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_i_f_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__i_f_f_s)8 int ostei_i_f_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__i_f_f_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__i_f_f_s);
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 iket;
26
27 // partition workspace
28 double * const INT__i_s_f_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__k_s_f_s = work + (SIMINT_NSHELL_SIMD * 280);
30 double * const INT__l_s_f_s = work + (SIMINT_NSHELL_SIMD * 640);
31 double * const INT__m_s_f_s = work + (SIMINT_NSHELL_SIMD * 1090);
32 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*1640);
33 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_s_s = primwork + 13;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 49;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 115;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 215;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_p_s = primwork + 350;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 485;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_p_s = primwork + 653;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_d_s = primwork + 842;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 1094;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_p_s = primwork + 1290;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_d_s = primwork + 1542;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_f_s = primwork + 1878;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_s_s = primwork + 2158;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_p_s = primwork + 2374;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_d_s = primwork + 2698;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_f_s = primwork + 3130;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_s_s = primwork + 3490;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_p_s = primwork + 3715;
52 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_d_s = primwork + 4120;
53 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_f_s = primwork + 4660;
54 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_s_s = primwork + 5110;
55 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_p_s = primwork + 5330;
56 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_d_s = primwork + 5825;
57 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_f_s = primwork + 6485;
58 double * const hrrwork = (double *)(primwork + 7035);
59 double * const HRR_INT__i_p_f_s = hrrwork + 0;
60 double * const HRR_INT__i_d_f_s = hrrwork + 840;
61 double * const HRR_INT__k_p_f_s = hrrwork + 2520;
62 double * const HRR_INT__k_d_f_s = hrrwork + 3600;
63 double * const HRR_INT__l_p_f_s = hrrwork + 5760;
64
65
66 // Create constants
67 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
68 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
69 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
70 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
71 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
72 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
73 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
74 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
75 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
76 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
77
78
79 ////////////////////////////////////////
80 // Loop over shells and primitives
81 ////////////////////////////////////////
82
83 real_abcd = 0;
84 istart = 0;
85 for(ab = 0; ab < P.nshell12_clip; ++ab)
86 {
87 const int iend = istart + P.nprim12[ab];
88
89 cd = 0;
90 jstart = 0;
91
92 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
93 {
94 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
95 int jend = jstart;
96 for(i = 0; i < nshellbatch; i++)
97 jend += Q.nprim12[cd+i];
98
99 // Clear the beginning of the workspace (where we are accumulating integrals)
100 memset(work, 0, SIMINT_NSHELL_SIMD * 1640 * sizeof(double));
101 abcd = 0;
102
103
104 for(i = istart; i < iend; ++i)
105 {
106 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
107
108 if(check_screen)
109 {
110 // Skip this whole thing if always insignificant
111 if((P.screen[i] * Q.screen_max) < screen_tol)
112 continue;
113 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
114 }
115
116 icd = 0;
117 iprimcd = 0;
118 nprim_icd = Q.nprim12[cd];
119 double * restrict PRIM_PTR_INT__i_s_f_s = INT__i_s_f_s + abcd * 280;
120 double * restrict PRIM_PTR_INT__k_s_f_s = INT__k_s_f_s + abcd * 360;
121 double * restrict PRIM_PTR_INT__l_s_f_s = INT__l_s_f_s + abcd * 450;
122 double * restrict PRIM_PTR_INT__m_s_f_s = INT__m_s_f_s + abcd * 550;
123
124
125
126 // Load these one per loop over i
127 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
128 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
129 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
130
131 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
132
133 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
134 {
135 // calculate the shell offsets
136 // these are the offset from the shell pointed to by cd
137 // for each element
138 int shelloffsets[SIMINT_SIMD_LEN] = {0};
139 int lastoffset = 0;
140 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
141
142 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
143 {
144 // Handle if the first element of the vector is a new shell
145 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
146 {
147 nprim_icd += Q.nprim12[cd + (++icd)];
148 PRIM_PTR_INT__i_s_f_s += 280;
149 PRIM_PTR_INT__k_s_f_s += 360;
150 PRIM_PTR_INT__l_s_f_s += 450;
151 PRIM_PTR_INT__m_s_f_s += 550;
152 }
153 iprimcd++;
154 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
155 {
156 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
157 {
158 shelloffsets[n] = shelloffsets[n-1] + 1;
159 lastoffset++;
160 nprim_icd += Q.nprim12[cd + (++icd)];
161 }
162 else
163 shelloffsets[n] = shelloffsets[n-1];
164 iprimcd++;
165 }
166 }
167 else
168 iprimcd += SIMINT_SIMD_LEN;
169
170 // Do we have to compute this vector (or has it been screened out)?
