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_g_p_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_g_p_s)8 int ostei_i_g_p_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_g_p_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__i_g_p_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_p_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__k_s_p_s = work + (SIMINT_NSHELL_SIMD * 84);
30 double * const INT__l_s_p_s = work + (SIMINT_NSHELL_SIMD * 192);
31 double * const INT__m_s_p_s = work + (SIMINT_NSHELL_SIMD * 327);
32 double * const INT__n_s_p_s = work + (SIMINT_NSHELL_SIMD * 492);
33 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*690);
34 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_s_s = primwork + 12;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 45;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 105;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 195;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 315;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 462;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_p_s = primwork + 630;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_s_s = primwork + 714;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_p_s = primwork + 894;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_s_s = primwork + 1002;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_p_s = primwork + 1182;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_s_s = primwork + 1317;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_p_s = primwork + 1482;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_s_s = primwork + 1647;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_p_s = primwork + 1779;
50 double * const hrrwork = (double *)(primwork + 1977);
51 double * const HRR_INT__i_p_p_s = hrrwork + 0;
52 double * const HRR_INT__i_d_p_s = hrrwork + 252;
53 double * const HRR_INT__i_f_p_s = hrrwork + 756;
54 double * const HRR_INT__k_p_p_s = hrrwork + 1596;
55 double * const HRR_INT__k_d_p_s = hrrwork + 1920;
56 double * const HRR_INT__k_f_p_s = hrrwork + 2568;
57 double * const HRR_INT__l_p_p_s = hrrwork + 3648;
58 double * const HRR_INT__l_d_p_s = hrrwork + 4053;
59 double * const HRR_INT__m_p_p_s = hrrwork + 4863;
60
61
62 // Create constants
63 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
64 const SIMINT_DBLTYPE const_10 = SIMINT_DBLSET1(10);
65 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
66 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
67 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
68 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
69 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
70 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
71 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
72 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
73 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
74
75
76 ////////////////////////////////////////
77 // Loop over shells and primitives
78 ////////////////////////////////////////
79
80 real_abcd = 0;
81 istart = 0;
82 for(ab = 0; ab < P.nshell12_clip; ++ab)
83 {
84 const int iend = istart + P.nprim12[ab];
85
86 cd = 0;
87 jstart = 0;
88
89 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
90 {
91 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
92 int jend = jstart;
93 for(i = 0; i < nshellbatch; i++)
94 jend += Q.nprim12[cd+i];
95
96 // Clear the beginning of the workspace (where we are accumulating integrals)
97 memset(work, 0, SIMINT_NSHELL_SIMD * 690 * sizeof(double));
98 abcd = 0;
99
100
101 for(i = istart; i < iend; ++i)
102 {
103 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
104
105 if(check_screen)
106 {
107 // Skip this whole thing if always insignificant
108 if((P.screen[i] * Q.screen_max) < screen_tol)
109 continue;
110 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
111 }
112
113 icd = 0;
114 iprimcd = 0;
115 nprim_icd = Q.nprim12[cd];
116 double * restrict PRIM_PTR_INT__i_s_p_s = INT__i_s_p_s + abcd * 84;
117 double * restrict PRIM_PTR_INT__k_s_p_s = INT__k_s_p_s + abcd * 108;
118 double * restrict PRIM_PTR_INT__l_s_p_s = INT__l_s_p_s + abcd * 135;
119 double * restrict PRIM_PTR_INT__m_s_p_s = INT__m_s_p_s + abcd * 165;
120 double * restrict PRIM_PTR_INT__n_s_p_s = INT__n_s_p_s + abcd * 198;
121
122
123
124 // Load these one per loop over i
125 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
126 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
127 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
128
129 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
130
131 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
132 {
133 // calculate the shell offsets
134 // these are the offset from the shell pointed to by cd
135 // for each element
136 int shelloffsets[SIMINT_SIMD_LEN] = {0};
137 int lastoffset = 0;
138 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
139
140 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
141 {
142 // Handle if the first element of the vector is a new shell
143 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
144 {
145 nprim_icd += Q.nprim12[cd + (++icd)];
146 PRIM_PTR_INT__i_s_p_s += 84;
147 PRIM_PTR_INT__k_s_p_s += 108;
148 PRIM_PTR_INT__l_s_p_s += 135;
149 PRIM_PTR_INT__m_s_p_s += 165;
150 PRIM_PTR_INT__n_s_p_s += 198;
151 }
152 iprimcd++;
153 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
154 {
155 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
156 {
157 shelloffsets[n] = shelloffsets[n-1] + 1;
158 lastoffset++;
159 nprim_icd += Q.nprim12[cd + (++icd)];
160 }
161 else
162 shelloffsets[n] = shelloffsets[n-1];
163 iprimcd++;
164 }
165 }
166 else
167 iprimcd += SIMINT_SIMD_LEN;
168
169 // Do we have to compute this vector (or has it been screened out)?
