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_g_s_i_f(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__g_s_i_f)8 int ostei_g_s_i_f(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__g_s_i_f)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__g_s_i_f);
17 int ab, cd, abcd;
18 int istart, jstart;
19 int iprimcd, nprim_icd, icd;
20 const int check_screen = (screen_tol > 0.0);
21 int i, j;
22 int n;
23 int not_screened;
24 int real_abcd;
25 int ibra;
26
27 // partition workspace
28 double * const INT__g_s_i_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__g_s_k_s = work + (SIMINT_NSHELL_SIMD * 420);
30 double * const INT__g_s_l_s = work + (SIMINT_NSHELL_SIMD * 960);
31 double * const INT__g_s_m_s = work + (SIMINT_NSHELL_SIMD * 1635);
32 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*2460);
33 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_p_s = primwork + 14;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_d_s = primwork + 53;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_f_s = primwork + 125;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_g_s = primwork + 235;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_h_s = primwork + 385;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_i_s = primwork + 574;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_k_s = primwork + 798;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_l_s = primwork + 1050;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_m_s = primwork + 1320;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_f_s = primwork + 1595;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_g_s = primwork + 1715;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_h_s = primwork + 1895;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_i_s = primwork + 2147;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_k_s = primwork + 2483;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_l_s = primwork + 2915;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_m_s = primwork + 3455;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_g_s = primwork + 4115;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_h_s = primwork + 4385;
52 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_i_s = primwork + 4763;
53 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_k_s = primwork + 5267;
54 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_l_s = primwork + 5915;
55 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_m_s = primwork + 6725;
56 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_h_s = primwork + 7715;
57 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_i_s = primwork + 8135;
58 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_k_s = primwork + 8695;
59 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_l_s = primwork + 9415;
60 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_m_s = primwork + 10315;
61 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_i_s = primwork + 11415;
62 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_k_s = primwork + 11835;
63 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_l_s = primwork + 12375;
64 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_m_s = primwork + 13050;
65 double * const hrrwork = (double *)(primwork + 13875);
66 double * const HRR_INT__g_s_i_p = hrrwork + 0;
67 double * const HRR_INT__g_s_i_d = hrrwork + 1260;
68 double * const HRR_INT__g_s_k_p = hrrwork + 3780;
69 double * const HRR_INT__g_s_k_d = hrrwork + 5400;
70 double * const HRR_INT__g_s_l_p = hrrwork + 8640;
71
72
73 // Create constants
74 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
75 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
76 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
77 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
78 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
79 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
80 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
81 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
82 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
83 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
84
85
86 ////////////////////////////////////////
87 // Loop over shells and primitives
88 ////////////////////////////////////////
89
90 real_abcd = 0;
91 istart = 0;
92 for(ab = 0; ab < P.nshell12_clip; ++ab)
93 {
94 const int iend = istart + P.nprim12[ab];
95
96 cd = 0;
97 jstart = 0;
98
99 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
100 {
101 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
102 int jend = jstart;
103 for(i = 0; i < nshellbatch; i++)
104 jend += Q.nprim12[cd+i];
105
106 // Clear the beginning of the workspace (where we are accumulating integrals)
107 memset(work, 0, SIMINT_NSHELL_SIMD * 2460 * sizeof(double));
108 abcd = 0;
109
110
111 for(i = istart; i < iend; ++i)
112 {
113 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
114
115 if(check_screen)
116 {
117 // Skip this whole thing if always insignificant
118 if((P.screen[i] * Q.screen_max) < screen_tol)
119 continue;
120 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
121 }
122
123 icd = 0;
124 iprimcd = 0;
125 nprim_icd = Q.nprim12[cd];
126 double * restrict PRIM_PTR_INT__g_s_i_s = INT__g_s_i_s + abcd * 420;
127 double * restrict PRIM_PTR_INT__g_s_k_s = INT__g_s_k_s + abcd * 540;
128 double * restrict PRIM_PTR_INT__g_s_l_s = INT__g_s_l_s + abcd * 675;
129 double * restrict PRIM_PTR_INT__g_s_m_s = INT__g_s_m_s + abcd * 825;
130
131
132
133 // Load these one per loop over i
134 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
135 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
136 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
137
138 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
139
140 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
141 {
142 // calculate the shell offsets
143 // these are the offset from the shell pointed to by cd
144 // for each element
145 int shelloffsets[SIMINT_SIMD_LEN] = {0};
146 int lastoffset = 0;
147 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
148
149 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
150 {
151 // Handle if the first element of the vector is a new shell
152 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
153 {
154 nprim_icd += Q.nprim12[cd + (++icd)];
155 PRIM_PTR_INT__g_s_i_s += 420;
156 PRIM_PTR_INT__g_s_k_s += 540;
157 PRIM_PTR_INT__g_s_l_s += 675;
158 PRIM_PTR_INT__g_s_m_s += 825;
159 }
160 iprimcd++;
161 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
162 {
163 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
164 {
165 shelloffsets[n] = shelloffsets[n-1] + 1;
166 lastoffset++;
167 nprim_icd += Q.nprim12[cd + (++icd)];
168 }
169 else
170 shelloffsets[n] = shelloffsets[n-1];
171 iprimcd++;
172 }
173 }
174 else
175 iprimcd += SIMINT_SIMD_LEN;
176
177 // Do we have to compute this vector (or has it been screened out)?
