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_f_p_k_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_p_k_s)8 int ostei_f_p_k_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__f_p_k_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__f_p_k_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__f_s_k_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__g_s_k_s = work + (SIMINT_NSHELL_SIMD * 360);
30 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*900);
31 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
32 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_p_s = primwork + 12;
33 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_d_s = primwork + 45;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_f_s = primwork + 105;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_g_s = primwork + 195;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_h_s = primwork + 315;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_i_s = primwork + 462;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_k_s = primwork + 630;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_g_s = primwork + 810;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_h_s = primwork + 990;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_i_s = primwork + 1242;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_k_s = primwork + 1578;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_h_s = primwork + 2010;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_i_s = primwork + 2388;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_k_s = primwork + 2892;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_i_s = primwork + 3540;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_k_s = primwork + 4100;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_k_s = primwork + 4820;
49 double * const hrrwork = (double *)(primwork + 5360);
50
51
52 // Create constants
53 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
54 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
55 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
56 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
57 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
58 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
59 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
60 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
61
62
63 ////////////////////////////////////////
64 // Loop over shells and primitives
65 ////////////////////////////////////////
66
67 real_abcd = 0;
68 istart = 0;
69 for(ab = 0; ab < P.nshell12_clip; ++ab)
70 {
71 const int iend = istart + P.nprim12[ab];
72
73 cd = 0;
74 jstart = 0;
75
76 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
77 {
78 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
79 int jend = jstart;
80 for(i = 0; i < nshellbatch; i++)
81 jend += Q.nprim12[cd+i];
82
83 // Clear the beginning of the workspace (where we are accumulating integrals)
84 memset(work, 0, SIMINT_NSHELL_SIMD * 900 * sizeof(double));
85 abcd = 0;
86
87
88 for(i = istart; i < iend; ++i)
89 {
90 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
91
92 if(check_screen)
93 {
94 // Skip this whole thing if always insignificant
95 if((P.screen[i] * Q.screen_max) < screen_tol)
96 continue;
97 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
98 }
99
100 icd = 0;
101 iprimcd = 0;
102 nprim_icd = Q.nprim12[cd];
103 double * restrict PRIM_PTR_INT__f_s_k_s = INT__f_s_k_s + abcd * 360;
104 double * restrict PRIM_PTR_INT__g_s_k_s = INT__g_s_k_s + abcd * 540;
105
106
107
108 // Load these one per loop over i
109 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
110 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
111 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
112
113 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
114
115 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
116 {
117 // calculate the shell offsets
118 // these are the offset from the shell pointed to by cd
119 // for each element
120 int shelloffsets[SIMINT_SIMD_LEN] = {0};
121 int lastoffset = 0;
122 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
123
124 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
125 {
126 // Handle if the first element of the vector is a new shell
127 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
128 {
129 nprim_icd += Q.nprim12[cd + (++icd)];
130 PRIM_PTR_INT__f_s_k_s += 360;
131 PRIM_PTR_INT__g_s_k_s += 540;
132 }
133 iprimcd++;
134 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
135 {
136 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
137 {
138 shelloffsets[n] = shelloffsets[n-1] + 1;
139 lastoffset++;
140 nprim_icd += Q.nprim12[cd + (++icd)];
141 }
142 else
143 shelloffsets[n] = shelloffsets[n-1];
144 iprimcd++;
145 }
146 }
147 else
148 iprimcd += SIMINT_SIMD_LEN;
149
150 // Do we have to compute this vector (or has it been screened out)?
151 // (not_screened != 0 means we have to do this vector)
152 if(check_screen)
153 {
154 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
155 if(vmax < screen_tol)
156 {
157 PRIM_PTR_INT__f_s_k_s += lastoffset*360;
158 PRIM_PTR_INT__g_s_k_s += lastoffset*540;
159 continue;
160 }
161 }
162
163 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
164 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
165 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
166 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
167
168
169 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
170 SIMINT_DBLTYPE PQ[3];
171 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
172 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
173 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
174 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
175 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
176 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
177
178 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
179 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
180 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
181 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
182 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
183 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
184 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
185
186 // NOTE: Minus sign!