170 // (not_screened != 0 means we have to do this vector)
171 if(check_screen)
172 {
173 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
174 if(vmax < screen_tol)
175 {
176 PRIM_PTR_INT__i_s_p_s += lastoffset*84;
177 PRIM_PTR_INT__k_s_p_s += lastoffset*108;
178 PRIM_PTR_INT__l_s_p_s += lastoffset*135;
179 PRIM_PTR_INT__m_s_p_s += lastoffset*165;
180 PRIM_PTR_INT__n_s_p_s += lastoffset*198;
181 continue;
182 }
183 }
184
185 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
186 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
187 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
188 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
189
190
191 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
192 SIMINT_DBLTYPE PQ[3];
193 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
194 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
195 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
196 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
197 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
198 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
199
200 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
201 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
202 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
203 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
204 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
205 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
206 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
207
208 // NOTE: Minus sign!
209 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
210 SIMINT_DBLTYPE aop_PQ[3];
211 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
212 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
213 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
214
215 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
216 SIMINT_DBLTYPE aoq_PQ[3];
217 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
218 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
219 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
220 // Put a minus sign here so we don't have to in RR routines
221 a_over_q = SIMINT_NEG(a_over_q);
222
223
224 //////////////////////////////////////////////
225 // Fjt function section
226 // Maximum v value: 11
227 //////////////////////////////////////////////
228 // The parameter to the Fjt function
229 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
230
231
232 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
233
234
235 boys_F_split(PRIM_INT__s_s_s_s, F_x, 11);
236 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
237 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
238 for(n = 0; n <= 11; n++)
239 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
240
241 //////////////////////////////////////////////
242 // Primitive integrals: Vertical recurrance
243 //////////////////////////////////////////////
244
245 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
246 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
247 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
248 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
249 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
250 const SIMINT_DBLTYPE vrr_const_6_over_2p = SIMINT_MUL(const_6, one_over_2p);
251 const SIMINT_DBLTYPE vrr_const_7_over_2p = SIMINT_MUL(const_7, one_over_2p);
252 const SIMINT_DBLTYPE vrr_const_8_over_2p = SIMINT_MUL(const_8, one_over_2p);
253 const SIMINT_DBLTYPE vrr_const_9_over_2p = SIMINT_MUL(const_9, one_over_2p);
254 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
255 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
256 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
257 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
258 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
259 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
260 const SIMINT_DBLTYPE vrr_const_7_over_2pq = SIMINT_MUL(const_7, one_over_2pq);
