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_s_d_p(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__i_s_d_p)8 int ostei_i_s_d_p(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_s_d_p)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__i_s_d_p);
17 int ab, cd, abcd;
18 int istart, jstart;
19 int iprimcd, nprim_icd, icd;
20 const int check_screen = (screen_tol > 0.0);
21 int i, j;
22 int n;
23 int not_screened;
24 int real_abcd;
25 int ibra;
26
27 // partition workspace
28 double * const INT__i_s_d_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__i_s_f_s = work + (SIMINT_NSHELL_SIMD * 168);
30 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*448);
31 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
32 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_s_s = primwork + 10;
33 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 37;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 85;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 155;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_p_s = primwork + 245;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 380;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_p_s = primwork + 485;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_d_s = primwork + 674;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 926;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_p_s = primwork + 1038;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_d_s = primwork + 1290;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_f_s = primwork + 1626;
44 double * const hrrwork = (double *)(primwork + 1906);
45
46
47 // Create constants
48 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
49 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
50 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
51 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
52 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
53 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
54 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
55
56
57 ////////////////////////////////////////
58 // Loop over shells and primitives
59 ////////////////////////////////////////
60
61 real_abcd = 0;
62 istart = 0;
63 for(ab = 0; ab < P.nshell12_clip; ++ab)
64 {
65 const int iend = istart + P.nprim12[ab];
66
67 cd = 0;
68 jstart = 0;
69
70 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
71 {
72 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
73 int jend = jstart;
74 for(i = 0; i < nshellbatch; i++)
75 jend += Q.nprim12[cd+i];
76
77 // Clear the beginning of the workspace (where we are accumulating integrals)
78 memset(work, 0, SIMINT_NSHELL_SIMD * 448 * sizeof(double));
79 abcd = 0;
80
81
82 for(i = istart; i < iend; ++i)
83 {
84 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
85
86 if(check_screen)
87 {
88 // Skip this whole thing if always insignificant
89 if((P.screen[i] * Q.screen_max) < screen_tol)
90 continue;
91 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
92 }
93
94 icd = 0;
95 iprimcd = 0;
96 nprim_icd = Q.nprim12[cd];
97 double * restrict PRIM_PTR_INT__i_s_d_s = INT__i_s_d_s + abcd * 168;
98 double * restrict PRIM_PTR_INT__i_s_f_s = INT__i_s_f_s + abcd * 280;
99
100
101
102 // Load these one per loop over i
103 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
104 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
105 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
106
107 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
108
109 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
110 {
111 // calculate the shell offsets
112 // these are the offset from the shell pointed to by cd
113 // for each element
114 int shelloffsets[SIMINT_SIMD_LEN] = {0};
115 int lastoffset = 0;
116 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
117
118 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
119 {
120 // Handle if the first element of the vector is a new shell
121 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
122 {
123 nprim_icd += Q.nprim12[cd + (++icd)];
124 PRIM_PTR_INT__i_s_d_s += 168;
125 PRIM_PTR_INT__i_s_f_s += 280;
126 }
127 iprimcd++;
128 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
129 {
130 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
131 {
132 shelloffsets[n] = shelloffsets[n-1] + 1;
133 lastoffset++;
134 nprim_icd += Q.nprim12[cd + (++icd)];
135 }
136 else
137 shelloffsets[n] = shelloffsets[n-1];
138 iprimcd++;
139 }
140 }
141 else
142 iprimcd += SIMINT_SIMD_LEN;
143
144 // Do we have to compute this vector (or has it been screened out)?
145 // (not_screened != 0 means we have to do this vector)
146 if(check_screen)
147 {
148 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
149 if(vmax < screen_tol)
150 {
151 PRIM_PTR_INT__i_s_d_s += lastoffset*168;
152 PRIM_PTR_INT__i_s_f_s += lastoffset*280;
153 continue;
154 }
155 }
156
157 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
158 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
159 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
160 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
161
162
163 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
164 SIMINT_DBLTYPE PQ[3];
165 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
166 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
167 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
168 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
169 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
170 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
171
172 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
173 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
174 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
175 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
176 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
177 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
178 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
179
180 // NOTE: Minus sign!
