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_g_d_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_g_d_s)8 int ostei_i_g_d_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_g_d_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__i_g_d_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_d_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__k_s_d_s = work + (SIMINT_NSHELL_SIMD * 168);
30 double * const INT__l_s_d_s = work + (SIMINT_NSHELL_SIMD * 384);
31 double * const INT__m_s_d_s = work + (SIMINT_NSHELL_SIMD * 654);
32 double * const INT__n_s_d_s = work + (SIMINT_NSHELL_SIMD * 984);
33 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*1380);
34 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_s_s = primwork + 13;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 49;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 115;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 215;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 350;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_p_s = primwork + 518;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 644;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_p_s = primwork + 840;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_d_s = primwork + 1008;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_s_s = primwork + 1176;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_p_s = primwork + 1392;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_d_s = primwork + 1608;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_s_s = primwork + 1824;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_p_s = primwork + 2049;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_d_s = primwork + 2319;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_s_s = primwork + 2589;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_p_s = primwork + 2809;
52 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_d_s = primwork + 3139;
53 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_s_s = primwork + 3469;
54 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_p_s = primwork + 3667;
55 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_d_s = primwork + 4063;
56 double * const hrrwork = (double *)(primwork + 4459);
57 double * const HRR_INT__i_p_d_s = hrrwork + 0;
58 double * const HRR_INT__i_d_d_s = hrrwork + 504;
59 double * const HRR_INT__i_f_d_s = hrrwork + 1512;
60 double * const HRR_INT__k_p_d_s = hrrwork + 3192;
61 double * const HRR_INT__k_d_d_s = hrrwork + 3840;
62 double * const HRR_INT__k_f_d_s = hrrwork + 5136;
63 double * const HRR_INT__l_p_d_s = hrrwork + 7296;
64 double * const HRR_INT__l_d_d_s = hrrwork + 8106;
65 double * const HRR_INT__m_p_d_s = hrrwork + 9726;
66
67
68 // Create constants
69 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
70 const SIMINT_DBLTYPE const_10 = SIMINT_DBLSET1(10);
71 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
72 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
73 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
74 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
75 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
76 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
77 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
78 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
79 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
80
81
82 ////////////////////////////////////////
83 // Loop over shells and primitives
84 ////////////////////////////////////////
85
86 real_abcd = 0;
87 istart = 0;
88 for(ab = 0; ab < P.nshell12_clip; ++ab)
89 {
90 const int iend = istart + P.nprim12[ab];
91
92 cd = 0;
93 jstart = 0;
94
95 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
96 {
97 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
98 int jend = jstart;
99 for(i = 0; i < nshellbatch; i++)
100 jend += Q.nprim12[cd+i];
101
102 // Clear the beginning of the workspace (where we are accumulating integrals)
103 memset(work, 0, SIMINT_NSHELL_SIMD * 1380 * sizeof(double));
104 abcd = 0;
105
106
107 for(i = istart; i < iend; ++i)
108 {
109 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
110
111 if(check_screen)
112 {
113 // Skip this whole thing if always insignificant
114 if((P.screen[i] * Q.screen_max) < screen_tol)
115 continue;
116 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
117 }
118
119 icd = 0;
120 iprimcd = 0;
121 nprim_icd = Q.nprim12[cd];
122 double * restrict PRIM_PTR_INT__i_s_d_s = INT__i_s_d_s + abcd * 168;
123 double * restrict PRIM_PTR_INT__k_s_d_s = INT__k_s_d_s + abcd * 216;
124 double * restrict PRIM_PTR_INT__l_s_d_s = INT__l_s_d_s + abcd * 270;
125 double * restrict PRIM_PTR_INT__m_s_d_s = INT__m_s_d_s + abcd * 330;
126 double * restrict PRIM_PTR_INT__n_s_d_s = INT__n_s_d_s + abcd * 396;
127
128
129
130 // Load these one per loop over i
131 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
132 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
133 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
134
135 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
136
137 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
138 {
139 // calculate the shell offsets
140 // these are the offset from the shell pointed to by cd
141 // for each element
142 int shelloffsets[SIMINT_SIMD_LEN] = {0};
143 int lastoffset = 0;
144 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
145
146 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
147 {
148 // Handle if the first element of the vector is a new shell
149 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
150 {
151 nprim_icd += Q.nprim12[cd + (++icd)];
152 PRIM_PTR_INT__i_s_d_s += 168;
153 PRIM_PTR_INT__k_s_d_s += 216;
154 PRIM_PTR_INT__l_s_d_s += 270;
155 PRIM_PTR_INT__m_s_d_s += 330;
156 PRIM_PTR_INT__n_s_d_s += 396;
157 }
158 iprimcd++;
159 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
160 {
161 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
162 {
163 shelloffsets[n] = shelloffsets[n-1] + 1;
164 lastoffset++;
165 nprim_icd += Q.nprim12[cd + (++icd)];
166 }
167 else
168 shelloffsets[n] = shelloffsets[n-1];
169 iprimcd++;
170 }
171 }
172 else
173 iprimcd += SIMINT_SIMD_LEN;
174
175 // Do we have to compute this vector (or has it been screened out)?