171 // (not_screened != 0 means we have to do this vector)
172 if(check_screen)
173 {
174 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
175 if(vmax < screen_tol)
176 {
177 PRIM_PTR_INT__i_s_f_s += lastoffset*280;
178 PRIM_PTR_INT__k_s_f_s += lastoffset*360;
179 PRIM_PTR_INT__l_s_f_s += lastoffset*450;
180 PRIM_PTR_INT__m_s_f_s += lastoffset*550;
181 continue;
182 }
183 }
184
185 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
186 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
187 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
188 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
189
190
191 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
192 SIMINT_DBLTYPE PQ[3];
193 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
194 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
195 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
196 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
197 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
198 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
199
200 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
201 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
202 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
203 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
204 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
205 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
206 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
207
208 // NOTE: Minus sign!
209 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
210 SIMINT_DBLTYPE aop_PQ[3];
211 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
212 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
213 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
214
215 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
216 SIMINT_DBLTYPE aoq_PQ[3];
217 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
218 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
219 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
220 // Put a minus sign here so we don't have to in RR routines
221 a_over_q = SIMINT_NEG(a_over_q);
222
223
224 //////////////////////////////////////////////
225 // Fjt function section
226 // Maximum v value: 12
227 //////////////////////////////////////////////
228 // The parameter to the Fjt function
229 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
230
231
232 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
233
234
235 boys_F_split(PRIM_INT__s_s_s_s, F_x, 12);
236 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
237 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
238 for(n = 0; n <= 12; n++)
239 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
240
241 //////////////////////////////////////////////
242 // Primitive integrals: Vertical recurrance
243 //////////////////////////////////////////////
244
245 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
246 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
247 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
248 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
249 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
250 const SIMINT_DBLTYPE vrr_const_6_over_2p = SIMINT_MUL(const_6, one_over_2p);
251 const SIMINT_DBLTYPE vrr_const_7_over_2p = SIMINT_MUL(const_7, one_over_2p);
252 const SIMINT_DBLTYPE vrr_const_8_over_2p = SIMINT_MUL(const_8, one_over_2p);
253 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
254 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