181 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
182 SIMINT_DBLTYPE aop_PQ[3];
183 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
184 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
185 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
186
187 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
188 SIMINT_DBLTYPE aoq_PQ[3];
189 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
190 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
191 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
192 // Put a minus sign here so we don't have to in RR routines
193 a_over_q = SIMINT_NEG(a_over_q);
194
195
196 //////////////////////////////////////////////
197 // Fjt function section
198 // Maximum v value: 9
199 //////////////////////////////////////////////
200 // The parameter to the Fjt function
201 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
202
203
204 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
205
206
207 boys_F_split(PRIM_INT__s_s_s_s, F_x, 9);
208 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
209 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
210 for(n = 0; n <= 9; n++)
211 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
212
213 //////////////////////////////////////////////
214 // Primitive integrals: Vertical recurrance
215 //////////////////////////////////////////////
216
217 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
218 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
219 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
220 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
221 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
222 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
223 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
224 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
225 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
226 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
227 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
228 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
229 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
230
231
232
233 // Forming PRIM_INT__p_s_s_s[9 * 3];
234 for(n = 0; n < 9; ++n) // loop over orders of auxiliary function
235 {
236
237 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
238 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]);
239
240 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
241 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]);
242
243 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
244 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]);
245
246 }
247
248
249
250 // Forming PRIM_INT__d_s_s_s[8 * 6];
251 for(n = 0; n < 8; ++n) // loop over orders of auxiliary function
252 {
253
254 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
255 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]);
256 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]);
257
258 PRIM_INT__d_s_s_s[n * 6 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 0]);
259 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]);
260
261 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
262 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]);
263 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]);
264
265 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
266 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]);
267 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]);
268
269 }
270
271
272
273 // Forming PRIM_INT__f_s_s_s[7 * 10];
274 for(n = 0; n < 7; ++n) // loop over orders of auxiliary function