176 // (not_screened != 0 means we have to do this vector)
177 if(check_screen)
178 {
179 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
180 if(vmax < screen_tol)
181 {
182 PRIM_PTR_INT__i_s_d_s += lastoffset*168;
183 PRIM_PTR_INT__k_s_d_s += lastoffset*216;
184 PRIM_PTR_INT__l_s_d_s += lastoffset*270;
185 PRIM_PTR_INT__m_s_d_s += lastoffset*330;
186 PRIM_PTR_INT__n_s_d_s += lastoffset*396;
187 continue;
188 }
189 }
190
191 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
192 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
193 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
194 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
195
196
197 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
198 SIMINT_DBLTYPE PQ[3];
199 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
200 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
201 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
202 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
203 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
204 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
205
206 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
207 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
208 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
209 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
210 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
211 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
212 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
213
214 // NOTE: Minus sign!
215 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
216 SIMINT_DBLTYPE aop_PQ[3];
217 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
218 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
219 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
220
221 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
222 SIMINT_DBLTYPE aoq_PQ[3];
223 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
224 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
225 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
226 // Put a minus sign here so we don't have to in RR routines
227 a_over_q = SIMINT_NEG(a_over_q);
228
229
230 //////////////////////////////////////////////
231 // Fjt function section
232 // Maximum v value: 12
233 //////////////////////////////////////////////
234 // The parameter to the Fjt function
235 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
236
237
238 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
239
240
241 boys_F_split(PRIM_INT__s_s_s_s, F_x, 12);
242 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
243 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
244 for(n = 0; n <= 12; n++)
245 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
246
247 //////////////////////////////////////////////
248 // Primitive integrals: Vertical recurrance
249 //////////////////////////////////////////////
250
251 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
252 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
253 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
254 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
255 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
256 const SIMINT_DBLTYPE vrr_const_6_over_2p = SIMINT_MUL(const_6, one_over_2p);
257 const SIMINT_DBLTYPE vrr_const_7_over_2p = SIMINT_MUL(const_7, one_over_2p);
258 const SIMINT_DBLTYPE vrr_const_8_over_2p = SIMINT_MUL(const_8, one_over_2p);
