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_i_s_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_i_s_s)8 int ostei_k_i_s_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_i_s_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__k_i_s_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_s_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__l_s_s_s = work + (SIMINT_NSHELL_SIMD * 36);
30 double * const INT__m_s_s_s = work + (SIMINT_NSHELL_SIMD * 81);
31 double * const INT__n_s_s_s = work + (SIMINT_NSHELL_SIMD * 136);
32 double * const INT__o_s_s_s = work + (SIMINT_NSHELL_SIMD * 202);
33 double * const INT__q_s_s_s = work + (SIMINT_NSHELL_SIMD * 280);
34 double * const INT__r_s_s_s = work + (SIMINT_NSHELL_SIMD * 371);
35 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*476);
36 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_s_s = primwork + 14;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 53;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 125;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 235;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 385;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 574;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_s_s = primwork + 798;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_s_s = primwork + 1050;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_s_s = primwork + 1320;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_s_s = primwork + 1595;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__o_s_s_s = primwork + 1859;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__q_s_s_s = primwork + 2093;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__r_s_s_s = primwork + 2275;
50 double * const hrrwork = (double *)(primwork + 2380);
51 double * const HRR_INT__k_p_s_s = hrrwork + 0;
52 double * const HRR_INT__k_d_s_s = hrrwork + 108;
53 double * const HRR_INT__k_f_s_s = hrrwork + 324;
54 double * const HRR_INT__k_g_s_s = hrrwork + 684;
55 double * const HRR_INT__k_h_s_s = hrrwork + 1224;
56 double * const HRR_INT__l_p_s_s = hrrwork + 1980;
57 double * const HRR_INT__l_d_s_s = hrrwork + 2115;
58 double * const HRR_INT__l_f_s_s = hrrwork + 2385;
59 double * const HRR_INT__l_g_s_s = hrrwork + 2835;
60 double * const HRR_INT__l_h_s_s = hrrwork + 3510;
61 double * const HRR_INT__m_p_s_s = hrrwork + 4455;
62 double * const HRR_INT__m_d_s_s = hrrwork + 4620;
63 double * const HRR_INT__m_f_s_s = hrrwork + 4950;
64 double * const HRR_INT__m_g_s_s = hrrwork + 5500;
65 double * const HRR_INT__n_p_s_s = hrrwork + 6325;
66 double * const HRR_INT__n_d_s_s = hrrwork + 6523;
67 double * const HRR_INT__n_f_s_s = hrrwork + 6919;
68 double * const HRR_INT__o_p_s_s = hrrwork + 7579;
69 double * const HRR_INT__o_d_s_s = hrrwork + 7813;
70 double * const HRR_INT__q_p_s_s = hrrwork + 8281;
71
72
73 // Create constants
74 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
75 const SIMINT_DBLTYPE const_10 = SIMINT_DBLSET1(10);
76 const SIMINT_DBLTYPE const_11 = SIMINT_DBLSET1(11);
77 const SIMINT_DBLTYPE const_12 = SIMINT_DBLSET1(12);
78 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
79 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
80 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
81 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
82 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
83 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
84 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
85 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
86 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
87
88
89 ////////////////////////////////////////
90 // Loop over shells and primitives
91 ////////////////////////////////////////
92
93 real_abcd = 0;
94 istart = 0;
95 for(ab = 0; ab < P.nshell12_clip; ++ab)
96 {
97 const int iend = istart + P.nprim12[ab];
98
99 cd = 0;
100 jstart = 0;
101
102 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
103 {
104 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
105 int jend = jstart;
106 for(i = 0; i < nshellbatch; i++)
107 jend += Q.nprim12[cd+i];
108