275 {
276
277 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
278 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]);
279 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]);
280
281 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
282 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]);
283
284 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
285 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]);
286
287 PRIM_INT__f_s_s_s[n * 10 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
288 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]);
289
290 PRIM_INT__f_s_s_s[n * 10 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 1]);
291 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]);
292
293 PRIM_INT__f_s_s_s[n * 10 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
294 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]);
295
296 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
297 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]);
298 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]);
299
300 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
301 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]);
302
303 PRIM_INT__f_s_s_s[n * 10 + 8] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 5]);
304 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]);
305
306 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
307 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]);
308 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]);
309
310 }
311
312
313 VRR_I_g_s_s_s(
314 PRIM_INT__g_s_s_s,
315 PRIM_INT__f_s_s_s,
316 PRIM_INT__d_s_s_s,
317 P_PA,
318 a_over_p,
319 aop_PQ,
320 one_over_2p,
321 6);
322
323
324 VRR_I_h_s_s_s(
325 PRIM_INT__h_s_s_s,
326 PRIM_INT__g_s_s_s,
327 PRIM_INT__f_s_s_s,
328 P_PA,
329 a_over_p,
330 aop_PQ,
331 one_over_2p,
332 5);
333
334
335 ostei_general_vrr1_I(6, 4,
336 one_over_2p, a_over_p, aop_PQ, P_PA,
337 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
338
339
340 ostei_general_vrr_K(6, 0, 1, 0, 3,
341 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
342 PRIM_INT__i_s_s_s, NULL, NULL, PRIM_INT__h_s_s_s, NULL, PRIM_INT__i_s_p_s);
343
344
345 ostei_general_vrr_K(5, 0, 1, 0, 3,
346 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
347 PRIM_INT__h_s_s_s, NULL, NULL, PRIM_INT__g_s_s_s, NULL, PRIM_INT__h_s_p_s);
348
349
350 ostei_general_vrr_K(6, 0, 2, 0, 2,
351 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
352 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);
353
354
355 VRR_K_g_s_p_s(
356 PRIM_INT__g_s_p_s,
357 PRIM_INT__g_s_s_s,
358 PRIM_INT__f_s_s_s,
359 Q_PA,
360 aoq_PQ,
361 one_over_2pq,
362 3);
363
364
365 ostei_general_vrr_K(5, 0, 2, 0, 2,
366 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
367 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);
368
369
370 ostei_general_vrr_K(6, 0, 3, 0, 1,
371 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
372 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);
373
374
375
376
377 ////////////////////////////////////
378 // Accumulate contracted integrals
379 ////////////////////////////////////
380 if(lastoffset == 0)
381 {
382 contract_all(168, PRIM_INT__i_s_d_s, PRIM_PTR_INT__i_s_d_s);
383 contract_all(280, PRIM_INT__i_s_f_s, PRIM_PTR_INT__i_s_f_s);
384 }
385 else
386 {
387 contract(168, shelloffsets, PRIM_INT__i_s_d_s, PRIM_PTR_INT__i_s_d_s);
388 contract(280, shelloffsets, PRIM_INT__i_s_f_s, PRIM_PTR_INT__i_s_f_s);
389 PRIM_PTR_INT__i_s_d_s += lastoffset*168;
390 PRIM_PTR_INT__i_s_f_s += lastoffset*280;
391 }
392
393 } // close loop over j
