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_k_f_g_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__k_f_g_s)8 int ostei_k_f_g_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__k_f_g_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__k_f_g_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__k_s_g_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__l_s_g_s = work + (SIMINT_NSHELL_SIMD * 540);
30 double * const INT__m_s_g_s = work + (SIMINT_NSHELL_SIMD * 1215);
31 double * const INT__n_s_g_s = work + (SIMINT_NSHELL_SIMD * 2040);
32 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*3030);
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 + 15;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 57;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 135;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 255;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_p_s = primwork + 420;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 600;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_p_s = primwork + 810;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_d_s = primwork + 1062;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 1440;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_p_s = primwork + 1692;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_d_s = primwork + 2028;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_f_s = primwork + 2532;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_s_s = primwork + 3092;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_p_s = primwork + 3380;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_d_s = primwork + 3812;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_f_s = primwork + 4460;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_g_s = primwork + 5180;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_s_s = primwork + 5720;
52 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_p_s = primwork + 6035;
53 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_d_s = primwork + 6575;
54 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_f_s = primwork + 7385;
55 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_g_s = primwork + 8285;
56 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_s_s = primwork + 8960;
57 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_p_s = primwork + 9290;
58 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_d_s = primwork + 9950;
59 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_f_s = primwork + 10940;
60 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_g_s = primwork + 12040;
61 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_s_s = primwork + 12865;
62 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_p_s = primwork + 13195;
63 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_d_s = primwork + 13987;
64 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_f_s = primwork + 15175;
65 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_g_s = primwork + 16495;
66 double * const hrrwork = (double *)(primwork + 17485);
67 double * const HRR_INT__k_p_g_s = hrrwork + 0;
68 double * const HRR_INT__k_d_g_s = hrrwork + 1620;
69 double * const HRR_INT__l_p_g_s = hrrwork + 4860;
70 double * const HRR_INT__l_d_g_s = hrrwork + 6885;
71 double * const HRR_INT__m_p_g_s = hrrwork + 10935;
72
73
74 // Create constants
75 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
76 const SIMINT_DBLTYPE const_10 = SIMINT_DBLSET1(10);
77 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
78 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
79 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
80 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
81 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
82 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
83 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
84 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
85 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
86
87
88 ////////////////////////////////////////
89 // Loop over shells and primitives
90 ////////////////////////////////////////
91
92 real_abcd = 0;
93 istart = 0;
94 for(ab = 0; ab < P.nshell12_clip; ++ab)
95 {
96 const int iend = istart + P.nprim12[ab];
97
98 cd = 0;
99 jstart = 0;
100
101 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
102 {
103 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
104 int jend = jstart;
105 for(i = 0; i < nshellbatch; i++)
106 jend += Q.nprim12[cd+i];
107
108 // Clear the beginning of the workspace (where we are accumulating integrals)
109 memset(work, 0, SIMINT_NSHELL_SIMD * 3030 * sizeof(double));
110 abcd = 0;
111
112
113 for(i = istart; i < iend; ++i)
114 {
115 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
116
117 if(check_screen)
118 {
119 // Skip this whole thing if always insignificant
120 if((P.screen[i] * Q.screen_max) < screen_tol)
121 continue;
122 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
123 }
124
125 icd = 0;
126 iprimcd = 0;
127 nprim_icd = Q.nprim12[cd];
128 double * restrict PRIM_PTR_INT__k_s_g_s = INT__k_s_g_s + abcd * 540;
129 double * restrict PRIM_PTR_INT__l_s_g_s = INT__l_s_g_s + abcd * 675;
130 double * restrict PRIM_PTR_INT__m_s_g_s = INT__m_s_g_s + abcd * 825;
131 double * restrict PRIM_PTR_INT__n_s_g_s = INT__n_s_g_s + abcd * 990;
132
133
134
135 // Load these one per loop over i
136 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
137 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
138 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
139
140 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
141
142 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
143 {
144 // calculate the shell offsets
145 // these are the offset from the shell pointed to by cd
146 // for each element
147 int shelloffsets[SIMINT_SIMD_LEN] = {0};
148 int lastoffset = 0;
149 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
150
151 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
152 {
153 // Handle if the first element of the vector is a new shell
154 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
155 {
156 nprim_icd += Q.nprim12[cd + (++icd)];
157 PRIM_PTR_INT__k_s_g_s += 540;
158 PRIM_PTR_INT__l_s_g_s += 675;
159 PRIM_PTR_INT__m_s_g_s += 825;
160 PRIM_PTR_INT__n_s_g_s += 990;
161 }
162 iprimcd++;
163 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
164 {
165 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
166 {
167 shelloffsets[n] = shelloffsets[n-1] + 1;
168 lastoffset++;
169 nprim_icd += Q.nprim12[cd + (++icd)];
170 }
171 else
172 shelloffsets[n] = shelloffsets[n-1];
173 iprimcd++;
174 }
175 }
176 else
177 iprimcd += SIMINT_SIMD_LEN;
178
179 // Do we have to compute this vector (or has it been screened out)?