109 // Clear the beginning of the workspace (where we are accumulating integrals)
110 memset(work, 0, SIMINT_NSHELL_SIMD * 476 * sizeof(double));
111 abcd = 0;
112
113
114 for(i = istart; i < iend; ++i)
115 {
116 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
117
118 if(check_screen)
119 {
120 // Skip this whole thing if always insignificant
121 if((P.screen[i] * Q.screen_max) < screen_tol)
122 continue;
123 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
124 }
125
126 icd = 0;
127 iprimcd = 0;
128 nprim_icd = Q.nprim12[cd];
129 double * restrict PRIM_PTR_INT__k_s_s_s = INT__k_s_s_s + abcd * 36;
130 double * restrict PRIM_PTR_INT__l_s_s_s = INT__l_s_s_s + abcd * 45;
131 double * restrict PRIM_PTR_INT__m_s_s_s = INT__m_s_s_s + abcd * 55;
132 double * restrict PRIM_PTR_INT__n_s_s_s = INT__n_s_s_s + abcd * 66;
133 double * restrict PRIM_PTR_INT__o_s_s_s = INT__o_s_s_s + abcd * 78;
134 double * restrict PRIM_PTR_INT__q_s_s_s = INT__q_s_s_s + abcd * 91;
135 double * restrict PRIM_PTR_INT__r_s_s_s = INT__r_s_s_s + abcd * 105;
136
137
138
139 // Load these one per loop over i
140 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
141 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
142 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
143
144 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
145
146 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
147 {
148 // calculate the shell offsets
149 // these are the offset from the shell pointed to by cd
150 // for each element
151 int shelloffsets[SIMINT_SIMD_LEN] = {0};
152 int lastoffset = 0;
153 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
154
155 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
156 {
157 // Handle if the first element of the vector is a new shell
158 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
159 {
160 nprim_icd += Q.nprim12[cd + (++icd)];
161 PRIM_PTR_INT__k_s_s_s += 36;
162 PRIM_PTR_INT__l_s_s_s += 45;
163 PRIM_PTR_INT__m_s_s_s += 55;
164 PRIM_PTR_INT__n_s_s_s += 66;
165 PRIM_PTR_INT__o_s_s_s += 78;
166 PRIM_PTR_INT__q_s_s_s += 91;
167 PRIM_PTR_INT__r_s_s_s += 105;
168 }
169 iprimcd++;
170 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
171 {
172 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
173 {
174 shelloffsets[n] = shelloffsets[n-1] + 1;
175 lastoffset++;
176 nprim_icd += Q.nprim12[cd + (++icd)];
177 }
178 else
179 shelloffsets[n] = shelloffsets[n-1];
180 iprimcd++;
181 }
182 }
183 else
184 iprimcd += SIMINT_SIMD_LEN;
185
186 // Do we have to compute this vector (or has it been screened out)?
187 // (not_screened != 0 means we have to do this vector)
188 if(check_screen)
189 {
190 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
191 if(vmax < screen_tol)
192 {
193 PRIM_PTR_INT__k_s_s_s += lastoffset*36;
194 PRIM_PTR_INT__l_s_s_s += lastoffset*45;
195 PRIM_PTR_INT__m_s_s_s += lastoffset*55;
196 PRIM_PTR_INT__n_s_s_s += lastoffset*66;
197 PRIM_PTR_INT__o_s_s_s += lastoffset*78;
198 PRIM_PTR_INT__q_s_s_s += lastoffset*91;
199 PRIM_PTR_INT__r_s_s_s += lastoffset*105;
200 continue;
201 }
202 }
203
204 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
205 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
206 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
207 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
208
209
210 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
211 SIMINT_DBLTYPE PQ[3];
212 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
213 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
214 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
215 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
216 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
217 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
218
219 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
220 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
221 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
222 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
223 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
224 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
225
226 // NOTE: Minus sign!