187 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
188 SIMINT_DBLTYPE aop_PQ[3];
189 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
190 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
191 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
192
193 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
194 SIMINT_DBLTYPE aoq_PQ[3];
195 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
196 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
197 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
198 // Put a minus sign here so we don't have to in RR routines
199 a_over_q = SIMINT_NEG(a_over_q);
200
201
202 //////////////////////////////////////////////
203 // Fjt function section
204 // Maximum v value: 11
205 //////////////////////////////////////////////
206 // The parameter to the Fjt function
207 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
208
209
210 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
211
212
213 boys_F_split(PRIM_INT__s_s_s_s, F_x, 11);
214 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
215 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
216 for(n = 0; n <= 11; n++)
217 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
218
219 //////////////////////////////////////////////
220 // Primitive integrals: Vertical recurrance
221 //////////////////////////////////////////////
222
223 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
224 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
225 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
226 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
227 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
228 const SIMINT_DBLTYPE vrr_const_3_over_2q = SIMINT_MUL(const_3, one_over_2q);
229 const SIMINT_DBLTYPE vrr_const_4_over_2q = SIMINT_MUL(const_4, one_over_2q);
230 const SIMINT_DBLTYPE vrr_const_5_over_2q = SIMINT_MUL(const_5, one_over_2q);
231 const SIMINT_DBLTYPE vrr_const_6_over_2q = SIMINT_MUL(const_6, one_over_2q);
232 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
233 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
234 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
235 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
236 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
237 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
238 const SIMINT_DBLTYPE vrr_const_7_over_2pq = SIMINT_MUL(const_7, one_over_2pq);
239
240
241
242 // Forming PRIM_INT__s_s_p_s[11 * 3];
243 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
244 {
245
246 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
247 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_p_s[n * 3 + 0]);
248
249 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
250 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_p_s[n * 3 + 1]);
251
252 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
253 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_p_s[n * 3 + 2]);
254
255 }
256
257
258
259 // Forming PRIM_INT__s_s_d_s[10 * 6];
260 for(n = 0; n < 10; ++n) // loop over orders of auxiliary function
261 {
262
263 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_p_s[n * 3 + 0]);
264 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_s_d_s[n * 6 + 0]);
265 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_FMADD( vrr_const_1_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__s_s_d_s[n * 6 + 0]);
266
267 PRIM_INT__s_s_d_s[n * 6 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 0]);
268 PRIM_INT__s_s_d_s[n * 6 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_s_d_s[n * 6 + 1]);
269
270 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 1]);
271 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_s_d_s[n * 6 + 3]);
272 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_FMADD( vrr_const_1_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__s_s_d_s[n * 6 + 3]);
273
274 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 2]);
275 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__s_s_d_s[n * 6 + 5]);
276 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_FMADD( vrr_const_1_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__s_s_d_s[n * 6 + 5]);