180 // (not_screened != 0 means we have to do this vector)
181 if(check_screen)
182 {
183 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
184 if(vmax < screen_tol)
185 {
186 PRIM_PTR_INT__k_s_g_s += lastoffset*540;
187 PRIM_PTR_INT__l_s_g_s += lastoffset*675;
188 PRIM_PTR_INT__m_s_g_s += lastoffset*825;
189 PRIM_PTR_INT__n_s_g_s += lastoffset*990;
190 continue;
191 }
192 }
193
194 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
195 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
196 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
197 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
198
199
200 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
201 SIMINT_DBLTYPE PQ[3];
202 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
203 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
204 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
205 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
206 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
207 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
208
209 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
210 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
211 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
212 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
213 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
214 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
215 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
216
217 // NOTE: Minus sign!
218 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
219 SIMINT_DBLTYPE aop_PQ[3];
220 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
221 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
222 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
223
224 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
225 SIMINT_DBLTYPE aoq_PQ[3];
226 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
227 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
228 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
229 // Put a minus sign here so we don't have to in RR routines
230 a_over_q = SIMINT_NEG(a_over_q);
231
232
233 //////////////////////////////////////////////
234 // Fjt function section
235 // Maximum v value: 14
236 //////////////////////////////////////////////
237 // The parameter to the Fjt function
238 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
239
240
241 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
242
243
244 boys_F_split(PRIM_INT__s_s_s_s, F_x, 14);
245 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
246 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
247 for(n = 0; n <= 14; n++)
248 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
249
250 //////////////////////////////////////////////
251 // Primitive integrals: Vertical recurrance
252 //////////////////////////////////////////////
253
254 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
255 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
256 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
257 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
258 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
259 const SIMINT_DBLTYPE vrr_const_6_over_2p = SIMINT_MUL(const_6, one_over_2p);
260 const SIMINT_DBLTYPE vrr_const_7_over_2p = SIMINT_MUL(const_7, one_over_2p);
261 const SIMINT_DBLTYPE vrr_const_8_over_2p = SIMINT_MUL(const_8, one_over_2p);
262 const SIMINT_DBLTYPE vrr_const_9_over_2p = SIMINT_MUL(const_9, one_over_2p);
263 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
264 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
265 const SIMINT_DBLTYPE vrr_const_3_over_2q = SIMINT_MUL(const_3, one_over_2q);
266 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
267 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
