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_h_d_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__i_h_d_s)8 int ostei_i_h_d_s(struct simint_multi_shellpair const P,
9 struct simint_multi_shellpair const Q,
10 double screen_tol,
11 double * const restrict work,
12 double * const restrict INT__i_h_d_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__i_h_d_s);
17 int ab, cd, abcd;
18 int istart, jstart;
19 int iprimcd, nprim_icd, icd;
20 const int check_screen = (screen_tol > 0.0);
21 int i, j;
22 int n;
23 int not_screened;
24 int real_abcd;
25 int iket;
26
27 // partition workspace
28 double * const INT__i_s_d_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__k_s_d_s = work + (SIMINT_NSHELL_SIMD * 168);
30 double * const INT__l_s_d_s = work + (SIMINT_NSHELL_SIMD * 384);
31 double * const INT__m_s_d_s = work + (SIMINT_NSHELL_SIMD * 654);
32 double * const INT__n_s_d_s = work + (SIMINT_NSHELL_SIMD * 984);
33 double * const INT__o_s_d_s = work + (SIMINT_NSHELL_SIMD * 1380);
34 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*1848);
35 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_s_s = primwork + 14;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 53;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 125;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 235;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 385;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_p_s = primwork + 574;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 700;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_p_s = primwork + 924;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_d_s = primwork + 1092;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_s_s = primwork + 1260;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_p_s = primwork + 1512;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_d_s = primwork + 1728;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_s_s = primwork + 1944;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_p_s = primwork + 2214;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_d_s = primwork + 2484;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_s_s = primwork + 2754;
52 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_p_s = primwork + 3029;
53 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_d_s = primwork + 3359;
54 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_s_s = primwork + 3689;
55 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_p_s = primwork + 3953;
56 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_d_s = primwork + 4349;
57 SIMINT_DBLTYPE * const restrict PRIM_INT__o_s_s_s = primwork + 4745;
58 SIMINT_DBLTYPE * const restrict PRIM_INT__o_s_p_s = primwork + 4979;
59 SIMINT_DBLTYPE * const restrict PRIM_INT__o_s_d_s = primwork + 5447;
60 double * const hrrwork = (double *)(primwork + 5915);
61 double * const HRR_INT__i_p_d_s = hrrwork + 0;
62 double * const HRR_INT__i_d_d_s = hrrwork + 504;
63 double * const HRR_INT__i_f_d_s = hrrwork + 1512;
64 double * const HRR_INT__i_g_d_s = hrrwork + 3192;
65 double * const HRR_INT__k_p_d_s = hrrwork + 5712;
66 double * const HRR_INT__k_d_d_s = hrrwork + 6360;
67 double * const HRR_INT__k_f_d_s = hrrwork + 7656;
68 double * const HRR_INT__k_g_d_s = hrrwork + 9816;
69 double * const HRR_INT__l_p_d_s = hrrwork + 13056;
70 double * const HRR_INT__l_d_d_s = hrrwork + 13866;
71 double * const HRR_INT__l_f_d_s = hrrwork + 15486;
72 double * const HRR_INT__m_p_d_s = hrrwork + 18186;
73 double * const HRR_INT__m_d_d_s = hrrwork + 19176;
74 double * const HRR_INT__n_p_d_s = hrrwork + 21156;
75
76
77 // Create constants
78 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
79 const SIMINT_DBLTYPE const_10 = SIMINT_DBLSET1(10);
80 const SIMINT_DBLTYPE const_11 = SIMINT_DBLSET1(11);
81 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
82 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