178 // (not_screened != 0 means we have to do this vector)
179 if(check_screen)
180 {
181 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
182 if(vmax < screen_tol)
183 {
184 PRIM_PTR_INT__g_s_i_s += lastoffset*420;
185 PRIM_PTR_INT__g_s_k_s += lastoffset*540;
186 PRIM_PTR_INT__g_s_l_s += lastoffset*675;
187 PRIM_PTR_INT__g_s_m_s += lastoffset*825;
188 continue;
189 }
190 }
191
192 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
193 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
194 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
195 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
196
197
198 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
199 SIMINT_DBLTYPE PQ[3];
200 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
201 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
202 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
203 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
204 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
205 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
206
207 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
208 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
209 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
210 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
211 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
212 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
213 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
214
215 // NOTE: Minus sign!
216 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
217 SIMINT_DBLTYPE aop_PQ[3];
218 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
219 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
220 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
221
222 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
223 SIMINT_DBLTYPE aoq_PQ[3];
224 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
225 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
226 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
227 // Put a minus sign here so we don't have to in RR routines
228 a_over_q = SIMINT_NEG(a_over_q);
229
230
231 //////////////////////////////////////////////
232 // Fjt function section
233 // Maximum v value: 13
234 //////////////////////////////////////////////
235 // The parameter to the Fjt function
236 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
237
238
239 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
240
241
242 boys_F_split(PRIM_INT__s_s_s_s, F_x, 13);
243 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
244 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
245 for(n = 0; n <= 13; n++)
246 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
247
248 //////////////////////////////////////////////
249 // Primitive integrals: Vertical recurrance
250 //////////////////////////////////////////////
251
252 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
253 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
254 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
255 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
256 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
257 const SIMINT_DBLTYPE vrr_const_3_over_2q = SIMINT_MUL(const_3, one_over_2q);
258 const SIMINT_DBLTYPE vrr_const_4_over_2q = SIMINT_MUL(const_4, one_over_2q);
259 const SIMINT_DBLTYPE vrr_const_5_over_2q = SIMINT_MUL(const_5, one_over_2q);
260 const SIMINT_DBLTYPE vrr_const_6_over_2q = SIMINT_MUL(const_6, one_over_2q);
261 const SIMINT_DBLTYPE vrr_const_7_over_2q = SIMINT_MUL(const_7, one_over_2q);
262 const SIMINT_DBLTYPE vrr_const_8_over_2q = SIMINT_MUL(const_8, one_over_2q);
263 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
264 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