268 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
269 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
270 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
271 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
272 const SIMINT_DBLTYPE vrr_const_7_over_2pq = SIMINT_MUL(const_7, one_over_2pq);
273 const SIMINT_DBLTYPE vrr_const_8_over_2pq = SIMINT_MUL(const_8, one_over_2pq);
274 const SIMINT_DBLTYPE vrr_const_9_over_2pq = SIMINT_MUL(const_9, one_over_2pq);
275 const SIMINT_DBLTYPE vrr_const_10_over_2pq = SIMINT_MUL(const_10, one_over_2pq);
276
277
278
279 // Forming PRIM_INT__p_s_s_s[14 * 3];
280 for(n = 0; n < 14; ++n) // loop over orders of auxiliary function
281 {
282
283 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
284 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]);
285
286 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
287 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]);
288
289 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
290 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]);
291
292 }
293
294
295
296 // Forming PRIM_INT__d_s_s_s[13 * 6];
297 for(n = 0; n < 13; ++n) // loop over orders of auxiliary function
298 {
299
300 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
301 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]);
302 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]);
303
304 PRIM_INT__d_s_s_s[n * 6 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 0]);
305 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]);
306
307 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
308 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]);
309 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]);
310
311 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
312 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]);
313 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]);
314
315 }
316
317
318
319 // Forming PRIM_INT__f_s_s_s[12 * 10];
320 for(n = 0; n < 12; ++n) // loop over orders of auxiliary function
321 {
322
323 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
324 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]);
325 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]);
326
327 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
328 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]);
329
330 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
331 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]);
332
333 PRIM_INT__f_s_s_s[n * 10 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
334 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]);
335
336 PRIM_INT__f_s_s_s[n * 10 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 1]);
337 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]);
338
339 PRIM_INT__f_s_s_s[n * 10 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
340 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]);
341
342 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
343 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]);
344 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]);
345
346 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
347 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]);
348
349 PRIM_INT__f_s_s_s[n * 10 + 8] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 5]);
350 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]);
351
352 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
353 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]);
354 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]);
355
356 }
357
358
359 VRR_I_g_s_s_s(
360 PRIM_INT__g_s_s_s,
361 PRIM_INT__f_s_s_s,
362 PRIM_INT__d_s_s_s,
363 P_PA,
364 a_over_p,
365 aop_PQ,
366 one_over_2p,
367 11);
368
369
370 VRR_I_h_s_s_s(
371 PRIM_INT__h_s_s_s,
372 PRIM_INT__g_s_s_s,
373 PRIM_INT__f_s_s_s,