255 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
256 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
257 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
258 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
259 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
260 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
261 const SIMINT_DBLTYPE vrr_const_7_over_2pq = SIMINT_MUL(const_7, one_over_2pq);
262 const SIMINT_DBLTYPE vrr_const_8_over_2pq = SIMINT_MUL(const_8, one_over_2pq);
263 const SIMINT_DBLTYPE vrr_const_9_over_2pq = SIMINT_MUL(const_9, one_over_2pq);
264
265
266
267 // Forming PRIM_INT__p_s_s_s[12 * 3];
268 for(n = 0; n < 12; ++n) // loop over orders of auxiliary function
269 {
270
271 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
272 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]);
273
274 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
275 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]);
276
277 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
278 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]);
279
280 }
281
282
283
284 // Forming PRIM_INT__d_s_s_s[11 * 6];
285 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
286 {
287
288 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
289 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]);
290 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]);
291
292 PRIM_INT__d_s_s_s[n * 6 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 0]);
293 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]);
294
295 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
296 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]);
297 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]);
298
299 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
300 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]);
301 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]);
302
303 }
304
305
306
307 // Forming PRIM_INT__f_s_s_s[10 * 10];
308 for(n = 0; n < 10; ++n) // loop over orders of auxiliary function
309 {
310
311 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
312 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]);
313 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]);
314
315 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
316 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]);
317
318 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
319 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]);
320
321 PRIM_INT__f_s_s_s[n * 10 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
322 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]);
323
324 PRIM_INT__f_s_s_s[n * 10 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 1]);
325 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]);
326
327 PRIM_INT__f_s_s_s[n * 10 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
328 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]);
329
330 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
331 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]);
332 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]);
333
334 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
335 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]);
336
337 PRIM_INT__f_s_s_s[n * 10 + 8] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 5]);
338 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]);
339
340 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
341 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]);