265 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
266 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
267 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
268 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
269 const SIMINT_DBLTYPE vrr_const_7_over_2pq = SIMINT_MUL(const_7, one_over_2pq);
270 const SIMINT_DBLTYPE vrr_const_8_over_2pq = SIMINT_MUL(const_8, one_over_2pq);
271 const SIMINT_DBLTYPE vrr_const_9_over_2pq = SIMINT_MUL(const_9, one_over_2pq);
272
273
274
275 // Forming PRIM_INT__s_s_p_s[13 * 3];
276 for(n = 0; n < 13; ++n) // loop over orders of auxiliary function
277 {
278
279 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
280 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]);
281
282 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
283 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]);
284
285 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
286 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]);
287
288 }
289
290
291
292 // Forming PRIM_INT__s_s_d_s[12 * 6];
293 for(n = 0; n < 12; ++n) // loop over orders of auxiliary function
294 {
295
296 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_p_s[n * 3 + 0]);
297 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]);
298 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]);
299
300 PRIM_INT__s_s_d_s[n * 6 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 0]);
301 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]);
302
303 PRIM_INT__s_s_d_s[n * 6 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 0]);
304 PRIM_INT__s_s_d_s[n * 6 + 2] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 0], PRIM_INT__s_s_d_s[n * 6 + 2]);
305
306 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 1]);
307 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]);
308 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]);
309
310 PRIM_INT__s_s_d_s[n * 6 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 1]);
311 PRIM_INT__s_s_d_s[n * 6 + 4] = SIMINT_FMADD( aoq_PQ[2], PRIM_INT__s_s_p_s[(n+1) * 3 + 1], PRIM_INT__s_s_d_s[n * 6 + 4]);
312
313 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 2]);
314 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]);
315 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]);
316
317 }
318
319
320
321 // Forming PRIM_INT__s_s_f_s[11 * 10];
322 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
323 {
324
325 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
326 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]);
327 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]);
328
329 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
330 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]);
331
332 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
333 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]);
334
335 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
336 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]);
337
338 PRIM_INT__s_s_f_s[n * 10 + 4] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 1]);
339 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]);
340
341 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
342 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]);
343
344 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
345 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]);
346 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]);
347
348 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
349 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]);
350
351 PRIM_INT__s_s_f_s[n * 10 + 8] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 5]);
352 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]);
353
354 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