261 const SIMINT_DBLTYPE vrr_const_8_over_2pq = SIMINT_MUL(const_8, one_over_2pq);
262 const SIMINT_DBLTYPE vrr_const_9_over_2pq = SIMINT_MUL(const_9, one_over_2pq);
263 const SIMINT_DBLTYPE vrr_const_10_over_2pq = SIMINT_MUL(const_10, one_over_2pq);
264
265
266
267 // Forming PRIM_INT__p_s_s_s[11 * 3];
268 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
269 {
270
271 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
272 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]);
273
274 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
275 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]);
276
277 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
278 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]);
279
280 }
281
282
283
284 // Forming PRIM_INT__d_s_s_s[10 * 6];
285 for(n = 0; n < 10; ++n) // loop over orders of auxiliary function
286 {
287
288 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
289 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]);
290 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]);
291
292 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
293 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]);
294 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]);
295
296 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
297 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]);
298 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]);
299
300 }
301
302
303
304 // Forming PRIM_INT__f_s_s_s[9 * 10];
305 for(n = 0; n < 9; ++n) // loop over orders of auxiliary function
306 {
307
308 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
309 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]);
310 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]);
311
312 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
313 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]);
314
315 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
316 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]);
317
318 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
319 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]);
320 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]);
321
322 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
323 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]);
324
325 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
326 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]);
327 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]);
328
329 }
330
331
332 VRR_I_g_s_s_s(
333 PRIM_INT__g_s_s_s,
334 PRIM_INT__f_s_s_s,
335 PRIM_INT__d_s_s_s,
336 P_PA,
337 a_over_p,
338 aop_PQ,
339 one_over_2p,
340 8);
341
342
343 VRR_I_h_s_s_s(
344 PRIM_INT__h_s_s_s,
345 PRIM_INT__g_s_s_s,
346 PRIM_INT__f_s_s_s,
347 P_PA,
348 a_over_p,
349 aop_PQ,
350 one_over_2p,
351 7);
352
353
354 ostei_general_vrr1_I(6, 6,
355 one_over_2p, a_over_p, aop_PQ, P_PA,
356 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
357
358
359 ostei_general_vrr_K(6, 0, 1, 0, 1,
360 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
361 PRIM_INT__i_s_s_s, NULL, NULL, PRIM_INT__h_s_s_s, NULL, PRIM_INT__i_s_p_s);
362
363
364 ostei_general_vrr1_I(7, 5,
365 one_over_2p, a_over_p, aop_PQ, P_PA,
366 PRIM_INT__i_s_s_s, PRIM_INT__h_s_s_s, PRIM_INT__k_s_s_s);
367
368
369 ostei_general_vrr_K(7, 0, 1, 0, 1,
370 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
371 PRIM_INT__k_s_s_s, NULL, NULL, PRIM_INT__i_s_s_s, NULL, PRIM_INT__k_s_p_s);
372
373
374 ostei_general_vrr1_I(8, 4,
375 one_over_2p, a_over_p, aop_PQ, P_PA,
376 PRIM_INT__k_s_s_s, PRIM_INT__i_s_s_s, PRIM_INT__l_s_s_s);
377
378
379 ostei_general_vrr_K(8, 0, 1, 0, 1,
380 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
381 PRIM_INT__l_s_s_s, NULL, NULL, PRIM_INT__k_s_s_s, NULL, PRIM_INT__l_s_p_s);
382
383
384 ostei_general_vrr1_I(9, 3,
385 one_over_2p, a_over_p, aop_PQ, P_PA,
386 PRIM_INT__l_s_s_s, PRIM_INT__k_s_s_s, PRIM_INT__m_s_s_s);
387
388
389 ostei_general_vrr_K(9, 0, 1, 0, 1,