342 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]);
343
344 }
345
346
347 VRR_I_g_s_s_s(
348 PRIM_INT__g_s_s_s,
349 PRIM_INT__f_s_s_s,
350 PRIM_INT__d_s_s_s,
351 P_PA,
352 a_over_p,
353 aop_PQ,
354 one_over_2p,
355 9);
356
357
358 VRR_I_h_s_s_s(
359 PRIM_INT__h_s_s_s,
360 PRIM_INT__g_s_s_s,
361 PRIM_INT__f_s_s_s,
362 P_PA,
363 a_over_p,
364 aop_PQ,
365 one_over_2p,
366 8);
367
368
369 ostei_general_vrr1_I(6, 7,
370 one_over_2p, a_over_p, aop_PQ, P_PA,
371 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
372
373
374 ostei_general_vrr_K(6, 0, 1, 0, 3,
375 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
376 PRIM_INT__i_s_s_s, NULL, NULL, PRIM_INT__h_s_s_s, NULL, PRIM_INT__i_s_p_s);
377
378
379 ostei_general_vrr_K(5, 0, 1, 0, 3,
380 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
381 PRIM_INT__h_s_s_s, NULL, NULL, PRIM_INT__g_s_s_s, NULL, PRIM_INT__h_s_p_s);
382
383
384 ostei_general_vrr_K(6, 0, 2, 0, 2,
385 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
386 PRIM_INT__i_s_p_s, PRIM_INT__i_s_s_s, NULL, PRIM_INT__h_s_p_s, NULL, PRIM_INT__i_s_d_s);
387
388
389 VRR_K_g_s_p_s(
390 PRIM_INT__g_s_p_s,
391 PRIM_INT__g_s_s_s,
392 PRIM_INT__f_s_s_s,
393 Q_PA,
394 aoq_PQ,
395 one_over_2pq,
396 3);
397
398
399 ostei_general_vrr_K(5, 0, 2, 0, 2,
400 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
401 PRIM_INT__h_s_p_s, PRIM_INT__h_s_s_s, NULL, PRIM_INT__g_s_p_s, NULL, PRIM_INT__h_s_d_s);
402
403
404 ostei_general_vrr_K(6, 0, 3, 0, 1,
405 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
406 PRIM_INT__i_s_d_s, PRIM_INT__i_s_p_s, NULL, PRIM_INT__h_s_d_s, NULL, PRIM_INT__i_s_f_s);
407
408
409 ostei_general_vrr1_I(7, 6,
410 one_over_2p, a_over_p, aop_PQ, P_PA,
411 PRIM_INT__i_s_s_s, PRIM_INT__h_s_s_s, PRIM_INT__k_s_s_s);
412
413
414 ostei_general_vrr_K(7, 0, 1, 0, 3,
415 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
416 PRIM_INT__k_s_s_s, NULL, NULL, PRIM_INT__i_s_s_s, NULL, PRIM_INT__k_s_p_s);
417
418
419 ostei_general_vrr_K(7, 0, 2, 0, 2,
420 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
421 PRIM_INT__k_s_p_s, PRIM_INT__k_s_s_s, NULL, PRIM_INT__i_s_p_s, NULL, PRIM_INT__k_s_d_s);
422
423
424 ostei_general_vrr_K(7, 0, 3, 0, 1,
425 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
426 PRIM_INT__k_s_d_s, PRIM_INT__k_s_p_s, NULL, PRIM_INT__i_s_d_s, NULL, PRIM_INT__k_s_f_s);
427
428
429 ostei_general_vrr1_I(8, 5,
430 one_over_2p, a_over_p, aop_PQ, P_PA,
431 PRIM_INT__k_s_s_s, PRIM_INT__i_s_s_s, PRIM_INT__l_s_s_s);
432
433
434 ostei_general_vrr_K(8, 0, 1, 0, 3,
435 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
436 PRIM_INT__l_s_s_s, NULL, NULL, PRIM_INT__k_s_s_s, NULL, PRIM_INT__l_s_p_s);
437
438
439 ostei_general_vrr_K(8, 0, 2, 0, 2,
440 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
441 PRIM_INT__l_s_p_s, PRIM_INT__l_s_s_s, NULL, PRIM_INT__k_s_p_s, NULL, PRIM_INT__l_s_d_s);
442
443
444 ostei_general_vrr_K(8, 0, 3, 0, 1,
445 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
446 PRIM_INT__l_s_d_s, PRIM_INT__l_s_p_s, NULL, PRIM_INT__k_s_d_s, NULL, PRIM_INT__l_s_f_s);
447
448
449 ostei_general_vrr1_I(9, 4,
450 one_over_2p, a_over_p, aop_PQ, P_PA,
451 PRIM_INT__l_s_s_s, PRIM_INT__k_s_s_s, PRIM_INT__m_s_s_s);
452
453
454 ostei_general_vrr_K(9, 0, 1, 0, 3,
455 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
456 PRIM_INT__m_s_s_s, NULL, NULL, PRIM_INT__l_s_s_s, NULL, PRIM_INT__m_s_p_s);
457
458
459 ostei_general_vrr_K(9, 0, 2, 0, 2,
460 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