227 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
228 SIMINT_DBLTYPE aop_PQ[3];
229 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
230 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
231 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
232
233
234 //////////////////////////////////////////////
235 // Fjt function section
236 // Maximum v value: 13
237 //////////////////////////////////////////////
238 // The parameter to the Fjt function
239 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
240
241
242 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
243
244
245 boys_F_split(PRIM_INT__s_s_s_s, F_x, 13);
246 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
247 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
248 for(n = 0; n <= 13; n++)
249 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
250
251 //////////////////////////////////////////////
252 // Primitive integrals: Vertical recurrance
253 //////////////////////////////////////////////
254
255 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
256 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
257 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
258 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
259 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
260 const SIMINT_DBLTYPE vrr_const_6_over_2p = SIMINT_MUL(const_6, one_over_2p);
261 const SIMINT_DBLTYPE vrr_const_7_over_2p = SIMINT_MUL(const_7, one_over_2p);
262 const SIMINT_DBLTYPE vrr_const_8_over_2p = SIMINT_MUL(const_8, one_over_2p);
263 const SIMINT_DBLTYPE vrr_const_9_over_2p = SIMINT_MUL(const_9, one_over_2p);
264 const SIMINT_DBLTYPE vrr_const_10_over_2p = SIMINT_MUL(const_10, one_over_2p);
265 const SIMINT_DBLTYPE vrr_const_11_over_2p = SIMINT_MUL(const_11, one_over_2p);
266 const SIMINT_DBLTYPE vrr_const_12_over_2p = SIMINT_MUL(const_12, one_over_2p);
267
268
269
270 // Forming PRIM_INT__p_s_s_s[13 * 3];
271 for(n = 0; n < 13; ++n) // loop over orders of auxiliary function
272 {
273
274 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
275 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]);
276
277 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
278 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]);
279
280 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
281 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]);
282
283 }
284
285
286
287 // Forming PRIM_INT__d_s_s_s[12 * 6];
288 for(n = 0; n < 12; ++n) // loop over orders of auxiliary function
289 {
290
291 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
292 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]);
293 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]);
294
295 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
296 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__p_s_s_s[(n+1) * 3 + 1], PRIM_INT__d_s_s_s[n * 6 + 3]);
297 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_FMADD( vrr_const_1_over_2p, SIMINT_FMADD(a_over_p, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__d_s_s_s[n * 6 + 3]);
298