83 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
84 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
85 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
86 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
87 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
88 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
89 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
90
91
92 ////////////////////////////////////////
93 // Loop over shells and primitives
94 ////////////////////////////////////////
95
96 real_abcd = 0;
97 istart = 0;
98 for(ab = 0; ab < P.nshell12_clip; ++ab)
99 {
100 const int iend = istart + P.nprim12[ab];
101
102 cd = 0;
103 jstart = 0;
104
105 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
106 {
107 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
108 int jend = jstart;
109 for(i = 0; i < nshellbatch; i++)
110 jend += Q.nprim12[cd+i];
111
112 // Clear the beginning of the workspace (where we are accumulating integrals)
113 memset(work, 0, SIMINT_NSHELL_SIMD * 1848 * sizeof(double));
114 abcd = 0;
115
116
117 for(i = istart; i < iend; ++i)
118 {
119 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
120
121 if(check_screen)
122 {
123 // Skip this whole thing if always insignificant
124 if((P.screen[i] * Q.screen_max) < screen_tol)
125 continue;
126 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
127 }
128
129 icd = 0;
130 iprimcd = 0;
131 nprim_icd = Q.nprim12[cd];
132 double * restrict PRIM_PTR_INT__i_s_d_s = INT__i_s_d_s + abcd * 168;
133 double * restrict PRIM_PTR_INT__k_s_d_s = INT__k_s_d_s + abcd * 216;
134 double * restrict PRIM_PTR_INT__l_s_d_s = INT__l_s_d_s + abcd * 270;
135 double * restrict PRIM_PTR_INT__m_s_d_s = INT__m_s_d_s + abcd * 330;
136 double * restrict PRIM_PTR_INT__n_s_d_s = INT__n_s_d_s + abcd * 396;
137 double * restrict PRIM_PTR_INT__o_s_d_s = INT__o_s_d_s + abcd * 468;
138
139
140
141 // Load these one per loop over i
142 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
143 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
144 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
145
146 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
147
148 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
149 {
150 // calculate the shell offsets
151 // these are the offset from the shell pointed to by cd
152 // for each element
153 int shelloffsets[SIMINT_SIMD_LEN] = {0};
154 int lastoffset = 0;
155 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
156
157 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
158 {
159 // Handle if the first element of the vector is a new shell
160 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
161 {
162 nprim_icd += Q.nprim12[cd + (++icd)];
163 PRIM_PTR_INT__i_s_d_s += 168;
164 PRIM_PTR_INT__k_s_d_s += 216;
165 PRIM_PTR_INT__l_s_d_s += 270;
166 PRIM_PTR_INT__m_s_d_s += 330;
167 PRIM_PTR_INT__n_s_d_s += 396;
168 PRIM_PTR_INT__o_s_d_s += 468;
169 }
170 iprimcd++;
171 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
172 {
173 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
174 {
175 shelloffsets[n] = shelloffsets[n-1] + 1;
176 lastoffset++;
177 nprim_icd += Q.nprim12[cd + (++icd)];
178 }
179 else
180 shelloffsets[n] = shelloffsets[n-1];
181 iprimcd++;
182 }
183 }
184 else
185 iprimcd += SIMINT_SIMD_LEN;
186
187 // Do we have to compute this vector (or has it been screened out)?
188 // (not_screened != 0 means we have to do this vector)
189 if(check_screen)
190 {
191 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
192 if(vmax < screen_tol)
193 {
194 PRIM_PTR_INT__i_s_d_s += lastoffset*168;
195 PRIM_PTR_INT__k_s_d_s += lastoffset*216;
196 PRIM_PTR_INT__l_s_d_s += lastoffset*270;
197 PRIM_PTR_INT__m_s_d_s += lastoffset*330;
198 PRIM_PTR_INT__n_s_d_s += lastoffset*396;
199 PRIM_PTR_INT__o_s_d_s += lastoffset*468;
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 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
226
227 // NOTE: Minus sign!