355 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]);
356 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]);
357
358 }
359
360
361 VRR_K_s_s_g_s(
362 PRIM_INT__s_s_g_s,
363 PRIM_INT__s_s_f_s,
364 PRIM_INT__s_s_d_s,
365 Q_PA,
366 a_over_q,
367 aoq_PQ,
368 one_over_2q,
369 10);
370
371
372 VRR_K_s_s_h_s(
373 PRIM_INT__s_s_h_s,
374 PRIM_INT__s_s_g_s,
375 PRIM_INT__s_s_f_s,
376 Q_PA,
377 a_over_q,
378 aoq_PQ,
379 one_over_2q,
380 9);
381
382
383 ostei_general_vrr1_K(6, 8,
384 one_over_2q, a_over_q, aoq_PQ, Q_PA,
385 PRIM_INT__s_s_h_s, PRIM_INT__s_s_g_s, PRIM_INT__s_s_i_s);
386
387
388 ostei_general_vrr_I(1, 0, 6, 0, 4,
389 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
390 PRIM_INT__s_s_i_s, NULL, NULL, PRIM_INT__s_s_h_s, NULL, PRIM_INT__p_s_i_s);
391
392
393 ostei_general_vrr_I(1, 0, 5, 0, 4,
394 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
395 PRIM_INT__s_s_h_s, NULL, NULL, PRIM_INT__s_s_g_s, NULL, PRIM_INT__p_s_h_s);
396
397
398 ostei_general_vrr_I(2, 0, 6, 0, 3,
399 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
400 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);
401
402
403 VRR_I_p_s_g_s(
404 PRIM_INT__p_s_g_s,
405 PRIM_INT__s_s_g_s,
406 PRIM_INT__s_s_f_s,
407 P_PA,
408 aop_PQ,
409 one_over_2pq,
410 4);
411
412
413 ostei_general_vrr_I(2, 0, 5, 0, 3,
414 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
415 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);
416
417
418 ostei_general_vrr_I(3, 0, 6, 0, 2,
419 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
420 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);
421
422
423 VRR_I_p_s_f_s(
424 PRIM_INT__p_s_f_s,
425 PRIM_INT__s_s_f_s,
426 PRIM_INT__s_s_d_s,
427 P_PA,
428 aop_PQ,
429 one_over_2pq,
430 4);
431
432
433 ostei_general_vrr_I(2, 0, 4, 0, 3,
434 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
435 PRIM_INT__p_s_g_s, PRIM_INT__s_s_g_s, NULL, PRIM_INT__p_s_f_s, NULL, PRIM_INT__d_s_g_s);
436
437
438 ostei_general_vrr_I(3, 0, 5, 0, 2,
439 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
440 PRIM_INT__d_s_h_s, PRIM_INT__p_s_h_s, NULL, PRIM_INT__d_s_g_s, NULL, PRIM_INT__f_s_h_s);
441
442
443 ostei_general_vrr_I(4, 0, 6, 0, 1,
444 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
445 PRIM_INT__f_s_i_s, PRIM_INT__d_s_i_s, NULL, PRIM_INT__f_s_h_s, NULL, PRIM_INT__g_s_i_s);
446
447
448 ostei_general_vrr1_K(7, 7,
449 one_over_2q, a_over_q, aoq_PQ, Q_PA,
450 PRIM_INT__s_s_i_s, PRIM_INT__s_s_h_s, PRIM_INT__s_s_k_s);
451
452
453 ostei_general_vrr_I(1, 0, 7, 0, 4,
454 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
455 PRIM_INT__s_s_k_s, NULL, NULL, PRIM_INT__s_s_i_s, NULL, PRIM_INT__p_s_k_s);
456
457
458 ostei_general_vrr_I(2, 0, 7, 0, 3,
459 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
460 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);
461
462
463 ostei_general_vrr_I(3, 0, 7, 0, 2,
464 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
465 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);
466
467
468 ostei_general_vrr_I(4, 0, 7, 0, 1,
469 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
470 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);
471
472
473 ostei_general_vrr1_K(8, 6,
474 one_over_2q, a_over_q, aoq_PQ, Q_PA,
475 PRIM_INT__s_s_k_s, PRIM_INT__s_s_i_s, PRIM_INT__s_s_l_s);
476
477
478 ostei_general_vrr_I(1, 0, 8, 0, 4,
479 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
480 PRIM_INT__s_s_l_s, NULL, NULL, PRIM_INT__s_s_k_s, NULL, PRIM_INT__p_s_l_s);
481
482
483 ostei_general_vrr_I(2, 0, 8, 0, 3,
484 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
485 PRIM_INT__p_s_l_s, PRIM_INT__s_s_l_s, NULL, PRIM_INT__p_s_k_s, NULL, PRIM_INT__d_s_l_s);
486
487
488 ostei_general_vrr_I(3, 0, 8, 0, 2,