299 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
300 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__p_s_s_s[(n+1) * 3 + 2], PRIM_INT__d_s_s_s[n * 6 + 5]);
301 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_FMADD( vrr_const_1_over_2p, SIMINT_FMADD(a_over_p, PRIM_INT__s_s_s_s[(n+1) * 1 + 0], PRIM_INT__s_s_s_s[n * 1 + 0]), PRIM_INT__d_s_s_s[n * 6 + 5]);
302
303 }
304
305
306
307 // Forming PRIM_INT__f_s_s_s[11 * 10];
308 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
309 {
310
311 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
312 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_FMADD( aop_PQ[0], PRIM_INT__d_s_s_s[(n+1) * 6 + 0], PRIM_INT__f_s_s_s[n * 10 + 0]);
313 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_FMADD( vrr_const_2_over_2p, SIMINT_FMADD(a_over_p, PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__p_s_s_s[n * 3 + 0]), PRIM_INT__f_s_s_s[n * 10 + 0]);
314
315 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
316 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]);
317 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]);
318
319 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
320 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]);
321 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]);
322
323 }
324
325
326 VRR_I_g_s_s_s(
327 PRIM_INT__g_s_s_s,
328 PRIM_INT__f_s_s_s,
329 PRIM_INT__d_s_s_s,
330 P_PA,
331 a_over_p,
332 aop_PQ,
333 one_over_2p,
334 10);
335
336
337 VRR_I_h_s_s_s(
338 PRIM_INT__h_s_s_s,
339 PRIM_INT__g_s_s_s,
340 PRIM_INT__f_s_s_s,
341 P_PA,
342 a_over_p,
343 aop_PQ,
344 one_over_2p,
345 9);
346
347
348 ostei_general_vrr1_I(6, 8,
349 one_over_2p, a_over_p, aop_PQ, P_PA,
350 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
351
352
353 ostei_general_vrr1_I(7, 7,
354 one_over_2p, a_over_p, aop_PQ, P_PA,
355 PRIM_INT__i_s_s_s, PRIM_INT__h_s_s_s, PRIM_INT__k_s_s_s);
356
357
358 ostei_general_vrr1_I(8, 6,
359 one_over_2p, a_over_p, aop_PQ, P_PA,
360 PRIM_INT__k_s_s_s, PRIM_INT__i_s_s_s, PRIM_INT__l_s_s_s);
361
362
363 ostei_general_vrr1_I(9, 5,
364 one_over_2p, a_over_p, aop_PQ, P_PA,
365 PRIM_INT__l_s_s_s, PRIM_INT__k_s_s_s, PRIM_INT__m_s_s_s);
366
367
368 ostei_general_vrr1_I(10, 4,
369 one_over_2p, a_over_p, aop_PQ, P_PA,
370 PRIM_INT__m_s_s_s, PRIM_INT__l_s_s_s, PRIM_INT__n_s_s_s);
371
372
373 ostei_general_vrr1_I(11, 3,
374 one_over_2p, a_over_p, aop_PQ, P_PA,
375 PRIM_INT__n_s_s_s, PRIM_INT__m_s_s_s, PRIM_INT__o_s_s_s);
376
377
378 ostei_general_vrr1_I(12, 2,
379 one_over_2p, a_over_p, aop_PQ, P_PA,
380 PRIM_INT__o_s_s_s, PRIM_INT__n_s_s_s, PRIM_INT__q_s_s_s);
381
382
383 ostei_general_vrr1_I(13, 1,
384 one_over_2p, a_over_p, aop_PQ, P_PA,
385 PRIM_INT__q_s_s_s, PRIM_INT__o_s_s_s, PRIM_INT__r_s_s_s);
386
387
388
389
390 ////////////////////////////////////
391 // Accumulate contracted integrals
392 ////////////////////////////////////
393 if(lastoffset == 0)
394 {
395 contract_all(36, PRIM_INT__k_s_s_s, PRIM_PTR_INT__k_s_s_s);
396 contract_all(45, PRIM_INT__l_s_s_s, PRIM_PTR_INT__l_s_s_s);
397 contract_all(55, PRIM_INT__m_s_s_s, PRIM_PTR_INT__m_s_s_s);
