_p[n * 18 + 5] = SIMINT_MUL(Q_PB[2], PRIM_INT__s_d_s_s[n * 6 + 1]);
332 PRIM_INT__s_d_s_p[n * 18 + 5] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_d_s_s[(n+1) * 6 + 1], PRIM_INT__s_d_s_p[n * 18 + 5]);
333
334 PRIM_INT__s_d_s_p[n * 18 + 6] = SIMINT_MUL(Q_PB[0], PRIM_INT__s_d_s_s[n * 6 + 2]);
335 PRIM_INT__s_d_s_p[n * 18 + 6] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_d_s_s[(n+1) * 6 + 2], PRIM_INT__s_d_s_p[n * 18 + 6]);
336 PRIM_INT__s_d_s_p[n * 18 + 6] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_p_s_s[(n+1) * 3 + 2], PRIM_INT__s_d_s_p[n * 18 + 6]);
337
338 PRIM_INT__s_d_s_p[n * 18 + 7] = SIMINT_MUL(Q_PB[1], PRIM_INT__s_d_s_s[n * 6 + 2]);
339 PRIM_INT__s_d_s_p[n * 18 + 7] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_d_s_s[(n+1) * 6 + 2], PRIM_INT__s_d_s_p[n * 18 + 7]);
340
341 PRIM_INT__s_d_s_p[n * 18 + 8] = SIMINT_MUL(Q_PB[2], PRIM_INT__s_d_s_s[n * 6 + 2]);
342 PRIM_INT__s_d_s_p[n * 18 + 8] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_d_s_s[(n+1) * 6 + 2], PRIM_INT__s_d_s_p[n * 18 + 8]);
343 PRIM_INT__s_d_s_p[n * 18 + 8] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_p_s_s[(n+1) * 3 + 0], PRIM_INT__s_d_s_p[n * 18 + 8]);
344
345 PRIM_INT__s_d_s_p[n * 18 + 9] = SIMINT_MUL(Q_PB[0], PRIM_INT__s_d_s_s[n * 6 + 3]);
346 PRIM_INT__s_d_s_p[n * 18 + 9] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_d_s_s[(n+1) * 6 + 3], PRIM_INT__s_d_s_p[n * 18 + 9]);
347
348 PRIM_INT__s_d_s_p[n * 18 + 10] = SIMINT_MUL(Q_PB[1], PRIM_INT__s_d_s_s[n * 6 + 3]);
349 PRIM_INT__s_d_s_p[n * 18 + 10] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_d_s_s[(n+1) * 6 + 3], PRIM_INT__s_d_s_p[n * 18 + 10]);
350 PRIM_INT__s_d_s_p[n * 18 + 10] = SIMINT_FMADD( vrr_const_2_over_2pq, PRIM_INT__s_p_s_s[(n+1) * 3 + 1], PRIM_INT__s_d_s_p[n * 18 + 10]);
351
352 PRIM_INT__s_d_s_p[n * 18 + 11] = SIMINT_MUL(Q_PB[2], PRIM_INT__s_d_s_s[n * 6 + 3]);
353 PRIM_INT__s_d_s_p[n * 18 + 11] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_d_s_s[(n+1) * 6 + 3], PRIM_INT__s_d_s_p[n * 18 + 11]);
354
355 PRIM_INT__s_d_s_p[n * 18 + 12] = SIMINT_MUL(Q_PB[0], PRIM_INT__s_d_s_s[n * 6 + 4]);
356 PRIM_INT__s_d_s_p[n * 18 + 12] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_d_s_s[(n+1) * 6 + 4], PRIM_INT__s_d_s_p[n * 18 + 12]);
357
358 PRIM_INT__s_d_s_p[n * 18 + 13] = SIMINT_MUL(Q_PB[1], PRIM_INT__s_d_s_s[n * 6 + 4]);
359 PRIM_INT__s_d_s_p[n * 18 + 13] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_d_s_s[(n+1) * 6 + 4], PRIM_INT__s_d_s_p[n * 18 + 13]);