374 P_PA,
375 a_over_p,
376 aop_PQ,
377 one_over_2p,
378 10);
379
380
381 ostei_general_vrr1_I(6, 9,
382 one_over_2p, a_over_p, aop_PQ, P_PA,
383 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
384
385
386 ostei_general_vrr1_I(7, 8,
387 one_over_2p, a_over_p, aop_PQ, P_PA,
388 PRIM_INT__i_s_s_s, PRIM_INT__h_s_s_s, PRIM_INT__k_s_s_s);
389
390
391 ostei_general_vrr_K(7, 0, 1, 0, 4,
392 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
393 PRIM_INT__k_s_s_s, NULL, NULL, PRIM_INT__i_s_s_s, NULL, PRIM_INT__k_s_p_s);
394
395
396 ostei_general_vrr_K(6, 0, 1, 0, 4,
397 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
398 PRIM_INT__i_s_s_s, NULL, NULL, PRIM_INT__h_s_s_s, NULL, PRIM_INT__i_s_p_s);
399
400
401 ostei_general_vrr_K(7, 0, 2, 0, 3,
402 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
403 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);
404
405
406 ostei_general_vrr_K(5, 0, 1, 0, 4,
407 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
408 PRIM_INT__h_s_s_s, NULL, NULL, PRIM_INT__g_s_s_s, NULL, PRIM_INT__h_s_p_s);
409
410
411 ostei_general_vrr_K(6, 0, 2, 0, 3,
412 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
413 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);
414
415
416 ostei_general_vrr_K(7, 0, 3, 0, 2,
417 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
418 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);
419
420
421 VRR_K_g_s_p_s(
422 PRIM_INT__g_s_p_s,
423 PRIM_INT__g_s_s_s,
424 PRIM_INT__f_s_s_s,
425 Q_PA,
426 aoq_PQ,
427 one_over_2pq,
428 4);
429
430
431 ostei_general_vrr_K(5, 0, 2, 0, 3,
432 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
433 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);
434
435
436 ostei_general_vrr_K(6, 0, 3, 0, 2,
437 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
438 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);
439
440
441 ostei_general_vrr_K(7, 0, 4, 0, 1,
442 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
443 PRIM_INT__k_s_f_s, PRIM_INT__k_s_d_s, NULL, PRIM_INT__i_s_f_s, NULL, PRIM_INT__k_s_g_s);
444
445
446 ostei_general_vrr1_I(8, 7,
447 one_over_2p, a_over_p, aop_PQ, P_PA,
448 PRIM_INT__k_s_s_s, PRIM_INT__i_s_s_s, PRIM_INT__l_s_s_s);
449
450
451 ostei_general_vrr_K(8, 0, 1, 0, 4,
452 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
453 PRIM_INT__l_s_s_s, NULL, NULL, PRIM_INT__k_s_s_s, NULL, PRIM_INT__l_s_p_s);
454
455
456 ostei_general_vrr_K(8, 0, 2, 0, 3,
457 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
458 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);
459
460
461 ostei_general_vrr_K(8, 0, 3, 0, 2,
462 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
463 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);
464
465
466 ostei_general_vrr_K(8, 0, 4, 0, 1,
467 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
468 PRIM_INT__l_s_f_s, PRIM_INT__l_s_d_s, NULL, PRIM_INT__k_s_f_s, NULL, PRIM_INT__l_s_g_s);
469
470
471 ostei_general_vrr1_I(9, 6,
472 one_over_2p, a_over_p, aop_PQ, P_PA,
473 PRIM_INT__l_s_s_s, PRIM_INT__k_s_s_s, PRIM_INT__m_s_s_s);
474
475
476 ostei_general_vrr_K(9, 0, 1, 0, 4,
477 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
478 PRIM_INT__m_s_s_s, NULL, NULL, PRIM_INT__l_s_s_s, NULL, PRIM_INT__m_s_p_s);
479
480
481 ostei_general_vrr_K(9, 0, 2, 0, 3,
482 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
483 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);
484
485
486 ostei_general_vrr_K(9, 0, 3, 0, 2,
487 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
488 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);
489
490
491 ostei_general_vrr_K(9, 0, 4, 0, 1,
492 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
493 PRIM_INT__m_s_f_s, PRIM_INT__m_s_d_s, NULL, PRIM_INT__l_s_f_s, NULL, PRIM_INT__m_s_g_s);
494
495
496 ostei_general_vrr1_I(10, 5,