398 contract_all(66, PRIM_INT__n_s_s_s, PRIM_PTR_INT__n_s_s_s);
399 contract_all(78, PRIM_INT__o_s_s_s, PRIM_PTR_INT__o_s_s_s);
400 contract_all(91, PRIM_INT__q_s_s_s, PRIM_PTR_INT__q_s_s_s);
401 contract_all(105, PRIM_INT__r_s_s_s, PRIM_PTR_INT__r_s_s_s);
402 }
403 else
404 {
405 contract(36, shelloffsets, PRIM_INT__k_s_s_s, PRIM_PTR_INT__k_s_s_s);
406 contract(45, shelloffsets, PRIM_INT__l_s_s_s, PRIM_PTR_INT__l_s_s_s);
407 contract(55, shelloffsets, PRIM_INT__m_s_s_s, PRIM_PTR_INT__m_s_s_s);
408 contract(66, shelloffsets, PRIM_INT__n_s_s_s, PRIM_PTR_INT__n_s_s_s);
409 contract(78, shelloffsets, PRIM_INT__o_s_s_s, PRIM_PTR_INT__o_s_s_s);
410 contract(91, shelloffsets, PRIM_INT__q_s_s_s, PRIM_PTR_INT__q_s_s_s);
411 contract(105, shelloffsets, PRIM_INT__r_s_s_s, PRIM_PTR_INT__r_s_s_s);
412 PRIM_PTR_INT__k_s_s_s += lastoffset*36;
413 PRIM_PTR_INT__l_s_s_s += lastoffset*45;
414 PRIM_PTR_INT__m_s_s_s += lastoffset*55;
415 PRIM_PTR_INT__n_s_s_s += lastoffset*66;
416 PRIM_PTR_INT__o_s_s_s += lastoffset*78;
417 PRIM_PTR_INT__q_s_s_s += lastoffset*91;
418 PRIM_PTR_INT__r_s_s_s += lastoffset*105;
419 }
420
421 } // close loop over j
422 } // close loop over i
423
424 //Advance to the next batch
425 jstart = SIMINT_SIMD_ROUND(jend);
426
427 //////////////////////////////////////////////
428 // Contracted integrals: Horizontal recurrance
429 //////////////////////////////////////////////
430
431
432 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
433
434
435 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
436 {
437
438 // set up HRR pointers
439 double const * restrict HRR_INT__k_s_s_s = INT__k_s_s_s + abcd * 36;
440 double const * restrict HRR_INT__l_s_s_s = INT__l_s_s_s + abcd * 45;
441 double const * restrict HRR_INT__m_s_s_s = INT__m_s_s_s + abcd * 55;
442 double const * restrict HRR_INT__n_s_s_s = INT__n_s_s_s + abcd * 66;
443 double const * restrict HRR_INT__o_s_s_s = INT__o_s_s_s + abcd * 78;
444 double const * restrict HRR_INT__q_s_s_s = INT__q_s_s_s + abcd * 91;
445 double const * restrict HRR_INT__r_s_s_s = INT__r_s_s_s + abcd * 105;
446 double * restrict HRR_INT__k_i_s_s = INT__k_i_s_s + real_abcd * 1008;
447
448 // form INT__k_p_s_s
449 ostei_general_hrr_J(7, 1, 0, 0, hAB, HRR_INT__l_s_s_s, HRR_INT__k_s_s_s, HRR_INT__k_p_s_s);
450
451 // form INT__l_p_s_s
452 ostei_general_hrr_J(8, 1, 0, 0, hAB, HRR_INT__m_s_s_s, HRR_INT__l_s_s_s, HRR_INT__l_p_s_s);
453
454 // form INT__m_p_s_s
455 ostei_general_hrr_J(9, 1, 0, 0, hAB, HRR_INT__n_s_s_s, HRR_INT__m_s_s_s, HRR_INT__m_p_s_s);
456
457 // form INT__n_p_s_s
458 ostei_general_hrr_J(10, 1, 0, 0, hAB, HRR_INT__o_s_s_s, HRR_INT__n_s_s_s, HRR_INT__n_p_s_s);
459
460 // form INT__o_p_s_s
461 ostei_general_hrr_J(11, 1, 0, 0, hAB, HRR_INT__q_s_s_s, HRR_INT__o_s_s_s, HRR_INT__o_p_s_s);
462
463 // form INT__q_p_s_s
464 ostei_general_hrr_J(12, 1, 0, 0, hAB, HRR_INT__r_s_s_s, HRR_INT__q_s_s_s, HRR_INT__q_p_s_s);
465
466 // form INT__k_d_s_s
467 ostei_general_hrr_J(7, 2, 0, 0, hAB, HRR_INT__l_p_s_s, HRR_INT__k_p_s_s, HRR_INT__k_d_s_s);
468
469 // form INT__l_d_s_s