360 PRIM_INT__s_d_s_p[n * 18 + 13] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_p_s_s[(n+1) * 3 + 2], PRIM_INT__s_d_s_p[n * 18 + 13]);
361
362 PRIM_INT__s_d_s_p[n * 18 + 14] = SIMINT_MUL(Q_PB[2], PRIM_INT__s_d_s_s[n * 6 + 4]);
363 PRIM_INT__s_d_s_p[n * 18 + 14] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_d_s_s[(n+1) * 6 + 4], PRIM_INT__s_d_s_p[n * 18 + 14]);
364 PRIM_INT__s_d_s_p[n * 18 + 14] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_p_s_s[(n+1) * 3 + 1], PRIM_INT__s_d_s_p[n * 18 + 14]);
365
366 PRIM_INT__s_d_s_p[n * 18 + 15] = SIMINT_MUL(Q_PB[0], PRIM_INT__s_d_s_s[n * 6 + 5]);
367 PRIM_INT__s_d_s_p[n * 18 + 15] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_d_s_s[(n+1) * 6 + 5], PRIM_INT__s_d_s_p[n * 18 + 15]);
368
369 PRIM_INT__s_d_s_p[n * 18 + 16] = SIMINT_MUL(Q_PB[1], PRIM_INT__s_d_s_s[n * 6 + 5]);
370 PRIM_INT__s_d_s_p[n * 18 + 16] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_d_s_s[(n+1) * 6 + 5], PRIM_INT__s_d_s_p[n * 18 + 16]);
371
372 PRIM_INT__s_d_s_p[n * 18 + 17] = SIMINT_MUL(Q_PB[2], PRIM_INT__s_d_s_s[n * 6 + 5]);
373 PRIM_INT__s_d_s_p[n * 18 + 17] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_d_s_s[(n+1) * 6 + 5], PRIM_INT__s_d_s_p[n * 18 + 17]);
374 PRIM_INT__s_d_s_p[n * 18 + 17] = SIMINT_FMADD( vrr_const_2_over_2pq, PRIM_INT__s_p_s_s[(n+1) * 3 + 2], PRIM_INT__s_d_s_p[n * 18 + 17]);
375
376 }
377
378
379 VRR_L_s_f_s_d(
380 PRIM_INT__s_f_s_d,
381 PRIM_INT__s_f_s_p,
382 PRIM_INT__s_f_s_s,
383 PRIM_INT__s_d_s_p,
384 Q_PB,
385 a_over_q,
386 aoq_PQ,
387 one_over_2pq,
388 one_over_2q,
389 2);
390
391
392
393 // Forming PRIM_INT__s_p_s_p[3 * 9];
394 for(n = 0; n < 3; ++n) // loop over orders of auxiliary function
395 {
396
397 PRIM_INT__s_p_s_p[n * 9 + 0] = SIMINT_MUL(Q_PB[0], PRIM_INT__s_p_s_s[n * 3 + 0]);
398 PRIM_INT__s_p_s_p[n * 9 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_p_s_s[(n+1) * 3 + 0], PRIM_INT__s_p_s_p[n * 9 + 0]);
399 PRIM_INT__s_p_s_p[n * 9 + 0] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_p_s_p[n * 9 + 0]);
400
401 PRIM_INT__s_p_s_p[n * 9 + 1] = SIMINT_MUL(Q_PB[1], PRIM_INT__s_p_s_s[n * 3 + 0]);
402 PRIM_INT__s_p_s_p[n * 9 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_p_s_s[(n+1) * 3 + 0], PRIM_INT__s_p_s_p[n * 9 + 1]);
403
404 PRIM_INT__s_p_s_p[n * 9 + 2] = SIMINT_MUL(Q_PB[2], PRIM_INT__s_p_s_s[n * 3 + 0]);
405 PRIM_INT__s_p_s_p[n * 9 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_p_s_s[(n+1) * 3 + 0], PRIM_INT__s_p_s_p[n * 9 + 2]);
406
407 PRIM_INT__s_p_s_p[n * 9 + 3] = SIMINT_MUL(Q_PB[0], PRIM_INT__s_p_s_s[n * 3 + 1]);