277
278 }
279
280
281
282 // Forming PRIM_INT__s_s_f_s[9 * 10];
283 for(n = 0; n < 9; ++n) // loop over orders of auxiliary function
284 {
285
286 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
287 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__s_s_f_s[n * 10 + 0]);
288 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_FMADD( vrr_const_2_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_s_p_s[n * 3 + 0]), PRIM_INT__s_s_f_s[n * 10 + 0]);
289
290 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
291 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__s_s_f_s[n * 10 + 1]);
292
293 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
294 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 0], PRIM_INT__s_s_f_s[n * 10 + 2]);
295
296 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
297 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__s_s_f_s[n * 10 + 3]);
298
299 PRIM_INT__s_s_f_s[n * 10 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
300 PRIM_INT__s_s_f_s[n * 10 + 4] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 1], PRIM_INT__s_s_f_s[n * 10 + 4]);
301
302 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
303 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_FMADD( aoq_PQ[0], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__s_s_f_s[n * 10 + 5]);
304
305 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
306 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__s_s_f_s[n * 10 + 6]);
307 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_FMADD( vrr_const_2_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_s_p_s[n * 3 + 1]), PRIM_INT__s_s_f_s[n * 10 + 6]);
308
309 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
310 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 3], PRIM_INT__s_s_f_s[n * 10 + 7]);
311
312 PRIM_INT__s_s_f_s[n * 10 + 8] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
313 PRIM_INT__s_s_f_s[n * 10 + 8] = SIMINT_FMADD( aoq_PQ[1], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__s_s_f_s[n * 10 + 8]);
314
315 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
316 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_d_s[(n+1) * 6 + 5], PRIM_INT__s_s_f_s[n * 10 + 9]);
317 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_FMADD( vrr_const_2_over_2q, SIMINT_FMADD(a_over_q, PRIM_INT__s_s_p_s[(n+1) * 3 + 2], PRIM_INT__s_s_p_s[n * 3 + 2]), PRIM_INT__s_s_f_s[n * 10 + 9]);
318
319 }
320
321
322 VRR_K_s_s_g_s(
323 PRIM_INT__s_s_g_s,
324 PRIM_INT__s_s_f_s,
325 PRIM_INT__s_s_d_s,
326 Q_PA,
327 a_over_q,
328 aoq_PQ,
329 one_over_2q,
330 8);
331
332
333 VRR_K_s_s_h_s(
334 PRIM_INT__s_s_h_s,
335 PRIM_INT__s_s_g_s,
336 PRIM_INT__s_s_f_s,
337 Q_PA,
338 a_over_q,
339 aoq_PQ,
340 one_over_2q,
341 7);
342
343
344 ostei_general_vrr1_K(6, 6,
345 one_over_2q, a_over_q, aoq_PQ, Q_PA,
346 PRIM_INT__s_s_h_s, PRIM_INT__s_s_g_s, PRIM_INT__s_s_i_s);
347
348
349 ostei_general_vrr1_K(7, 5,
350 one_over_2q, a_over_q, aoq_PQ, Q_PA,
351 PRIM_INT__s_s_i_s, PRIM_INT__s_s_h_s, PRIM_INT__s_s_k_s);
352
353
354 ostei_general_vrr_I(1, 0, 7, 0, 4,
355 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
356 PRIM_INT__s_s_k_s, NULL, NULL, PRIM_INT__s_s_i_s, NULL, PRIM_INT__p_s_k_s);
357
358
359 ostei_general_vrr_I(1, 0, 6, 0, 4,
360 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
361 PRIM_INT__s_s_i_s, NULL, NULL, PRIM_INT__s_s_h_s, NULL, PRIM_INT__p_s_i_s);
362
363
364 ostei_general_vrr_I(2, 0, 7, 0, 3,
365 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
366 PRIM_INT__p_s_k_s, PRIM_INT__s_s_k_s, NULL, PRIM_INT__p_s_i_s, NULL, PRIM_INT__d_s_k_s);
367
368
369 ostei_general_vrr_I(1, 0, 5, 0, 4,
370 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
371 PRIM_INT__s_s_h_s, NULL, NULL, PRIM_INT__s_s_g_s, NULL, PRIM_INT__p_s_h_s);
372
373
374 ostei_general_vrr_I(2, 0, 6, 0, 3,
375 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
376 PRIM_INT__p_s_i_s, PRIM_INT__s_s_i_s, NULL, PRIM_INT__p_s_h_s, NULL, PRIM_INT__d_s_i_s);
377
378
379 ostei_general_vrr_I(3, 0, 7, 0, 2,
380 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
381 PRIM_INT__d_s_k_s, PRIM_INT__p_s_k_s, NULL, PRIM_INT__d_s_i_s, NULL, PRIM_INT__f_s_k_s);