259 const SIMINT_DBLTYPE vrr_const_9_over_2p = SIMINT_MUL(const_9, one_over_2p);
260 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
261 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
262 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
263 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
264 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
265 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
266 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
267 const SIMINT_DBLTYPE vrr_const_7_over_2pq = SIMINT_MUL(const_7, one_over_2pq);
268 const SIMINT_DBLTYPE vrr_const_8_over_2pq = SIMINT_MUL(const_8, one_over_2pq);
269 const SIMINT_DBLTYPE vrr_const_9_over_2pq = SIMINT_MUL(const_9, one_over_2pq);
270 const SIMINT_DBLTYPE vrr_const_10_over_2pq = SIMINT_MUL(const_10, one_over_2pq);
271
272
273
274 // Forming PRIM_INT__p_s_s_s[12 * 3];
275 for(n = 0; n < 12; ++n) // loop over orders of auxiliary function
276 {
277
278 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
279 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]);
280
281 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
282 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]);
283
284 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
285 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]);
286
287 }
288
289
290
291 // Forming PRIM_INT__d_s_s_s[11 * 6];
292 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
293 {
294
295 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
296 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]);
297 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]);
298
299 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
300 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]);
301 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]);
302
303 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
304 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]);
305 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]);
306
307 }
308
309
310
311 // Forming PRIM_INT__f_s_s_s[10 * 10];
312 for(n = 0; n < 10; ++n) // loop over orders of auxiliary function
313 {
314
315 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
316 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]);
317 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]);
318
319 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
320 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]);
321
322 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
323 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]);
324
325 PRIM_INT__f_s_s_s[n * 10 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
326 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]);
327
328 PRIM_INT__f_s_s_s[n * 10 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
329 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]);
330
331 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
332 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]);
333 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]);
334
335 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
336 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]);
337
338 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
339 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]);
340 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]);
341
342 }
343
344
345 VRR_I_g_s_s_s(
346 PRIM_INT__g_s_s_s,
347 PRIM_INT__f_s_s_s,
348 PRIM_INT__d_s_s_s,
349 P_PA,
350 a_over_p,