408 PRIM_INT__s_p_s_p[n * 9 + 3] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_p_s_s[(n+1) * 3 + 1], PRIM_INT__s_p_s_p[n * 9 + 3]);
409
410 PRIM_INT__s_p_s_p[n * 9 + 4] = SIMINT_MUL(Q_PB[1], PRIM_INT__s_p_s_s[n * 3 + 1]);
411 PRIM_INT__s_p_s_p[n * 9 + 4] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_p_s_s[(n+1) * 3 + 1], PRIM_INT__s_p_s_p[n * 9 + 4]);
412 PRIM_INT__s_p_s_p[n * 9 + 4] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_p_s_p[n * 9 + 4]);
413
414 PRIM_INT__s_p_s_p[n * 9 + 5] = SIMINT_MUL(Q_PB[2], PRIM_INT__s_p_s_s[n * 3 + 1]);
415 PRIM_INT__s_p_s_p[n * 9 + 5] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_p_s_s[(n+1) * 3 + 1], PRIM_INT__s_p_s_p[n * 9 + 5]);
416
417 PRIM_INT__s_p_s_p[n * 9 + 6] = SIMINT_MUL(Q_PB[0], PRIM_INT__s_p_s_s[n * 3 + 2]);
418 PRIM_INT__s_p_s_p[n * 9 + 6] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_p_s_s[(n+1) * 3 + 2], PRIM_INT__s_p_s_p[n * 9 + 6]);
419
420 PRIM_INT__s_p_s_p[n * 9 + 7] = SIMINT_MUL(Q_PB[1], PRIM_INT__s_p_s_s[n * 3 + 2]);
421 PRIM_INT__s_p_s_p[n * 9 + 7] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_p_s_s[(n+1) * 3 + 2], PRIM_INT__s_p_s_p[n * 9 + 7]);
422
423 PRIM_INT__s_p_s_p[n * 9 + 8] = SIMINT_MUL(Q_PB[2], PRIM_INT__s_p_s_s[n * 3 + 2]);
424 PRIM_INT__s_p_s_p[n * 9 + 8] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_p_s_s[(n+1) * 3 + 2], PRIM_INT__s_p_s_p[n * 9 + 8]);
425 PRIM_INT__s_p_s_p[n * 9 + 8] = SIMINT_FMADD( vrr_const_1_over_2pq, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_p_s_p[n * 9 + 8]);
426
427 }
428
429
430 VRR_L_s_d_s_d(
431 PRIM_INT__s_d_s_d,
432 PRIM_INT__s_d_s_p,
433 PRIM_INT__s_d_s_s,
434 PRIM_INT__s_p_s_p,
435 Q_PB,
436 a_over_q,
437 aoq_PQ,
438 one_over_2pq,
439 one_over_2q,
440 2);
441
442
443 ostei_general_vrr_L(0, 3, 0, 3, 1,
444 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PB,
445 PRIM_INT__s_f_s_d, NULL, PRIM_INT__s_f_s_p, NULL, PRIM_INT__s_d_s_d, PRIM_INT__s_f_s_f);
446
447
448
449
450 ////////////////////////////////////
451 // Accumulate contracted integrals
452 ////////////////////////////////////
453 if(lastoffset == 0)
454 {
455 contract_all(100, PRIM_INT__s_f_s_f, PRIM_PTR_INT__s_f_s_f);
456 }
457 else
458 {
459 contract(100, shelloffsets, PRIM_INT__s_f_s_f, PRIM_PTR_INT__s_f_s_f);
460 PRIM_PTR_INT__s_f_s_f += lastoffset*100;
461 }
462
463 } // close loop over j
464 } // close loop over i
465
466 //Advance to the next batch
467 jstart = SIMINT_SIMD_ROUND(jend);
468 abcd += nshellbatch;
469
470 } // close loop cdbatch
471
472 istart = iend;
473 } // close loop over ab
474
475 return P.nshell12_clip * Q.nshell12_clip;
476 }
477
478