461 PRIM_INT__m_s_p_s, PRIM_INT__m_s_s_s, NULL, PRIM_INT__l_s_p_s, NULL, PRIM_INT__m_s_d_s);
462
463
464 ostei_general_vrr_K(9, 0, 3, 0, 1,
465 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
466 PRIM_INT__m_s_d_s, PRIM_INT__m_s_p_s, NULL, PRIM_INT__l_s_d_s, NULL, PRIM_INT__m_s_f_s);
467
468
469
470
471 ////////////////////////////////////
472 // Accumulate contracted integrals
473 ////////////////////////////////////
474 if(lastoffset == 0)
475 {
476 contract_all(280, PRIM_INT__i_s_f_s, PRIM_PTR_INT__i_s_f_s);
477 contract_all(360, PRIM_INT__k_s_f_s, PRIM_PTR_INT__k_s_f_s);
478 contract_all(450, PRIM_INT__l_s_f_s, PRIM_PTR_INT__l_s_f_s);
479 contract_all(550, PRIM_INT__m_s_f_s, PRIM_PTR_INT__m_s_f_s);
480 }
481 else
482 {
483 contract(280, shelloffsets, PRIM_INT__i_s_f_s, PRIM_PTR_INT__i_s_f_s);
484 contract(360, shelloffsets, PRIM_INT__k_s_f_s, PRIM_PTR_INT__k_s_f_s);
485 contract(450, shelloffsets, PRIM_INT__l_s_f_s, PRIM_PTR_INT__l_s_f_s);
486 contract(550, shelloffsets, PRIM_INT__m_s_f_s, PRIM_PTR_INT__m_s_f_s);
487 PRIM_PTR_INT__i_s_f_s += lastoffset*280;
488 PRIM_PTR_INT__k_s_f_s += lastoffset*360;
489 PRIM_PTR_INT__l_s_f_s += lastoffset*450;
490 PRIM_PTR_INT__m_s_f_s += lastoffset*550;
491 }
492
493 } // close loop over j
494 } // close loop over i
495
496 //Advance to the next batch
497 jstart = SIMINT_SIMD_ROUND(jend);
498
499 //////////////////////////////////////////////
500 // Contracted integrals: Horizontal recurrance
501 //////////////////////////////////////////////
502
503
504 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
505
506
507 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
508 {
509
510 // set up HRR pointers
511 double const * restrict HRR_INT__i_s_f_s = INT__i_s_f_s + abcd * 280;
512 double const * restrict HRR_INT__k_s_f_s = INT__k_s_f_s + abcd * 360;
513 double const * restrict HRR_INT__l_s_f_s = INT__l_s_f_s + abcd * 450;
514 double const * restrict HRR_INT__m_s_f_s = INT__m_s_f_s + abcd * 550;
515 double * restrict HRR_INT__i_f_f_s = INT__i_f_f_s + real_abcd * 2800;
516
517 // form INT__i_p_f_s
518 ostei_general_hrr_J(6, 1, 3, 0, hAB, HRR_INT__k_s_f_s, HRR_INT__i_s_f_s, HRR_INT__i_p_f_s);
519
520 // form INT__k_p_f_s
521 ostei_general_hrr_J(7, 1, 3, 0, hAB, HRR_INT__l_s_f_s, HRR_INT__k_s_f_s, HRR_INT__k_p_f_s);
522
523 // form INT__l_p_f_s
524 ostei_general_hrr_J(8, 1, 3, 0, hAB, HRR_INT__m_s_f_s, HRR_INT__l_s_f_s, HRR_INT__l_p_f_s);
525
526 // form INT__i_d_f_s
527 ostei_general_hrr_J(6, 2, 3, 0, hAB, HRR_INT__k_p_f_s, HRR_INT__i_p_f_s, HRR_INT__i_d_f_s);
528
529 // form INT__k_d_f_s
530 ostei_general_hrr_J(7, 2, 3, 0, hAB, HRR_INT__l_p_f_s, HRR_INT__k_p_f_s, HRR_INT__k_d_f_s);
531
532 // form INT__i_f_f_s
533 ostei_general_hrr_J(6, 3, 3, 0, hAB, HRR_INT__k_d_f_s, HRR_INT__i_d_f_s, HRR_INT__i_f_f_s);
534
535
536 } // close HRR loop
537
538
539 } // close loop cdbatch
540
541 istart = iend;
542 } // close loop over ab
543
544 return P.nshell12_clip * Q.nshell12_clip;
545 }
546
ostei_f_i_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__f_i_f_s)547 int ostei_f_i_f_s(struct simint_multi_shellpair const P,
548 struct simint_multi_shellpair const Q,
549 double screen_tol,
550 double * const restrict work,
551 double * const restrict INT__f_i_f_s)
552 {
553 double P_AB[3*P.nshell12];
554 struct simint_multi_shellpair P_tmp = P;
555 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
556 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
557 P_tmp.AB_x = P_AB;
558 P_tmp.AB_y = P_AB + P.nshell12;