228 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
229 SIMINT_DBLTYPE aop_PQ[3];
230 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
231 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
232 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
233
234 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
235 SIMINT_DBLTYPE aoq_PQ[3];
236 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
237 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
238 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
239 // Put a minus sign here so we don't have to in RR routines
240 a_over_q = SIMINT_NEG(a_over_q);
241
242
243 //////////////////////////////////////////////
244 // Fjt function section
245 // Maximum v value: 13
246 //////////////////////////////////////////////
247 // The parameter to the Fjt function
248 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
249
250
251 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
252
253
254 boys_F_split(PRIM_INT__s_s_s_s, F_x, 13);
255 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
256 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
257 for(n = 0; n <= 13; n++)
258 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
259
260 //////////////////////////////////////////////
261 // Primitive integrals: Vertical recurrance
262 //////////////////////////////////////////////
263
264 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
265 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
266 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
267 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
268 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
269 const SIMINT_DBLTYPE vrr_const_6_over_2p = SIMINT_MUL(const_6, one_over_2p);
270 const SIMINT_DBLTYPE vrr_const_7_over_2p = SIMINT_MUL(const_7, one_over_2p);
271 const SIMINT_DBLTYPE vrr_const_8_over_2p = SIMINT_MUL(const_8, one_over_2p);
272 const SIMINT_DBLTYPE vrr_const_9_over_2p = SIMINT_MUL(const_9, one_over_2p);
273 const SIMINT_DBLTYPE vrr_const_10_over_2p = SIMINT_MUL(const_10, one_over_2p);
274 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
275 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
276 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
277 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
278 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
279 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
280 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
281 const SIMINT_DBLTYPE vrr_const_7_over_2pq = SIMINT_MUL(const_7, one_over_2pq);
282 const SIMINT_DBLTYPE vrr_const_8_over_2pq = SIMINT_MUL(const_8, one_over_2pq);
283 const SIMINT_DBLTYPE vrr_const_9_over_2pq = SIMINT_MUL(const_9, one_over_2pq);
284 const SIMINT_DBLTYPE vrr_const_10_over_2pq = SIMINT_MUL(const_10, one_over_2pq);
285 const SIMINT_DBLTYPE vrr_const_11_over_2pq = SIMINT_MUL(const_11, one_over_2pq);
286
287
288
289 // Forming PRIM_INT__p_s_s_s[13 * 3];
290 for(n = 0; n < 13; ++n) // loop over orders of auxiliary function
291 {
292
293 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
294 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]);
295
296 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
297 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]);
298
299 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
300 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]);
301
302 }
303
304
305
306 // Forming PRIM_INT__d_s_s_s[12 * 6];