470 ostei_general_hrr_J(8, 2, 0, 0, hAB, HRR_INT__m_p_s_s, HRR_INT__l_p_s_s, HRR_INT__l_d_s_s);
471
472 // form INT__m_d_s_s
473 ostei_general_hrr_J(9, 2, 0, 0, hAB, HRR_INT__n_p_s_s, HRR_INT__m_p_s_s, HRR_INT__m_d_s_s);
474
475 // form INT__n_d_s_s
476 ostei_general_hrr_J(10, 2, 0, 0, hAB, HRR_INT__o_p_s_s, HRR_INT__n_p_s_s, HRR_INT__n_d_s_s);
477
478 // form INT__o_d_s_s
479 ostei_general_hrr_J(11, 2, 0, 0, hAB, HRR_INT__q_p_s_s, HRR_INT__o_p_s_s, HRR_INT__o_d_s_s);
480
481 // form INT__k_f_s_s
482 ostei_general_hrr_J(7, 3, 0, 0, hAB, HRR_INT__l_d_s_s, HRR_INT__k_d_s_s, HRR_INT__k_f_s_s);
483
484 // form INT__l_f_s_s
485 ostei_general_hrr_J(8, 3, 0, 0, hAB, HRR_INT__m_d_s_s, HRR_INT__l_d_s_s, HRR_INT__l_f_s_s);
486
487 // form INT__m_f_s_s
488 ostei_general_hrr_J(9, 3, 0, 0, hAB, HRR_INT__n_d_s_s, HRR_INT__m_d_s_s, HRR_INT__m_f_s_s);
489
490 // form INT__n_f_s_s
491 ostei_general_hrr_J(10, 3, 0, 0, hAB, HRR_INT__o_d_s_s, HRR_INT__n_d_s_s, HRR_INT__n_f_s_s);
492
493 // form INT__k_g_s_s
494 ostei_general_hrr_J(7, 4, 0, 0, hAB, HRR_INT__l_f_s_s, HRR_INT__k_f_s_s, HRR_INT__k_g_s_s);
495
496 // form INT__l_g_s_s
497 ostei_general_hrr_J(8, 4, 0, 0, hAB, HRR_INT__m_f_s_s, HRR_INT__l_f_s_s, HRR_INT__l_g_s_s);
498
499 // form INT__m_g_s_s
500 ostei_general_hrr_J(9, 4, 0, 0, hAB, HRR_INT__n_f_s_s, HRR_INT__m_f_s_s, HRR_INT__m_g_s_s);
501
502 // form INT__k_h_s_s
503 ostei_general_hrr_J(7, 5, 0, 0, hAB, HRR_INT__l_g_s_s, HRR_INT__k_g_s_s, HRR_INT__k_h_s_s);
504
505 // form INT__l_h_s_s
506 ostei_general_hrr_J(8, 5, 0, 0, hAB, HRR_INT__m_g_s_s, HRR_INT__l_g_s_s, HRR_INT__l_h_s_s);
507
508 // form INT__k_i_s_s
509 ostei_general_hrr_J(7, 6, 0, 0, hAB, HRR_INT__l_h_s_s, HRR_INT__k_h_s_s, HRR_INT__k_i_s_s);
510
511
512 } // close HRR loop
513
514
515 } // close loop cdbatch
516
517 istart = iend;
518 } // close loop over ab
519
520 return P.nshell12_clip * Q.nshell12_clip;
521 }
522
ostei_i_k_s_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_k_s_s)523 int ostei_i_k_s_s(struct simint_multi_shellpair const P,
524 struct simint_multi_shellpair const Q,
525 double screen_tol,
526 double * const restrict work,
527 double * const restrict INT__i_k_s_s)
528 {
529 double P_AB[3*P.nshell12];
530 struct simint_multi_shellpair P_tmp = P;
531 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
532 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
533 P_tmp.AB_x = P_AB;
534 P_tmp.AB_y = P_AB + P.nshell12;
535 P_tmp.AB_z = P_AB + 2*P.nshell12;
536
537 for(int i = 0; i < P.nshell12; i++)
538 {
539 P_tmp.AB_x[i] = -P.AB_x[i];
540 P_tmp.AB_y[i] = -P.AB_y[i];
541 P_tmp.AB_z[i] = -P.AB_z[i];
542 }
543
544 int ret = ostei_k_i_s_s(P_tmp, Q, screen_tol, work, INT__i_k_s_s);
545 double buffer[1008] SIMINT_ALIGN_ARRAY_DBL;
546
547 for(int q = 0; q < ret; q++)
548 {
549 int idx = 0;
550 for(int a = 0; a < 28; ++a)
551 for(int b = 0; b < 36; ++b)
552 for(int c = 0; c < 1; ++c)
553 for(int d = 0; d < 1; ++d)
554 buffer[idx++] = INT__i_k_s_s[q*1008+b*28+a*1+c*1+d];
555
556 memcpy(INT__i_k_s_s+q*1008, buffer, 1008*sizeof(double));
557 }
558
559 return ret;
560 }
561
562