351 aop_PQ,
352 one_over_2p,
353 9);
354
355
356 VRR_I_h_s_s_s(
357 PRIM_INT__h_s_s_s,
358 PRIM_INT__g_s_s_s,
359 PRIM_INT__f_s_s_s,
360 P_PA,
361 a_over_p,
362 aop_PQ,
363 one_over_2p,
364 8);
365
366
367 ostei_general_vrr1_I(6, 7,
368 one_over_2p, a_over_p, aop_PQ, P_PA,
369 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
370
371
372 ostei_general_vrr_K(6, 0, 1, 0, 2,
373 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
374 PRIM_INT__i_s_s_s, NULL, NULL, PRIM_INT__h_s_s_s, NULL, PRIM_INT__i_s_p_s);
375
376
377 ostei_general_vrr_K(5, 0, 1, 0, 2,
378 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
379 PRIM_INT__h_s_s_s, NULL, NULL, PRIM_INT__g_s_s_s, NULL, PRIM_INT__h_s_p_s);
380
381
382 ostei_general_vrr_K(6, 0, 2, 0, 1,
383 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
384 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);
385
386
387 ostei_general_vrr1_I(7, 6,
388 one_over_2p, a_over_p, aop_PQ, P_PA,
389 PRIM_INT__i_s_s_s, PRIM_INT__h_s_s_s, PRIM_INT__k_s_s_s);
390
391
392 ostei_general_vrr_K(7, 0, 1, 0, 2,
393 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
394 PRIM_INT__k_s_s_s, NULL, NULL, PRIM_INT__i_s_s_s, NULL, PRIM_INT__k_s_p_s);
395
396
397 ostei_general_vrr_K(7, 0, 2, 0, 1,
398 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
399 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);
400
401
402 ostei_general_vrr1_I(8, 5,
403 one_over_2p, a_over_p, aop_PQ, P_PA,
404 PRIM_INT__k_s_s_s, PRIM_INT__i_s_s_s, PRIM_INT__l_s_s_s);
405
406
407 ostei_general_vrr_K(8, 0, 1, 0, 2,
408 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
409 PRIM_INT__l_s_s_s, NULL, NULL, PRIM_INT__k_s_s_s, NULL, PRIM_INT__l_s_p_s);
410
411
412 ostei_general_vrr_K(8, 0, 2, 0, 1,
413 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
414 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);
415
416
417 ostei_general_vrr1_I(9, 4,
418 one_over_2p, a_over_p, aop_PQ, P_PA,
419 PRIM_INT__l_s_s_s, PRIM_INT__k_s_s_s, PRIM_INT__m_s_s_s);
420
421
422 ostei_general_vrr_K(9, 0, 1, 0, 2,
423 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
424 PRIM_INT__m_s_s_s, NULL, NULL, PRIM_INT__l_s_s_s, NULL, PRIM_INT__m_s_p_s);
425
426
427 ostei_general_vrr_K(9, 0, 2, 0, 1,
428 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
429 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);
430
431
432 ostei_general_vrr1_I(10, 3,
433 one_over_2p, a_over_p, aop_PQ, P_PA,
434 PRIM_INT__m_s_s_s, PRIM_INT__l_s_s_s, PRIM_INT__n_s_s_s);
435
436
437 ostei_general_vrr_K(10, 0, 1, 0, 2,
438 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
439 PRIM_INT__n_s_s_s, NULL, NULL, PRIM_INT__m_s_s_s, NULL, PRIM_INT__n_s_p_s);
440
441
442 ostei_general_vrr_K(10, 0, 2, 0, 1,
443 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
444 PRIM_INT__n_s_p_s, PRIM_INT__n_s_s_s, NULL, PRIM_INT__m_s_p_s, NULL, PRIM_INT__n_s_d_s);
445
446
447
448
449 ////////////////////////////////////
450 // Accumulate contracted integrals
451 ////////////////////////////////////
452 if(lastoffset == 0)
453 {
454 contract_all(168, PRIM_INT__i_s_d_s, PRIM_PTR_INT__i_s_d_s);
455 contract_all(216, PRIM_INT__k_s_d_s, PRIM_PTR_INT__k_s_d_s);
456 contract_all(270, PRIM_INT__l_s_d_s, PRIM_PTR_INT__l_s_d_s);
457 contract_all(330, PRIM_INT__m_s_d_s, PRIM_PTR_INT__m_s_d_s);
458 contract_all(396, PRIM_INT__n_s_d_s, PRIM_PTR_INT__n_s_d_s);
459 }
460 else
461 {
462 contract(168, shelloffsets, PRIM_INT__i_s_d_s, PRIM_PTR_INT__i_s_d_s);