390 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
391 PRIM_INT__m_s_s_s, NULL, NULL, PRIM_INT__l_s_s_s, NULL, PRIM_INT__m_s_p_s);
392
393
394 ostei_general_vrr1_I(10, 2,
395 one_over_2p, a_over_p, aop_PQ, P_PA,
396 PRIM_INT__m_s_s_s, PRIM_INT__l_s_s_s, PRIM_INT__n_s_s_s);
397
398
399 ostei_general_vrr_K(10, 0, 1, 0, 1,
400 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
401 PRIM_INT__n_s_s_s, NULL, NULL, PRIM_INT__m_s_s_s, NULL, PRIM_INT__n_s_p_s);
402
403
404
405
406 ////////////////////////////////////
407 // Accumulate contracted integrals
408 ////////////////////////////////////
409 if(lastoffset == 0)
410 {
411 contract_all(84, PRIM_INT__i_s_p_s, PRIM_PTR_INT__i_s_p_s);
412 contract_all(108, PRIM_INT__k_s_p_s, PRIM_PTR_INT__k_s_p_s);
413 contract_all(135, PRIM_INT__l_s_p_s, PRIM_PTR_INT__l_s_p_s);
414 contract_all(165, PRIM_INT__m_s_p_s, PRIM_PTR_INT__m_s_p_s);
415 contract_all(198, PRIM_INT__n_s_p_s, PRIM_PTR_INT__n_s_p_s);
416 }
417 else
418 {
419 contract(84, shelloffsets, PRIM_INT__i_s_p_s, PRIM_PTR_INT__i_s_p_s);
420 contract(108, shelloffsets, PRIM_INT__k_s_p_s, PRIM_PTR_INT__k_s_p_s);
421 contract(135, shelloffsets, PRIM_INT__l_s_p_s, PRIM_PTR_INT__l_s_p_s);
422 contract(165, shelloffsets, PRIM_INT__m_s_p_s, PRIM_PTR_INT__m_s_p_s);
423 contract(198, shelloffsets, PRIM_INT__n_s_p_s, PRIM_PTR_INT__n_s_p_s);
424 PRIM_PTR_INT__i_s_p_s += lastoffset*84;
425 PRIM_PTR_INT__k_s_p_s += lastoffset*108;
426 PRIM_PTR_INT__l_s_p_s += lastoffset*135;
427 PRIM_PTR_INT__m_s_p_s += lastoffset*165;
428 PRIM_PTR_INT__n_s_p_s += lastoffset*198;
429 }
430
431 } // close loop over j
432 } // close loop over i
433
434 //Advance to the next batch
435 jstart = SIMINT_SIMD_ROUND(jend);
436
437 //////////////////////////////////////////////
438 // Contracted integrals: Horizontal recurrance
439 //////////////////////////////////////////////
440
441
442 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
443
444
445 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
446 {
447
448 // set up HRR pointers
449 double const * restrict HRR_INT__i_s_p_s = INT__i_s_p_s + abcd * 84;
450 double const * restrict HRR_INT__k_s_p_s = INT__k_s_p_s + abcd * 108;
451 double const * restrict HRR_INT__l_s_p_s = INT__l_s_p_s + abcd * 135;
452 double const * restrict HRR_INT__m_s_p_s = INT__m_s_p_s + abcd * 165;
453 double const * restrict HRR_INT__n_s_p_s = INT__n_s_p_s + abcd * 198;
454 double * restrict HRR_INT__i_g_p_s = INT__i_g_p_s + real_abcd * 1260;
455
456 // form INT__i_p_p_s
457 ostei_general_hrr_J(6, 1, 1, 0, hAB, HRR_INT__k_s_p_s, HRR_INT__i_s_p_s, HRR_INT__i_p_p_s);
458
459 // form INT__k_p_p_s
460 ostei_general_hrr_J(7, 1, 1, 0, hAB, HRR_INT__l_s_p_s, HRR_INT__k_s_p_s, HRR_INT__k_p_p_s);
461
462 // form INT__l_p_p_s
463 ostei_general_hrr_J(8, 1, 1, 0, hAB, HRR_INT__m_s_p_s, HRR_INT__l_s_p_s, HRR_INT__l_p_p_s);
464
465 // form INT__m_p_p_s
466 ostei_general_hrr_J(9, 1, 1, 0, hAB, HRR_INT__n_s_p_s, HRR_INT__m_s_p_s, HRR_INT__m_p_p_s);
467
468 // form INT__i_d_p_s
469 ostei_general_hrr_J(6, 2, 1, 0, hAB, HRR_INT__k_p_p_s, HRR_INT__i_p_p_s, HRR_INT__i_d_p_s);
470
471 // form INT__k_d_p_s
472 ostei_general_hrr_J(7, 2, 1, 0, hAB, HRR_INT__l_p_p_s, HRR_INT__k_p_p_s, HRR_INT__k_d_p_s);
473
474 // form INT__l_d_p_s
475 ostei_general_hrr_J(8, 2, 1, 0, hAB, HRR_INT__m_p_p_s, HRR_INT__l_p_p_s, HRR_INT__l_d_p_s);
476
477 // form INT__i_f_p_s
478 ostei_general_hrr_J(6, 3, 1, 0, hAB, HRR_INT__k_d_p_s, HRR_INT__i_d_p_s, HRR_INT__i_f_p_s);
479
480 // form INT__k_f_p_s
481 ostei_general_hrr_J(7, 3, 1, 0, hAB, HRR_INT__l_d_p_s, HRR_INT__k_d_p_s, HRR_INT__k_f_p_s);
482
483 // form INT__i_g_p_s
484 ostei_general_hrr_J(6, 4, 1, 0, hAB, HRR_INT__k_f_p_s, HRR_INT__i_f_p_s, HRR_INT__i_g_p_s);
485
486
487 } // close HRR loop
488
489
490 } // close loop cdbatch
491
492 istart = iend;
493 } // close loop over ab
494
495 return P.nshell12_clip * Q.nshell12_clip;
496 }
497
ostei_g_i_p_s(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__g_i_p_s)498 int ostei_g_i_p_s(struct simint_multi_shellpair const P,