497 one_over_2p, a_over_p, aop_PQ, P_PA,
498 PRIM_INT__m_s_s_s, PRIM_INT__l_s_s_s, PRIM_INT__n_s_s_s);
499
500
501 ostei_general_vrr_K(10, 0, 1, 0, 4,
502 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
503 PRIM_INT__n_s_s_s, NULL, NULL, PRIM_INT__m_s_s_s, NULL, PRIM_INT__n_s_p_s);
504
505
506 ostei_general_vrr_K(10, 0, 2, 0, 3,
507 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
508 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);
509
510
511 ostei_general_vrr_K(10, 0, 3, 0, 2,
512 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
513 PRIM_INT__n_s_d_s, PRIM_INT__n_s_p_s, NULL, PRIM_INT__m_s_d_s, NULL, PRIM_INT__n_s_f_s);
514
515
516 ostei_general_vrr_K(10, 0, 4, 0, 1,
517 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
518 PRIM_INT__n_s_f_s, PRIM_INT__n_s_d_s, NULL, PRIM_INT__m_s_f_s, NULL, PRIM_INT__n_s_g_s);
519
520
521
522
523 ////////////////////////////////////
524 // Accumulate contracted integrals
525 ////////////////////////////////////
526 if(lastoffset == 0)
527 {
528 contract_all(540, PRIM_INT__k_s_g_s, PRIM_PTR_INT__k_s_g_s);
529 contract_all(675, PRIM_INT__l_s_g_s, PRIM_PTR_INT__l_s_g_s);
530 contract_all(825, PRIM_INT__m_s_g_s, PRIM_PTR_INT__m_s_g_s);
531 contract_all(990, PRIM_INT__n_s_g_s, PRIM_PTR_INT__n_s_g_s);
532 }
533 else
534 {
535 contract(540, shelloffsets, PRIM_INT__k_s_g_s, PRIM_PTR_INT__k_s_g_s);
536 contract(675, shelloffsets, PRIM_INT__l_s_g_s, PRIM_PTR_INT__l_s_g_s);
537 contract(825, shelloffsets, PRIM_INT__m_s_g_s, PRIM_PTR_INT__m_s_g_s);
538 contract(990, shelloffsets, PRIM_INT__n_s_g_s, PRIM_PTR_INT__n_s_g_s);
539 PRIM_PTR_INT__k_s_g_s += lastoffset*540;
540 PRIM_PTR_INT__l_s_g_s += lastoffset*675;
541 PRIM_PTR_INT__m_s_g_s += lastoffset*825;
542 PRIM_PTR_INT__n_s_g_s += lastoffset*990;
543 }
544
545 } // close loop over j
546 } // close loop over i
547
548 //Advance to the next batch
549 jstart = SIMINT_SIMD_ROUND(jend);
550
551 //////////////////////////////////////////////
552 // Contracted integrals: Horizontal recurrance
553 //////////////////////////////////////////////
554
555
556 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
557
558
559 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
560 {
561
562 // set up HRR pointers
563 double const * restrict HRR_INT__k_s_g_s = INT__k_s_g_s + abcd * 540;
564 double const * restrict HRR_INT__l_s_g_s = INT__l_s_g_s + abcd * 675;
565 double const * restrict HRR_INT__m_s_g_s = INT__m_s_g_s + abcd * 825;
566 double const * restrict HRR_INT__n_s_g_s = INT__n_s_g_s + abcd * 990;
567 double * restrict HRR_INT__k_f_g_s = INT__k_f_g_s + real_abcd * 5400;
568
569 // form INT__k_p_g_s
570 ostei_general_hrr_J(7, 1, 4, 0, hAB, HRR_INT__l_s_g_s, HRR_INT__k_s_g_s, HRR_INT__k_p_g_s);
571
572 // form INT__l_p_g_s
573 ostei_general_hrr_J(8, 1, 4, 0, hAB, HRR_INT__m_s_g_s, HRR_INT__l_s_g_s, HRR_INT__l_p_g_s);
574
575 // form INT__m_p_g_s
576 ostei_general_hrr_J(9, 1, 4, 0, hAB, HRR_INT__n_s_g_s, HRR_INT__m_s_g_s, HRR_INT__m_p_g_s);
577
578 // form INT__k_d_g_s
579 ostei_general_hrr_J(7, 2, 4, 0, hAB, HRR_INT__l_p_g_s, HRR_INT__k_p_g_s, HRR_INT__k_d_g_s);
580
581 // form INT__l_d_g_s
582 ostei_general_hrr_J(8, 2, 4, 0, hAB, HRR_INT__m_p_g_s, HRR_INT__l_p_g_s, HRR_INT__l_d_g_s);
583
584 // form INT__k_f_g_s
585 ostei_general_hrr_J(7, 3, 4, 0, hAB, HRR_INT__l_d_g_s, HRR_INT__k_d_g_s, HRR_INT__k_f_g_s);
586
587
588 } // close HRR loop
589
590
591 } // close loop cdbatch
592
593 istart = iend;
594 } // close loop over ab
595
596 return P.nshell12_clip * Q.nshell12_clip;
597 }
598
ostei_f_k_g_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_k_g_s)599 int ostei_f_k_g_s(struct simint_multi_shellpair const P,
600 struct simint_multi_shellpair const Q,
601 double screen_tol,
602 double * const restrict work,
603 double * const restrict INT__f_k_g_s)