463 contract(216, shelloffsets, PRIM_INT__k_s_d_s, PRIM_PTR_INT__k_s_d_s);
464 contract(270, shelloffsets, PRIM_INT__l_s_d_s, PRIM_PTR_INT__l_s_d_s);
465 contract(330, shelloffsets, PRIM_INT__m_s_d_s, PRIM_PTR_INT__m_s_d_s);
466 contract(396, shelloffsets, PRIM_INT__n_s_d_s, PRIM_PTR_INT__n_s_d_s);
467 PRIM_PTR_INT__i_s_d_s += lastoffset*168;
468 PRIM_PTR_INT__k_s_d_s += lastoffset*216;
469 PRIM_PTR_INT__l_s_d_s += lastoffset*270;
470 PRIM_PTR_INT__m_s_d_s += lastoffset*330;
471 PRIM_PTR_INT__n_s_d_s += lastoffset*396;
472 }
473
474 } // close loop over j
475 } // close loop over i
476
477 //Advance to the next batch
478 jstart = SIMINT_SIMD_ROUND(jend);
479
480 //////////////////////////////////////////////
481 // Contracted integrals: Horizontal recurrance
482 //////////////////////////////////////////////
483
484
485 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
486
487
488 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
489 {
490
491 // set up HRR pointers
492 double const * restrict HRR_INT__i_s_d_s = INT__i_s_d_s + abcd * 168;
493 double const * restrict HRR_INT__k_s_d_s = INT__k_s_d_s + abcd * 216;
494 double const * restrict HRR_INT__l_s_d_s = INT__l_s_d_s + abcd * 270;
495 double const * restrict HRR_INT__m_s_d_s = INT__m_s_d_s + abcd * 330;
496 double const * restrict HRR_INT__n_s_d_s = INT__n_s_d_s + abcd * 396;
497 double * restrict HRR_INT__i_g_d_s = INT__i_g_d_s + real_abcd * 2520;
498
499 // form INT__i_p_d_s
500 ostei_general_hrr_J(6, 1, 2, 0, hAB, HRR_INT__k_s_d_s, HRR_INT__i_s_d_s, HRR_INT__i_p_d_s);
501
502 // form INT__k_p_d_s
503 ostei_general_hrr_J(7, 1, 2, 0, hAB, HRR_INT__l_s_d_s, HRR_INT__k_s_d_s, HRR_INT__k_p_d_s);
504
505 // form INT__l_p_d_s
506 ostei_general_hrr_J(8, 1, 2, 0, hAB, HRR_INT__m_s_d_s, HRR_INT__l_s_d_s, HRR_INT__l_p_d_s);
507
508 // form INT__m_p_d_s
509 ostei_general_hrr_J(9, 1, 2, 0, hAB, HRR_INT__n_s_d_s, HRR_INT__m_s_d_s, HRR_INT__m_p_d_s);
510
511 // form INT__i_d_d_s
512 ostei_general_hrr_J(6, 2, 2, 0, hAB, HRR_INT__k_p_d_s, HRR_INT__i_p_d_s, HRR_INT__i_d_d_s);
513
514 // form INT__k_d_d_s
515 ostei_general_hrr_J(7, 2, 2, 0, hAB, HRR_INT__l_p_d_s, HRR_INT__k_p_d_s, HRR_INT__k_d_d_s);
516
517 // form INT__l_d_d_s
518 ostei_general_hrr_J(8, 2, 2, 0, hAB, HRR_INT__m_p_d_s, HRR_INT__l_p_d_s, HRR_INT__l_d_d_s);
519
520 // form INT__i_f_d_s
521 ostei_general_hrr_J(6, 3, 2, 0, hAB, HRR_INT__k_d_d_s, HRR_INT__i_d_d_s, HRR_INT__i_f_d_s);
522
523 // form INT__k_f_d_s
524 ostei_general_hrr_J(7, 3, 2, 0, hAB, HRR_INT__l_d_d_s, HRR_INT__k_d_d_s, HRR_INT__k_f_d_s);
525
526 // form INT__i_g_d_s
527 ostei_general_hrr_J(6, 4, 2, 0, hAB, HRR_INT__k_f_d_s, HRR_INT__i_f_d_s, HRR_INT__i_g_d_s);
528
529
530 } // close HRR loop
531
532
533 } // close loop cdbatch
534
535 istart = iend;
536 } // close loop over ab
537
538 return P.nshell12_clip * Q.nshell12_clip;
539 }
540
ostei_g_i_d_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__g_i_d_s)541 int ostei_g_i_d_s(struct simint_multi_shellpair const P,
542 struct simint_multi_shellpair const Q,
543 double screen_tol,
544 double * const restrict work,
545 double * const restrict INT__g_i_d_s)
546 {
547 double P_AB[3*P.nshell12];
548 struct simint_multi_shellpair P_tmp = P;
549 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
550 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
551 P_tmp.AB_x = P_AB;
552 P_tmp.AB_y = P_AB + P.nshell12;
553 P_tmp.AB_z = P_AB + 2*P.nshell12;
554
555 for(int i = 0; i < P.nshell12; i++)