394 } // close loop over i
395
396 //Advance to the next batch
397 jstart = SIMINT_SIMD_ROUND(jend);
398
399 //////////////////////////////////////////////
400 // Contracted integrals: Horizontal recurrance
401 //////////////////////////////////////////////
402
403
404
405
406 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
407 {
408 const double hCD[3] = { Q.AB_x[cd+abcd], Q.AB_y[cd+abcd], Q.AB_z[cd+abcd] };
409
410 // set up HRR pointers
411 double const * restrict HRR_INT__i_s_d_s = INT__i_s_d_s + abcd * 168;
412 double const * restrict HRR_INT__i_s_f_s = INT__i_s_f_s + abcd * 280;
413 double * restrict HRR_INT__i_s_d_p = INT__i_s_d_p + real_abcd * 504;
414
415 // form INT__i_s_d_p
416 for(ibra = 0; ibra < 28; ++ibra)
417 {
418 HRR_INT__i_s_d_p[ibra * 18 + 0] = HRR_INT__i_s_f_s[ibra * 10 + 0] + ( hCD[0] * HRR_INT__i_s_d_s[ibra * 6 + 0] );
419
420 HRR_INT__i_s_d_p[ibra * 18 + 1] = HRR_INT__i_s_f_s[ibra * 10 + 1] + ( hCD[1] * HRR_INT__i_s_d_s[ibra * 6 + 0] );
421
422 HRR_INT__i_s_d_p[ibra * 18 + 2] = HRR_INT__i_s_f_s[ibra * 10 + 2] + ( hCD[2] * HRR_INT__i_s_d_s[ibra * 6 + 0] );
423
424 HRR_INT__i_s_d_p[ibra * 18 + 3] = HRR_INT__i_s_f_s[ibra * 10 + 1] + ( hCD[0] * HRR_INT__i_s_d_s[ibra * 6 + 1] );
425
426 HRR_INT__i_s_d_p[ibra * 18 + 4] = HRR_INT__i_s_f_s[ibra * 10 + 3] + ( hCD[1] * HRR_INT__i_s_d_s[ibra * 6 + 1] );
427
428 HRR_INT__i_s_d_p[ibra * 18 + 5] = HRR_INT__i_s_f_s[ibra * 10 + 4] + ( hCD[2] * HRR_INT__i_s_d_s[ibra * 6 + 1] );
429
430 HRR_INT__i_s_d_p[ibra * 18 + 6] = HRR_INT__i_s_f_s[ibra * 10 + 2] + ( hCD[0] * HRR_INT__i_s_d_s[ibra * 6 + 2] );
431
432 HRR_INT__i_s_d_p[ibra * 18 + 7] = HRR_INT__i_s_f_s[ibra * 10 + 4] + ( hCD[1] * HRR_INT__i_s_d_s[ibra * 6 + 2] );
433
434 HRR_INT__i_s_d_p[ibra * 18 + 8] = HRR_INT__i_s_f_s[ibra * 10 + 5] + ( hCD[2] * HRR_INT__i_s_d_s[ibra * 6 + 2] );
435
436 HRR_INT__i_s_d_p[ibra * 18 + 9] = HRR_INT__i_s_f_s[ibra * 10 + 3] + ( hCD[0] * HRR_INT__i_s_d_s[ibra * 6 + 3] );
437
438 HRR_INT__i_s_d_p[ibra * 18 + 10] = HRR_INT__i_s_f_s[ibra * 10 + 6] + ( hCD[1] * HRR_INT__i_s_d_s[ibra * 6 + 3] );
439
440 HRR_INT__i_s_d_p[ibra * 18 + 11] = HRR_INT__i_s_f_s[ibra * 10 + 7] + ( hCD[2] * HRR_INT__i_s_d_s[ibra * 6 + 3] );
441
442 HRR_INT__i_s_d_p[ibra * 18 + 12] = HRR_INT__i_s_f_s[ibra * 10 + 4] + ( hCD[0] * HRR_INT__i_s_d_s[ibra * 6 + 4] );
443
444 HRR_INT__i_s_d_p[ibra * 18 + 13] = HRR_INT__i_s_f_s[ibra * 10 + 7] + ( hCD[1] * HRR_INT__i_s_d_s[ibra * 6 + 4] );
445
446 HRR_INT__i_s_d_p[ibra * 18 + 14] = HRR_INT__i_s_f_s[ibra * 10 + 8] + ( hCD[2] * HRR_INT__i_s_d_s[ibra * 6 + 4] );
447
448 HRR_INT__i_s_d_p[ibra * 18 + 15] = HRR_INT__i_s_f_s[ibra * 10 + 5] + ( hCD[0] * HRR_INT__i_s_d_s[ibra * 6 + 5] );
449
450 HRR_INT__i_s_d_p[ibra * 18 + 16] = HRR_INT__i_s_f_s[ibra * 10 + 8] + ( hCD[1] * HRR_INT__i_s_d_s[ibra * 6 + 5] );
451
452 HRR_INT__i_s_d_p[ibra * 18 + 17] = HRR_INT__i_s_f_s[ibra * 10 + 9] + ( hCD[2] * HRR_INT__i_s_d_s[ibra * 6 + 5] );
453
454 }
455
456
457 } // close HRR loop
458
459
460 } // close loop cdbatch
461
462 istart = iend;
463 } // close loop over ab
464
465 return P.nshell12_clip * Q.nshell12_clip;
466 }
467
ostei_s_i_d_p(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__s_i_d_p)468 int ostei_s_i_d_p(struct simint_multi_shellpair const P,
469 struct simint_multi_shellpair const Q,
470 double screen_tol,
471 double * const restrict work,
472 double * const restrict INT__s_i_d_p)
473 {
474 double P_AB[3*P.nshell12];
475 struct simint_multi_shellpair P_tmp = P;
476 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
477 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