489 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
490 PRIM_INT__d_s_l_s, PRIM_INT__p_s_l_s, NULL, PRIM_INT__d_s_k_s, NULL, PRIM_INT__f_s_l_s);
491
492
493 ostei_general_vrr_I(4, 0, 8, 0, 1,
494 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
495 PRIM_INT__f_s_l_s, PRIM_INT__d_s_l_s, NULL, PRIM_INT__f_s_k_s, NULL, PRIM_INT__g_s_l_s);
496
497
498 ostei_general_vrr1_K(9, 5,
499 one_over_2q, a_over_q, aoq_PQ, Q_PA,
500 PRIM_INT__s_s_l_s, PRIM_INT__s_s_k_s, PRIM_INT__s_s_m_s);
501
502
503 ostei_general_vrr_I(1, 0, 9, 0, 4,
504 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
505 PRIM_INT__s_s_m_s, NULL, NULL, PRIM_INT__s_s_l_s, NULL, PRIM_INT__p_s_m_s);
506
507
508 ostei_general_vrr_I(2, 0, 9, 0, 3,
509 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
510 PRIM_INT__p_s_m_s, PRIM_INT__s_s_m_s, NULL, PRIM_INT__p_s_l_s, NULL, PRIM_INT__d_s_m_s);
511
512
513 ostei_general_vrr_I(3, 0, 9, 0, 2,
514 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
515 PRIM_INT__d_s_m_s, PRIM_INT__p_s_m_s, NULL, PRIM_INT__d_s_l_s, NULL, PRIM_INT__f_s_m_s);
516
517
518 ostei_general_vrr_I(4, 0, 9, 0, 1,
519 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
520 PRIM_INT__f_s_m_s, PRIM_INT__d_s_m_s, NULL, PRIM_INT__f_s_l_s, NULL, PRIM_INT__g_s_m_s);
521
522
523
524
525 ////////////////////////////////////
526 // Accumulate contracted integrals
527 ////////////////////////////////////
528 if(lastoffset == 0)
529 {
530 contract_all(420, PRIM_INT__g_s_i_s, PRIM_PTR_INT__g_s_i_s);
531 contract_all(540, PRIM_INT__g_s_k_s, PRIM_PTR_INT__g_s_k_s);
532 contract_all(675, PRIM_INT__g_s_l_s, PRIM_PTR_INT__g_s_l_s);
533 contract_all(825, PRIM_INT__g_s_m_s, PRIM_PTR_INT__g_s_m_s);
534 }
535 else
536 {
537 contract(420, shelloffsets, PRIM_INT__g_s_i_s, PRIM_PTR_INT__g_s_i_s);
538 contract(540, shelloffsets, PRIM_INT__g_s_k_s, PRIM_PTR_INT__g_s_k_s);
539 contract(675, shelloffsets, PRIM_INT__g_s_l_s, PRIM_PTR_INT__g_s_l_s);
540 contract(825, shelloffsets, PRIM_INT__g_s_m_s, PRIM_PTR_INT__g_s_m_s);
541 PRIM_PTR_INT__g_s_i_s += lastoffset*420;
542 PRIM_PTR_INT__g_s_k_s += lastoffset*540;
543 PRIM_PTR_INT__g_s_l_s += lastoffset*675;
544 PRIM_PTR_INT__g_s_m_s += lastoffset*825;
545 }
546
547 } // close loop over j
548 } // close loop over i
549
550 //Advance to the next batch
551 jstart = SIMINT_SIMD_ROUND(jend);
552
553 //////////////////////////////////////////////
554 // Contracted integrals: Horizontal recurrance
555 //////////////////////////////////////////////
556
557
558
559
560 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
561 {
562 const double hCD[3] = { Q.AB_x[cd+abcd], Q.AB_y[cd+abcd], Q.AB_z[cd+abcd] };
563
564 // set up HRR pointers
565 double const * restrict HRR_INT__g_s_i_s = INT__g_s_i_s + abcd * 420;
566 double const * restrict HRR_INT__g_s_k_s = INT__g_s_k_s + abcd * 540;
567 double const * restrict HRR_INT__g_s_l_s = INT__g_s_l_s + abcd * 675;
568 double const * restrict HRR_INT__g_s_m_s = INT__g_s_m_s + abcd * 825;
569 double * restrict HRR_INT__g_s_i_f = INT__g_s_i_f + real_abcd * 4200;
570
571 // form INT__g_s_i_p
572 ostei_general_hrr_L(4, 0, 6, 1, hCD, HRR_INT__g_s_k_s, HRR_INT__g_s_i_s, HRR_INT__g_s_i_p);
573
574 // form INT__g_s_k_p
575 ostei_general_hrr_L(4, 0, 7, 1, hCD, HRR_INT__g_s_l_s, HRR_INT__g_s_k_s, HRR_INT__g_s_k_p);
576
577 // form INT__g_s_l_p
578 ostei_general_hrr_L(4, 0, 8, 1, hCD, HRR_INT__g_s_m_s, HRR_INT__g_s_l_s, HRR_INT__g_s_l_p);
579
580 // form INT__g_s_i_d
581 ostei_general_hrr_L(4, 0, 6, 2, hCD, HRR_INT__g_s_k_p, HRR_INT__g_s_i_p, HRR_INT__g_s_i_d);
582
583 // form INT__g_s_k_d
584 ostei_general_hrr_L(4, 0, 7, 2, hCD, HRR_INT__g_s_l_p, HRR_INT__g_s_k_p, HRR_INT__g_s_k_d);
585
586 // form INT__g_s_i_f
587 ostei_general_hrr_L(4, 0, 6, 3, hCD, HRR_INT__g_s_k_d, HRR_INT__g_s_i_d, HRR_INT__g_s_i_f);