604 {
605 double P_AB[3*P.nshell12];
606 struct simint_multi_shellpair P_tmp = P;
607 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
608 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
609 P_tmp.AB_x = P_AB;
610 P_tmp.AB_y = P_AB + P.nshell12;
611 P_tmp.AB_z = P_AB + 2*P.nshell12;
612
613 for(int i = 0; i < P.nshell12; i++)
614 {
615 P_tmp.AB_x[i] = -P.AB_x[i];
616 P_tmp.AB_y[i] = -P.AB_y[i];
617 P_tmp.AB_z[i] = -P.AB_z[i];
618 }
619
620 int ret = ostei_k_f_g_s(P_tmp, Q, screen_tol, work, INT__f_k_g_s);
621 double buffer[5400] SIMINT_ALIGN_ARRAY_DBL;
622
623 for(int q = 0; q < ret; q++)
624 {
625 int idx = 0;
626 for(int a = 0; a < 10; ++a)
627 for(int b = 0; b < 36; ++b)
628 for(int c = 0; c < 15; ++c)
629 for(int d = 0; d < 1; ++d)
630 buffer[idx++] = INT__f_k_g_s[q*5400+b*150+a*15+c*1+d];
631
632 memcpy(INT__f_k_g_s+q*5400, buffer, 5400*sizeof(double));
633 }
634
635 return ret;
636 }
637
ostei_k_f_s_g(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__k_f_s_g)638 int ostei_k_f_s_g(struct simint_multi_shellpair const P,
639 struct simint_multi_shellpair const Q,
640 double screen_tol,
641 double * const restrict work,
642 double * const restrict INT__k_f_s_g)
643 {
644 double Q_AB[3*Q.nshell12];
645 struct simint_multi_shellpair Q_tmp = Q;
646 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
647 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
648 Q_tmp.AB_x = Q_AB;
649 Q_tmp.AB_y = Q_AB + Q.nshell12;
650 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
651
652 for(int i = 0; i < Q.nshell12; i++)
653 {
654 Q_tmp.AB_x[i] = -Q.AB_x[i];
655 Q_tmp.AB_y[i] = -Q.AB_y[i];
656 Q_tmp.AB_z[i] = -Q.AB_z[i];
657 }
658
659 int ret = ostei_k_f_g_s(P, Q_tmp, screen_tol, work, INT__k_f_s_g);
660 double buffer[5400] SIMINT_ALIGN_ARRAY_DBL;
661
662 for(int q = 0; q < ret; q++)
663 {
664 int idx = 0;
665 for(int a = 0; a < 36; ++a)
666 for(int b = 0; b < 10; ++b)
667 for(int c = 0; c < 1; ++c)
668 for(int d = 0; d < 15; ++d)
669 buffer[idx++] = INT__k_f_s_g[q*5400+a*150+b*15+d*1+c];
670
671 memcpy(INT__k_f_s_g+q*5400, buffer, 5400*sizeof(double));
672 }
673
674 return ret;
675 }
676
ostei_f_k_s_g(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__f_k_s_g)677 int ostei_f_k_s_g(struct simint_multi_shellpair const P,
678 struct simint_multi_shellpair const Q,
679 double screen_tol,
680 double * const restrict work,
681 double * const restrict INT__f_k_s_g)
682 {
683 double P_AB[3*P.nshell12];
684 struct simint_multi_shellpair P_tmp = P;
685 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
686 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
687 P_tmp.AB_x = P_AB;
688 P_tmp.AB_y = P_AB + P.nshell12;
689 P_tmp.AB_z = P_AB + 2*P.nshell12;
690
691 for(int i = 0; i < P.nshell12; i++)
692 {
693 P_tmp.AB_x[i] = -P.AB_x[i];
694 P_tmp.AB_y[i] = -P.AB_y[i];
695 P_tmp.AB_z[i] = -P.AB_z[i];
696 }
697
698 double Q_AB[3*Q.nshell12];
699 struct simint_multi_shellpair Q_tmp = Q;
700 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
701 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
702 Q_tmp.AB_x = Q_AB;
703 Q_tmp.AB_y = Q_AB + Q.nshell12;
704 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
705
706 for(int i = 0; i < Q.nshell12; i++)
707 {
708 Q_tmp.AB_x[i] = -Q.AB_x[i];
709 Q_tmp.AB_y[i] = -Q.AB_y[i];
710 Q_tmp.AB_z[i] = -Q.AB_z[i];
711 }
712
713 int ret = ostei_k_f_g_s(P_tmp, Q_tmp, screen_tol, work, INT__f_k_s_g);
714 double buffer[5400] SIMINT_ALIGN_ARRAY_DBL;
715
716 for(int q = 0; q < ret; q++)
717 {
718 int idx = 0;
719 for(int a = 0; a < 10; ++a)
720 for(int b = 0; b < 36; ++b)
721 for(int c = 0; c < 1; ++c)
722 for(int d = 0; d < 15; ++d)
723 buffer[idx++] = INT__f_k_s_g[q*5400+b*150+a*15+d*1+c];
724
725 memcpy(INT__f_k_s_g+q*5400, buffer, 5400*sizeof(double));
726 }
727
728 return ret;
729 }
730
731