307 for(n = 0; n < 12; ++n) // loop over orders of auxiliary function
308 {
309
310 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
311 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]);
312 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]);
313
314 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
315 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]);
316 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]);
317
318 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
319 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]);
320 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]);
321
322 }
323
324
325
326 // Forming PRIM_INT__f_s_s_s[11 * 10];
327 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
328 {
329
330 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
331 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]);
332 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]);
333
334 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
335 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]);
336
337 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
338 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]);
339
340 PRIM_INT__f_s_s_s[n * 10 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
341 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]);
342
343 PRIM_INT__f_s_s_s[n * 10 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
344 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]);
345
346 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
347 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]);
348 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]);
349
350 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
351 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]);
352
353 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
354 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]);
355 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]);
356
357 }
358
359
360 VRR_I_g_s_s_s(
361 PRIM_INT__g_s_s_s,
362 PRIM_INT__f_s_s_s,
363 PRIM_INT__d_s_s_s,
364 P_PA,
365 a_over_p,
366 aop_PQ,
367 one_over_2p,
368 10);
369
370
371 VRR_I_h_s_s_s(
372 PRIM_INT__h_s_s_s,
373 PRIM_INT__g_s_s_s,
374 PRIM_INT__f_s_s_s,
375 P_PA,
376 a_over_p,
377 aop_PQ,
378 one_over_2p,
379 9);
380
381
382 ostei_general_vrr1_I(6, 8,
383 one_over_2p, a_over_p, aop_PQ, P_PA,
384 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
385
386
387 ostei_general_vrr_K(6, 0, 1, 0, 2,
388 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
389 PRIM_INT__i_s_s_s, NULL, NULL, PRIM_INT__h_s_s_s, NULL, PRIM_INT__i_s_p_s);
390
391
392 ostei_general_vrr_K(5, 0, 1, 0, 2,
393 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
394 PRIM_INT__h_s_s_s, NULL, NULL, PRIM_INT__g_s_s_s, NULL, PRIM_INT__h_s_p_s);
395
396
397 ostei_general_vrr_K(6, 0, 2, 0, 1,
398 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
399 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);
400
401
402 ostei_general_vrr1_I(7, 7,
403 one_over_2p, a_over_p, aop_PQ, P_PA,
404 PRIM_INT__i_s_s_s, PRIM_INT__h_s_s_s, PRIM_INT__k_s_s_s);
405
406
407 ostei_general_vrr_K(7, 0, 1, 0, 2,
408 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
409 PRIM_INT__k_s_s_s, NULL, NULL, PRIM_INT__i_s_s_s, NULL, PRIM_INT__k_s_p_s);
410
411
412 ostei_general_vrr_K(7, 0, 2, 0, 1,
413 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
414 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);
415
416