499 struct simint_multi_shellpair const Q,
500 double screen_tol,
501 double * const restrict work,
502 double * const restrict INT__g_i_p_s)
503 {
504 double P_AB[3*P.nshell12];
505 struct simint_multi_shellpair P_tmp = P;
506 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
507 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
508 P_tmp.AB_x = P_AB;
509 P_tmp.AB_y = P_AB + P.nshell12;
510 P_tmp.AB_z = P_AB + 2*P.nshell12;
511
512 for(int i = 0; i < P.nshell12; i++)
513 {
514 P_tmp.AB_x[i] = -P.AB_x[i];
515 P_tmp.AB_y[i] = -P.AB_y[i];
516 P_tmp.AB_z[i] = -P.AB_z[i];
517 }
518
519 int ret = ostei_i_g_p_s(P_tmp, Q, screen_tol, work, INT__g_i_p_s);
520 double buffer[1260] SIMINT_ALIGN_ARRAY_DBL;
521
522 for(int q = 0; q < ret; q++)
523 {
524 int idx = 0;
525 for(int a = 0; a < 15; ++a)
526 for(int b = 0; b < 28; ++b)
527 for(int c = 0; c < 3; ++c)
528 for(int d = 0; d < 1; ++d)
529 buffer[idx++] = INT__g_i_p_s[q*1260+b*45+a*3+c*1+d];
530
531 memcpy(INT__g_i_p_s+q*1260, buffer, 1260*sizeof(double));
532 }
533
534 return ret;
535 }
536
ostei_i_g_s_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_g_s_p)537 int ostei_i_g_s_p(struct simint_multi_shellpair const P,
538 struct simint_multi_shellpair const Q,
539 double screen_tol,
540 double * const restrict work,
541 double * const restrict INT__i_g_s_p)
542 {
543 double Q_AB[3*Q.nshell12];
544 struct simint_multi_shellpair Q_tmp = Q;
545 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
546 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
547 Q_tmp.AB_x = Q_AB;
548 Q_tmp.AB_y = Q_AB + Q.nshell12;
549 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
550
551 for(int i = 0; i < Q.nshell12; i++)
552 {
553 Q_tmp.AB_x[i] = -Q.AB_x[i];
554 Q_tmp.AB_y[i] = -Q.AB_y[i];
555 Q_tmp.AB_z[i] = -Q.AB_z[i];
556 }
557
558 int ret = ostei_i_g_p_s(P, Q_tmp, screen_tol, work, INT__i_g_s_p);
559 double buffer[1260] SIMINT_ALIGN_ARRAY_DBL;
560
561 for(int q = 0; q < ret; q++)
562 {
563 int idx = 0;
564 for(int a = 0; a < 28; ++a)
565 for(int b = 0; b < 15; ++b)
566 for(int c = 0; c < 1; ++c)
567 for(int d = 0; d < 3; ++d)
568 buffer[idx++] = INT__i_g_s_p[q*1260+a*45+b*3+d*1+c];
569
570 memcpy(INT__i_g_s_p+q*1260, buffer, 1260*sizeof(double));
571 }
572
573 return ret;
574 }
575
ostei_g_i_s_p(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__g_i_s_p)576 int ostei_g_i_s_p(struct simint_multi_shellpair const P,
577 struct simint_multi_shellpair const Q,
578 double screen_tol,
579 double * const restrict work,
580 double * const restrict INT__g_i_s_p)
581 {
582 double P_AB[3*P.nshell12];
583 struct simint_multi_shellpair P_tmp = P;
584 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
585 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
586 P_tmp.AB_x = P_AB;
587 P_tmp.AB_y = P_AB + P.nshell12;
588 P_tmp.AB_z = P_AB + 2*P.nshell12;
589
590 for(int i = 0; i < P.nshell12; i++)
591 {
592 P_tmp.AB_x[i] = -P.AB_x[i];
593 P_tmp.AB_y[i] = -P.AB_y[i];
594 P_tmp.AB_z[i] = -P.AB_z[i];
595 }
596
597 double Q_AB[3*Q.nshell12];
598 struct simint_multi_shellpair Q_tmp = Q;
599 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
600 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
601 Q_tmp.AB_x = Q_AB;
602 Q_tmp.AB_y = Q_AB + Q.nshell12;
603 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
604
605 for(int i = 0; i < Q.nshell12; i++)
606 {
607 Q_tmp.AB_x[i] = -Q.AB_x[i];
608 Q_tmp.AB_y[i] = -Q.AB_y[i];
609 Q_tmp.AB_z[i] = -Q.AB_z[i];
610 }
611
612 int ret = ostei_i_g_p_s(P_tmp, Q_tmp, screen_tol, work, INT__g_i_s_p);
613 double buffer[1260] SIMINT_ALIGN_ARRAY_DBL;
614
615 for(int q = 0; q < ret; q++)
616 {
617 int idx = 0;
618 for(int a = 0; a < 15; ++a)
619 for(int b = 0; b < 28; ++b)
620 for(int c = 0; c < 1; ++c)
621 for(int d = 0; d < 3; ++d)
622 buffer[idx++] = INT__g_i_s_p[q*1260+b*45+a*3+d*1+c];
623
624 memcpy(INT__g_i_s_p+q*1260, buffer, 1260*sizeof(double));
625 }
626
627 return ret;
628 }
629
630