559 P_tmp.AB_z = P_AB + 2*P.nshell12;
560
561 for(int i = 0; i < P.nshell12; i++)
562 {
563 P_tmp.AB_x[i] = -P.AB_x[i];
564 P_tmp.AB_y[i] = -P.AB_y[i];
565 P_tmp.AB_z[i] = -P.AB_z[i];
566 }
567
568 int ret = ostei_i_f_f_s(P_tmp, Q, screen_tol, work, INT__f_i_f_s);
569 double buffer[2800] SIMINT_ALIGN_ARRAY_DBL;
570
571 for(int q = 0; q < ret; q++)
572 {
573 int idx = 0;
574 for(int a = 0; a < 10; ++a)
575 for(int b = 0; b < 28; ++b)
576 for(int c = 0; c < 10; ++c)
577 for(int d = 0; d < 1; ++d)
578 buffer[idx++] = INT__f_i_f_s[q*2800+b*100+a*10+c*1+d];
579
580 memcpy(INT__f_i_f_s+q*2800, buffer, 2800*sizeof(double));
581 }
582
583 return ret;
584 }
585
ostei_i_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__i_f_s_f)586 int ostei_i_f_s_f(struct simint_multi_shellpair const P,
587 struct simint_multi_shellpair const Q,
588 double screen_tol,
589 double * const restrict work,
590 double * const restrict INT__i_f_s_f)
591 {
592 double Q_AB[3*Q.nshell12];
593 struct simint_multi_shellpair Q_tmp = Q;
594 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
595 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
596 Q_tmp.AB_x = Q_AB;
597 Q_tmp.AB_y = Q_AB + Q.nshell12;
598 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
599
600 for(int i = 0; i < Q.nshell12; i++)
601 {
602 Q_tmp.AB_x[i] = -Q.AB_x[i];
603 Q_tmp.AB_y[i] = -Q.AB_y[i];
604 Q_tmp.AB_z[i] = -Q.AB_z[i];
605 }
606
607 int ret = ostei_i_f_f_s(P, Q_tmp, screen_tol, work, INT__i_f_s_f);
608 double buffer[2800] SIMINT_ALIGN_ARRAY_DBL;
609
610 for(int q = 0; q < ret; q++)
611 {
612 int idx = 0;
613 for(int a = 0; a < 28; ++a)
614 for(int b = 0; b < 10; ++b)
615 for(int c = 0; c < 1; ++c)
616 for(int d = 0; d < 10; ++d)
617 buffer[idx++] = INT__i_f_s_f[q*2800+a*100+b*10+d*1+c];
618
619 memcpy(INT__i_f_s_f+q*2800, buffer, 2800*sizeof(double));
620 }
621
622 return ret;
623 }
624
ostei_f_i_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__f_i_s_f)625 int ostei_f_i_s_f(struct simint_multi_shellpair const P,
626 struct simint_multi_shellpair const Q,
627 double screen_tol,
628 double * const restrict work,
629 double * const restrict INT__f_i_s_f)
630 {
631 double P_AB[3*P.nshell12];
632 struct simint_multi_shellpair P_tmp = P;
633 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
634 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
635 P_tmp.AB_x = P_AB;
636 P_tmp.AB_y = P_AB + P.nshell12;
637 P_tmp.AB_z = P_AB + 2*P.nshell12;
638
639 for(int i = 0; i < P.nshell12; i++)
640 {
641 P_tmp.AB_x[i] = -P.AB_x[i];
642 P_tmp.AB_y[i] = -P.AB_y[i];
643 P_tmp.AB_z[i] = -P.AB_z[i];
644 }
645
646 double Q_AB[3*Q.nshell12];
647 struct simint_multi_shellpair Q_tmp = Q;
648 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
649 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
650 Q_tmp.AB_x = Q_AB;
651 Q_tmp.AB_y = Q_AB + Q.nshell12;
652 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
653
654 for(int i = 0; i < Q.nshell12; i++)
655 {
656 Q_tmp.AB_x[i] = -Q.AB_x[i];
657 Q_tmp.AB_y[i] = -Q.AB_y[i];
658 Q_tmp.AB_z[i] = -Q.AB_z[i];
659 }
660
661 int ret = ostei_i_f_f_s(P_tmp, Q_tmp, screen_tol, work, INT__f_i_s_f);
662 double buffer[2800] SIMINT_ALIGN_ARRAY_DBL;
663
664 for(int q = 0; q < ret; q++)
665 {
666 int idx = 0;
667 for(int a = 0; a < 10; ++a)
668 for(int b = 0; b < 28; ++b)
669 for(int c = 0; c < 1; ++c)
670 for(int d = 0; d < 10; ++d)
671 buffer[idx++] = INT__f_i_s_f[q*2800+b*100+a*10+d*1+c];
672
673 memcpy(INT__f_i_s_f+q*2800, buffer, 2800*sizeof(double));
674 }
675
676 return ret;
677 }
678
679