478 P_tmp.AB_x = P_AB;
479 P_tmp.AB_y = P_AB + P.nshell12;
480 P_tmp.AB_z = P_AB + 2*P.nshell12;
481
482 for(int i = 0; i < P.nshell12; i++)
483 {
484 P_tmp.AB_x[i] = -P.AB_x[i];
485 P_tmp.AB_y[i] = -P.AB_y[i];
486 P_tmp.AB_z[i] = -P.AB_z[i];
487 }
488
489 int ret = ostei_i_s_d_p(P_tmp, Q, screen_tol, work, INT__s_i_d_p);
490 double buffer[504] SIMINT_ALIGN_ARRAY_DBL;
491
492 for(int q = 0; q < ret; q++)
493 {
494 int idx = 0;
495 for(int a = 0; a < 1; ++a)
496 for(int b = 0; b < 28; ++b)
497 for(int c = 0; c < 6; ++c)
498 for(int d = 0; d < 3; ++d)
499 buffer[idx++] = INT__s_i_d_p[q*504+b*18+a*18+c*3+d];
500
501 memcpy(INT__s_i_d_p+q*504, buffer, 504*sizeof(double));
502 }
503
504 return ret;
505 }
506
ostei_i_s_p_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_s_p_d)507 int ostei_i_s_p_d(struct simint_multi_shellpair const P,
508 struct simint_multi_shellpair const Q,
509 double screen_tol,
510 double * const restrict work,
511 double * const restrict INT__i_s_p_d)
512 {
513 double Q_AB[3*Q.nshell12];
514 struct simint_multi_shellpair Q_tmp = Q;
515 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
516 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
517 Q_tmp.AB_x = Q_AB;
518 Q_tmp.AB_y = Q_AB + Q.nshell12;
519 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
520
521 for(int i = 0; i < Q.nshell12; i++)
522 {
523 Q_tmp.AB_x[i] = -Q.AB_x[i];
524 Q_tmp.AB_y[i] = -Q.AB_y[i];
525 Q_tmp.AB_z[i] = -Q.AB_z[i];
526 }
527
528 int ret = ostei_i_s_d_p(P, Q_tmp, screen_tol, work, INT__i_s_p_d);
529 double buffer[504] SIMINT_ALIGN_ARRAY_DBL;
530
531 for(int q = 0; q < ret; q++)
532 {
533 int idx = 0;
534 for(int a = 0; a < 28; ++a)
535 for(int b = 0; b < 1; ++b)
536 for(int c = 0; c < 3; ++c)
537 for(int d = 0; d < 6; ++d)
538 buffer[idx++] = INT__i_s_p_d[q*504+a*18+b*18+d*3+c];
539
540 memcpy(INT__i_s_p_d+q*504, buffer, 504*sizeof(double));
541 }
542
543 return ret;
544 }
545
ostei_s_i_p_d(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__s_i_p_d)546 int ostei_s_i_p_d(struct simint_multi_shellpair const P,
547 struct simint_multi_shellpair const Q,
548 double screen_tol,
549 double * const restrict work,
550 double * const restrict INT__s_i_p_d)
551 {
552 double P_AB[3*P.nshell12];
553 struct simint_multi_shellpair P_tmp = P;
554 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
555 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
556 P_tmp.AB_x = P_AB;
557 P_tmp.AB_y = P_AB + P.nshell12;
558 P_tmp.AB_z = P_AB + 2*P.nshell12;
559
560 for(int i = 0; i < P.nshell12; i++)
561 {
562 P_tmp.AB_x[i] = -P.AB_x[i];
563 P_tmp.AB_y[i] = -P.AB_y[i];
564 P_tmp.AB_z[i] = -P.AB_z[i];
565 }
566
567 double Q_AB[3*Q.nshell12];
568 struct simint_multi_shellpair Q_tmp = Q;
569 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
570 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
571 Q_tmp.AB_x = Q_AB;
572 Q_tmp.AB_y = Q_AB + Q.nshell12;
573 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
574
575 for(int i = 0; i < Q.nshell12; i++)
576 {
577 Q_tmp.AB_x[i] = -Q.AB_x[i];
578 Q_tmp.AB_y[i] = -Q.AB_y[i];
579 Q_tmp.AB_z[i] = -Q.AB_z[i];
580 }
581
582 int ret = ostei_i_s_d_p(P_tmp, Q_tmp, screen_tol, work, INT__s_i_p_d);
583 double buffer[504] SIMINT_ALIGN_ARRAY_DBL;
584
585 for(int q = 0; q < ret; q++)
586 {
587 int idx = 0;
588 for(int a = 0; a < 1; ++a)
589 for(int b = 0; b < 28; ++b)
590 for(int c = 0; c < 3; ++c)
591 for(int d = 0; d < 6; ++d)
592 buffer[idx++] = INT__s_i_p_d[q*504+b*18+a*18+d*3+c];
593
594 memcpy(INT__s_i_p_d+q*504, buffer, 504*sizeof(double));
595 }
596
597 return ret;
598 }
599
600