556 {
557 P_tmp.AB_x[i] = -P.AB_x[i];
558 P_tmp.AB_y[i] = -P.AB_y[i];
559 P_tmp.AB_z[i] = -P.AB_z[i];
560 }
561
562 int ret = ostei_i_g_d_s(P_tmp, Q, screen_tol, work, INT__g_i_d_s);
563 double buffer[2520] SIMINT_ALIGN_ARRAY_DBL;
564
565 for(int q = 0; q < ret; q++)
566 {
567 int idx = 0;
568 for(int a = 0; a < 15; ++a)
569 for(int b = 0; b < 28; ++b)
570 for(int c = 0; c < 6; ++c)
571 for(int d = 0; d < 1; ++d)
572 buffer[idx++] = INT__g_i_d_s[q*2520+b*90+a*6+c*1+d];
573
574 memcpy(INT__g_i_d_s+q*2520, buffer, 2520*sizeof(double));
575 }
576
577 return ret;
578 }
579
ostei_i_g_s_d(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__i_g_s_d)580 int ostei_i_g_s_d(struct simint_multi_shellpair const P,
581 struct simint_multi_shellpair const Q,
582 double screen_tol,
583 double * const restrict work,
584 double * const restrict INT__i_g_s_d)
585 {
586 double Q_AB[3*Q.nshell12];
587 struct simint_multi_shellpair Q_tmp = Q;
588 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
589 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
590 Q_tmp.AB_x = Q_AB;
591 Q_tmp.AB_y = Q_AB + Q.nshell12;
592 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
593
594 for(int i = 0; i < Q.nshell12; i++)
595 {
596 Q_tmp.AB_x[i] = -Q.AB_x[i];
597 Q_tmp.AB_y[i] = -Q.AB_y[i];
598 Q_tmp.AB_z[i] = -Q.AB_z[i];
599 }
600
601 int ret = ostei_i_g_d_s(P, Q_tmp, screen_tol, work, INT__i_g_s_d);
602 double buffer[2520] SIMINT_ALIGN_ARRAY_DBL;
603
604 for(int q = 0; q < ret; q++)
605 {
606 int idx = 0;
607 for(int a = 0; a < 28; ++a)
608 for(int b = 0; b < 15; ++b)
609 for(int c = 0; c < 1; ++c)
610 for(int d = 0; d < 6; ++d)
611 buffer[idx++] = INT__i_g_s_d[q*2520+a*90+b*6+d*1+c];
612
613 memcpy(INT__i_g_s_d+q*2520, buffer, 2520*sizeof(double));
614 }
615
616 return ret;
617 }
618
ostei_g_i_s_d(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__g_i_s_d)619 int ostei_g_i_s_d(struct simint_multi_shellpair const P,
620 struct simint_multi_shellpair const Q,
621 double screen_tol,
622 double * const restrict work,
623 double * const restrict INT__g_i_s_d)
624 {
625 double P_AB[3*P.nshell12];
626 struct simint_multi_shellpair P_tmp = P;
627 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
628 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
629 P_tmp.AB_x = P_AB;
630 P_tmp.AB_y = P_AB + P.nshell12;
631 P_tmp.AB_z = P_AB + 2*P.nshell12;
632
633 for(int i = 0; i < P.nshell12; i++)
634 {
635 P_tmp.AB_x[i] = -P.AB_x[i];
636 P_tmp.AB_y[i] = -P.AB_y[i];
637 P_tmp.AB_z[i] = -P.AB_z[i];
638 }
639
640 double Q_AB[3*Q.nshell12];
641 struct simint_multi_shellpair Q_tmp = Q;
642 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
643 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
644 Q_tmp.AB_x = Q_AB;
645 Q_tmp.AB_y = Q_AB + Q.nshell12;
646 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
647
648 for(int i = 0; i < Q.nshell12; i++)
649 {
650 Q_tmp.AB_x[i] = -Q.AB_x[i];
651 Q_tmp.AB_y[i] = -Q.AB_y[i];
652 Q_tmp.AB_z[i] = -Q.AB_z[i];
653 }
654
655 int ret = ostei_i_g_d_s(P_tmp, Q_tmp, screen_tol, work, INT__g_i_s_d);
656 double buffer[2520] SIMINT_ALIGN_ARRAY_DBL;
657
658 for(int q = 0; q < ret; q++)
659 {
660 int idx = 0;
661 for(int a = 0; a < 15; ++a)
662 for(int b = 0; b < 28; ++b)
663 for(int c = 0; c < 1; ++c)
664 for(int d = 0; d < 6; ++d)
665 buffer[idx++] = INT__g_i_s_d[q*2520+b*90+a*6+d*1+c];
666
667 memcpy(INT__g_i_s_d+q*2520, buffer, 2520*sizeof(double));
668 }
669
670 return ret;
671 }
672
673