417 ostei_general_vrr1_I(8, 6,
418 one_over_2p, a_over_p, aop_PQ, P_PA,
419 PRIM_INT__k_s_s_s, PRIM_INT__i_s_s_s, PRIM_INT__l_s_s_s);
420
421
422 ostei_general_vrr_K(8, 0, 1, 0, 2,
423 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
424 PRIM_INT__l_s_s_s, NULL, NULL, PRIM_INT__k_s_s_s, NULL, PRIM_INT__l_s_p_s);
425
426
427 ostei_general_vrr_K(8, 0, 2, 0, 1,
428 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
429 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);
430
431
432 ostei_general_vrr1_I(9, 5,
433 one_over_2p, a_over_p, aop_PQ, P_PA,
434 PRIM_INT__l_s_s_s, PRIM_INT__k_s_s_s, PRIM_INT__m_s_s_s);
435
436
437 ostei_general_vrr_K(9, 0, 1, 0, 2,
438 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
439 PRIM_INT__m_s_s_s, NULL, NULL, PRIM_INT__l_s_s_s, NULL, PRIM_INT__m_s_p_s);
440
441
442 ostei_general_vrr_K(9, 0, 2, 0, 1,
443 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
444 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);
445
446
447 ostei_general_vrr1_I(10, 4,
448 one_over_2p, a_over_p, aop_PQ, P_PA,
449 PRIM_INT__m_s_s_s, PRIM_INT__l_s_s_s, PRIM_INT__n_s_s_s);
450
451
452 ostei_general_vrr_K(10, 0, 1, 0, 2,
453 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
454 PRIM_INT__n_s_s_s, NULL, NULL, PRIM_INT__m_s_s_s, NULL, PRIM_INT__n_s_p_s);
455
456
457 ostei_general_vrr_K(10, 0, 2, 0, 1,
458 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
459 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);
460
461
462 ostei_general_vrr1_I(11, 3,
463 one_over_2p, a_over_p, aop_PQ, P_PA,
464 PRIM_INT__n_s_s_s, PRIM_INT__m_s_s_s, PRIM_INT__o_s_s_s);
465
466
467 ostei_general_vrr_K(11, 0, 1, 0, 2,
468 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
469 PRIM_INT__o_s_s_s, NULL, NULL, PRIM_INT__n_s_s_s, NULL, PRIM_INT__o_s_p_s);
470
471
472 ostei_general_vrr_K(11, 0, 2, 0, 1,
473 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
474 PRIM_INT__o_s_p_s, PRIM_INT__o_s_s_s, NULL, PRIM_INT__n_s_p_s, NULL, PRIM_INT__o_s_d_s);
475
476
477
478
479 ////////////////////////////////////
480 // Accumulate contracted integrals
481 ////////////////////////////////////
482 if(lastoffset == 0)
483 {
484 contract_all(168, PRIM_INT__i_s_d_s, PRIM_PTR_INT__i_s_d_s);
485 contract_all(216, PRIM_INT__k_s_d_s, PRIM_PTR_INT__k_s_d_s);
486 contract_all(270, PRIM_INT__l_s_d_s, PRIM_PTR_INT__l_s_d_s);
487 contract_all(330, PRIM_INT__m_s_d_s, PRIM_PTR_INT__m_s_d_s);
488 contract_all(396, PRIM_INT__n_s_d_s, PRIM_PTR_INT__n_s_d_s);
489 contract_all(468, PRIM_INT__o_s_d_s, PRIM_PTR_INT__o_s_d_s);
490 }
491 else
492 {
493 contract(168, shelloffsets, PRIM_INT__i_s_d_s, PRIM_PTR_INT__i_s_d_s);
494 contract(216, shelloffsets, PRIM_INT__k_s_d_s, PRIM_PTR_INT__k_s_d_s);
495 contract(270, shelloffsets, PRIM_INT__l_s_d_s, PRIM_PTR_INT__l_s_d_s);
496 contract(330, shelloffsets, PRIM_INT__m_s_d_s, PRIM_PTR_INT__m_s_d_s);
497 contract(396, shelloffsets, PRIM_INT__n_s_d_s, PRIM_PTR_INT__n_s_d_s);
498 contract(468, shelloffsets, PRIM_INT__o_s_d_s, PRIM_PTR_INT__o_s_d_s);
499 PRIM_PTR_INT__i_s_d_s += lastoffset*168;
500 PRIM_PTR_INT__k_s_d_s += lastoffset*216;
501 PRIM_PTR_INT__l_s_d_s += lastoffset*270;
502 PRIM_PTR_INT__m_s_d_s += lastoffset*330;
503 PRIM_PTR_INT__n_s_d_s += lastoffset*396;
504 PRIM_PTR_INT__o_s_d_s += lastoffset*468;
505 }
506
507 } // close loop over j
508 } // close loop over i
509
510 //Advance to the next batch
511 jstart = SIMINT_SIMD_ROUND(jend);
512
513 //////////////////////////////////////////////