382
383
384 VRR_I_p_s_g_s(
385 PRIM_INT__p_s_g_s,
386 PRIM_INT__s_s_g_s,
387 PRIM_INT__s_s_f_s,
388 P_PA,
389 aop_PQ,
390 one_over_2pq,
391 4);
392
393
394 ostei_general_vrr_I(2, 0, 5, 0, 3,
395 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
396 PRIM_INT__p_s_h_s, PRIM_INT__s_s_h_s, NULL, PRIM_INT__p_s_g_s, NULL, PRIM_INT__d_s_h_s);
397
398
399 ostei_general_vrr_I(3, 0, 6, 0, 2,
400 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
401 PRIM_INT__d_s_i_s, PRIM_INT__p_s_i_s, NULL, PRIM_INT__d_s_h_s, NULL, PRIM_INT__f_s_i_s);
402
403
404 ostei_general_vrr_I(4, 0, 7, 0, 1,
405 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
406 PRIM_INT__f_s_k_s, PRIM_INT__d_s_k_s, NULL, PRIM_INT__f_s_i_s, NULL, PRIM_INT__g_s_k_s);
407
408
409
410
411 ////////////////////////////////////
412 // Accumulate contracted integrals
413 ////////////////////////////////////
414 if(lastoffset == 0)
415 {
416 contract_all(360, PRIM_INT__f_s_k_s, PRIM_PTR_INT__f_s_k_s);
417 contract_all(540, PRIM_INT__g_s_k_s, PRIM_PTR_INT__g_s_k_s);
418 }
419 else
420 {
421 contract(360, shelloffsets, PRIM_INT__f_s_k_s, PRIM_PTR_INT__f_s_k_s);
422 contract(540, shelloffsets, PRIM_INT__g_s_k_s, PRIM_PTR_INT__g_s_k_s);
423 PRIM_PTR_INT__f_s_k_s += lastoffset*360;
424 PRIM_PTR_INT__g_s_k_s += lastoffset*540;
425 }
426
427 } // close loop over j
428 } // close loop over i
429
430 //Advance to the next batch
431 jstart = SIMINT_SIMD_ROUND(jend);
432
433 //////////////////////////////////////////////
434 // Contracted integrals: Horizontal recurrance
435 //////////////////////////////////////////////
436
437
438 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
439
440
441 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
442 {
443
444 // set up HRR pointers
445 double const * restrict HRR_INT__f_s_k_s = INT__f_s_k_s + abcd * 360;
446 double const * restrict HRR_INT__g_s_k_s = INT__g_s_k_s + abcd * 540;
447 double * restrict HRR_INT__f_p_k_s = INT__f_p_k_s + real_abcd * 1080;
448
449 // form INT__f_p_k_s
450 HRR_J_f_p(
451 HRR_INT__f_p_k_s,
452 HRR_INT__f_s_k_s,
453 HRR_INT__g_s_k_s,
454 hAB, 36);
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_p_f_k_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__p_f_k_s)468 int ostei_p_f_k_s(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__p_f_k_s)
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_f_p_k_s(P_tmp, Q, screen_tol, work, INT__p_f_k_s);
490 double buffer[1080] SIMINT_ALIGN_ARRAY_DBL;
491
492 for(int q = 0; q < ret; q++)
493 {
494 int idx = 0;
495 for(int a = 0; a < 3; ++a)
496 for(int b = 0; b < 10; ++b)
497 for(int c = 0; c < 36; ++c)
498 for(int d = 0; d < 1; ++d)
499 buffer[idx++] = INT__p_f_k_s[q*1080+b*108+a*36+c*1+d];
500
501 memcpy(INT__p_f_k_s+q*1080, buffer, 1080*sizeof(double));
502 }
503
504 return ret;
505 }
506
ostei_f_p_s_k(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__f_p_s_k)507 int ostei_f_p_s_k(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__f_p_s_k)
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_f_p_k_s(P, Q_tmp, screen_tol, work, INT__f_p_s_k);
529 double buffer[1080] SIMINT_ALIGN_ARRAY_DBL;
530
531 for(int q = 0; q < ret; q++)
532 {
533 int idx = 0;
534 for(int a = 0; a < 10; ++a)
535 for(int b = 0; b < 3; ++b)
536 for(int c = 0; c < 1; ++c)
537 for(int d = 0; d < 36; ++d)
538 buffer[idx++] = INT__f_p_s_k[q*1080+a*108+b*36+d*1+c];
539
540 memcpy(INT__f_p_s_k+q*1080, buffer, 1080*sizeof(double));
541 }
542
543 return ret;
544 }
545
ostei_p_f_s_k(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__p_f_s_k)546 int ostei_p_f_s_k(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__p_f_s_k)
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_f_p_k_s(P_tmp, Q_tmp, screen_tol, work, INT__p_f_s_k);
583 double buffer[1080] SIMINT_ALIGN_ARRAY_DBL;
584
585 for(int q = 0; q < ret; q++)
586 {
587 int idx = 0;
588 for(int a = 0; a < 3; ++a)
589 for(int b = 0; b < 10; ++b)
590 for(int c = 0; c < 1; ++c)
591 for(int d = 0; d < 36; ++d)
592 buffer[idx++] = INT__p_f_s_k[q*1080+b*108+a*36+d*1+c];
593
594 memcpy(INT__p_f_s_k+q*1080, buffer, 1080*sizeof(double));
595 }
596
597 return ret;
598 }
599
600