514 // Contracted integrals: Horizontal recurrance
515 //////////////////////////////////////////////
516
517
518 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
519
520
521 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
522 {
523
524 // set up HRR pointers
525 double const * restrict HRR_INT__i_s_d_s = INT__i_s_d_s + abcd * 168;
526 double const * restrict HRR_INT__k_s_d_s = INT__k_s_d_s + abcd * 216;
527 double const * restrict HRR_INT__l_s_d_s = INT__l_s_d_s + abcd * 270;
528 double const * restrict HRR_INT__m_s_d_s = INT__m_s_d_s + abcd * 330;
529 double const * restrict HRR_INT__n_s_d_s = INT__n_s_d_s + abcd * 396;
530 double const * restrict HRR_INT__o_s_d_s = INT__o_s_d_s + abcd * 468;
531 double * restrict HRR_INT__i_h_d_s = INT__i_h_d_s + real_abcd * 3528;
532
533 // form INT__i_p_d_s
534 ostei_general_hrr_J(6, 1, 2, 0, hAB, HRR_INT__k_s_d_s, HRR_INT__i_s_d_s, HRR_INT__i_p_d_s);
535
536 // form INT__k_p_d_s
537 ostei_general_hrr_J(7, 1, 2, 0, hAB, HRR_INT__l_s_d_s, HRR_INT__k_s_d_s, HRR_INT__k_p_d_s);
538
539 // form INT__l_p_d_s
540 ostei_general_hrr_J(8, 1, 2, 0, hAB, HRR_INT__m_s_d_s, HRR_INT__l_s_d_s, HRR_INT__l_p_d_s);
541
542 // form INT__m_p_d_s
543 ostei_general_hrr_J(9, 1, 2, 0, hAB, HRR_INT__n_s_d_s, HRR_INT__m_s_d_s, HRR_INT__m_p_d_s);
544
545 // form INT__n_p_d_s
546 ostei_general_hrr_J(10, 1, 2, 0, hAB, HRR_INT__o_s_d_s, HRR_INT__n_s_d_s, HRR_INT__n_p_d_s);
547
548 // form INT__i_d_d_s
549 ostei_general_hrr_J(6, 2, 2, 0, hAB, HRR_INT__k_p_d_s, HRR_INT__i_p_d_s, HRR_INT__i_d_d_s);
550
551 // form INT__k_d_d_s
552 ostei_general_hrr_J(7, 2, 2, 0, hAB, HRR_INT__l_p_d_s, HRR_INT__k_p_d_s, HRR_INT__k_d_d_s);
553
554 // form INT__l_d_d_s
555 ostei_general_hrr_J(8, 2, 2, 0, hAB, HRR_INT__m_p_d_s, HRR_INT__l_p_d_s, HRR_INT__l_d_d_s);
556
557 // form INT__m_d_d_s
558 ostei_general_hrr_J(9, 2, 2, 0, hAB, HRR_INT__n_p_d_s, HRR_INT__m_p_d_s, HRR_INT__m_d_d_s);
559
560 // form INT__i_f_d_s
561 ostei_general_hrr_J(6, 3, 2, 0, hAB, HRR_INT__k_d_d_s, HRR_INT__i_d_d_s, HRR_INT__i_f_d_s);
562
563 // form INT__k_f_d_s
564 ostei_general_hrr_J(7, 3, 2, 0, hAB, HRR_INT__l_d_d_s, HRR_INT__k_d_d_s, HRR_INT__k_f_d_s);
565
566 // form INT__l_f_d_s
567 ostei_general_hrr_J(8, 3, 2, 0, hAB, HRR_INT__m_d_d_s, HRR_INT__l_d_d_s, HRR_INT__l_f_d_s);
568
569 // form INT__i_g_d_s
570 ostei_general_hrr_J(6, 4, 2, 0, hAB, HRR_INT__k_f_d_s, HRR_INT__i_f_d_s, HRR_INT__i_g_d_s);
571
572 // form INT__k_g_d_s
573 ostei_general_hrr_J(7, 4, 2, 0, hAB, HRR_INT__l_f_d_s, HRR_INT__k_f_d_s, HRR_INT__k_g_d_s);
574
575 // form INT__i_h_d_s
576 ostei_general_hrr_J(6, 5, 2, 0, hAB, HRR_INT__k_g_d_s, HRR_INT__i_g_d_s, HRR_INT__i_h_d_s);
577
578
579 } // close HRR loop
580
581
582 } // close loop cdbatch
583
584 istart = iend;
585 } // close loop over ab
586
587 return P.nshell12_clip * Q.nshell12_clip;
588 }
589
ostei_h_i_d_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__h_i_d_s)590 int ostei_h_i_d_s(struct simint_multi_shellpair const P,
591 struct simint_multi_shellpair const Q,
592 double screen_tol,
593 double * const restrict work,
594 double * const restrict INT__h_i_d_s)
595 {
596 double P_AB[3*P.nshell12];
597 struct simint_multi_shellpair P_tmp = P;
598 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
599 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
600 P_tmp.AB_x = P_AB;
601 P_tmp.AB_y = P_AB + P.nshell12;
602 P_tmp.AB_z = P_AB + 2*P.nshell12;
603
604 for(int i = 0; i < P.nshell12; i++)
605 {
606 P_tmp.AB_x[i] = -P.AB_x[i];
607 P_tmp.AB_y[i] = -P.AB_y[i];
608 P_tmp.AB_z[i] = -P.AB_z[i];
609 }
610
611 int ret = ostei_i_h_d_s(P_tmp, Q, screen_tol, work, INT__h_i_d_s);
612 double buffer[3528] SIMINT_ALIGN_ARRAY_DBL;
613
614 for(int q = 0; q < ret; q++)
615 {
616 int idx = 0;
617 for(int a = 0; a < 21; ++a)
618 for(int b = 0; b < 28; ++b)
619 for(int c = 0; c < 6; ++c)
620 for(int d = 0; d < 1; ++d)
621 buffer[idx++] = INT__h_i_d_s[q*3528+b*126+a*6+c*1+d];
622
623 memcpy(INT__h_i_d_s+q*3528, buffer, 3528*sizeof(double));
624 }
625
626 return ret;
627 }
628
ostei_i_h_s_d(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__i_h_s_d)629 int ostei_i_h_s_d(struct simint_multi_shellpair const P,
630 struct simint_multi_shellpair const Q,
631 double screen_tol,
632 double * const restrict work,
633 double * const restrict INT__i_h_s_d)
634 {
635 double Q_AB[3*Q.nshell12];
636 struct simint_multi_shellpair Q_tmp = Q;
637 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
638 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
639 Q_tmp.AB_x = Q_AB;
640 Q_tmp.AB_y = Q_AB + Q.nshell12;
641 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
642
643 for(int i = 0; i < Q.nshell12; i++)
644 {
645 Q_tmp.AB_x[i] = -Q.AB_x[i];
646 Q_tmp.AB_y[i] = -Q.AB_y[i];
647 Q_tmp.AB_z[i] = -Q.AB_z[i];
648 }
649
650 int ret = ostei_i_h_d_s(P, Q_tmp, screen_tol, work, INT__i_h_s_d);
651 double buffer[3528] SIMINT_ALIGN_ARRAY_DBL;
652
653 for(int q = 0; q < ret; q++)
654 {
655 int idx = 0;
656 for(int a = 0; a < 28; ++a)
657 for(int b = 0; b < 21; ++b)
658 for(int c = 0; c < 1; ++c)
659 for(int d = 0; d < 6; ++d)
660 buffer[idx++] = INT__i_h_s_d[q*3528+a*126+b*6+d*1+c];
661
662 memcpy(INT__i_h_s_d+q*3528, buffer, 3528*sizeof(double));
663 }
664
665 return ret;
666 }
667
ostei_h_i_s_d(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__h_i_s_d)668 int ostei_h_i_s_d(struct simint_multi_shellpair const P,
669 struct simint_multi_shellpair const Q,
670 double screen_tol,
671 double * const restrict work,
672 double * const restrict INT__h_i_s_d)
673 {
674 double P_AB[3*P.nshell12];
675 struct simint_multi_shellpair P_tmp = P;
676 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
677 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
678 P_tmp.AB_x = P_AB;
679 P_tmp.AB_y = P_AB + P.nshell12;
680 P_tmp.AB_z = P_AB + 2*P.nshell12;
681
682 for(int i = 0; i < P.nshell12; i++)
683 {
684 P_tmp.AB_x[i] = -P.AB_x[i];
685 P_tmp.AB_y[i] = -P.AB_y[i];
686 P_tmp.AB_z[i] = -P.AB_z[i];
687 }
688
689 double Q_AB[3*Q.nshell12];
690 struct simint_multi_shellpair Q_tmp = Q;
691 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
692 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
693 Q_tmp.AB_x = Q_AB;
694 Q_tmp.AB_y = Q_AB + Q.nshell12;
695 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
696
697 for(int i = 0; i < Q.nshell12; i++)
698 {
699 Q_tmp.AB_x[i] = -Q.AB_x[i];
700 Q_tmp.AB_y[i] = -Q.AB_y[i];
701 Q_tmp.AB_z[i] = -Q.AB_z[i];
702 }
703
704 int ret = ostei_i_h_d_s(P_tmp, Q_tmp, screen_tol, work, INT__h_i_s_d);
705 double buffer[3528] SIMINT_ALIGN_ARRAY_DBL;
706
707 for(int q = 0; q < ret; q++)
708 {
709 int idx = 0;
710 for(int a = 0; a < 21; ++a)
711 for(int b = 0; b < 28; ++b)
712 for(int c = 0; c < 1; ++c)
713 for(int d = 0; d < 6; ++d)
714 buffer[idx++] = INT__h_i_s_d[q*3528+b*126+a*6+d*1+c];
715
716 memcpy(INT__h_i_s_d+q*3528, buffer, 3528*sizeof(double));
717 }
718
719 return ret;
720 }
721
722