588
589
590 } // close HRR loop
591
592
593 } // close loop cdbatch
594
595 istart = iend;
596 } // close loop over ab
597
598 return P.nshell12_clip * Q.nshell12_clip;
599 }
600
ostei_s_g_i_f(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__s_g_i_f)601 int ostei_s_g_i_f(struct simint_multi_shellpair const P,
602 struct simint_multi_shellpair const Q,
603 double screen_tol,
604 double * const restrict work,
605 double * const restrict INT__s_g_i_f)
606 {
607 double P_AB[3*P.nshell12];
608 struct simint_multi_shellpair P_tmp = P;
609 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
610 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
611 P_tmp.AB_x = P_AB;
612 P_tmp.AB_y = P_AB + P.nshell12;
613 P_tmp.AB_z = P_AB + 2*P.nshell12;
614
615 for(int i = 0; i < P.nshell12; i++)
616 {
617 P_tmp.AB_x[i] = -P.AB_x[i];
618 P_tmp.AB_y[i] = -P.AB_y[i];
619 P_tmp.AB_z[i] = -P.AB_z[i];
620 }
621
622 int ret = ostei_g_s_i_f(P_tmp, Q, screen_tol, work, INT__s_g_i_f);
623 double buffer[4200] SIMINT_ALIGN_ARRAY_DBL;
624
625 for(int q = 0; q < ret; q++)
626 {
627 int idx = 0;
628 for(int a = 0; a < 1; ++a)
629 for(int b = 0; b < 15; ++b)
630 for(int c = 0; c < 28; ++c)
631 for(int d = 0; d < 10; ++d)
632 buffer[idx++] = INT__s_g_i_f[q*4200+b*280+a*280+c*10+d];
633
634 memcpy(INT__s_g_i_f+q*4200, buffer, 4200*sizeof(double));
635 }
636
637 return ret;
638 }
639
ostei_g_s_f_i(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__g_s_f_i)640 int ostei_g_s_f_i(struct simint_multi_shellpair const P,
641 struct simint_multi_shellpair const Q,
642 double screen_tol,
643 double * const restrict work,
644 double * const restrict INT__g_s_f_i)
645 {
646 double Q_AB[3*Q.nshell12];
647 struct simint_multi_shellpair Q_tmp = Q;
648 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
649 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
650 Q_tmp.AB_x = Q_AB;
651 Q_tmp.AB_y = Q_AB + Q.nshell12;
652 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
653
654 for(int i = 0; i < Q.nshell12; i++)
655 {
656 Q_tmp.AB_x[i] = -Q.AB_x[i];
657 Q_tmp.AB_y[i] = -Q.AB_y[i];
658 Q_tmp.AB_z[i] = -Q.AB_z[i];
659 }
660
661 int ret = ostei_g_s_i_f(P, Q_tmp, screen_tol, work, INT__g_s_f_i);
662 double buffer[4200] SIMINT_ALIGN_ARRAY_DBL;
663
664 for(int q = 0; q < ret; q++)
665 {
666 int idx = 0;
667 for(int a = 0; a < 15; ++a)
668 for(int b = 0; b < 1; ++b)
669 for(int c = 0; c < 10; ++c)
670 for(int d = 0; d < 28; ++d)
671 buffer[idx++] = INT__g_s_f_i[q*4200+a*280+b*280+d*10+c];
672
673 memcpy(INT__g_s_f_i+q*4200, buffer, 4200*sizeof(double));
674 }
675
676 return ret;
677 }
678
ostei_s_g_f_i(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__s_g_f_i)679 int ostei_s_g_f_i(struct simint_multi_shellpair const P,
680 struct simint_multi_shellpair const Q,
681 double screen_tol,
682 double * const restrict work,
683 double * const restrict INT__s_g_f_i)
684 {
685 double P_AB[3*P.nshell12];
686 struct simint_multi_shellpair P_tmp = P;
687 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
688 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
689 P_tmp.AB_x = P_AB;
690 P_tmp.AB_y = P_AB + P.nshell12;
691 P_tmp.AB_z = P_AB + 2*P.nshell12;
692
693 for(int i = 0; i < P.nshell12; i++)
694 {
695 P_tmp.AB_x[i] = -P.AB_x[i];
696 P_tmp.AB_y[i] = -P.AB_y[i];
697 P_tmp.AB_z[i] = -P.AB_z[i];
698 }
699
700 double Q_AB[3*Q.nshell12];
701 struct simint_multi_shellpair Q_tmp = Q;
702 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
703 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
704 Q_tmp.AB_x = Q_AB;
705 Q_tmp.AB_y = Q_AB + Q.nshell12;
706 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
707
708 for(int i = 0; i < Q.nshell12; i++)
709 {
710 Q_tmp.AB_x[i] = -Q.AB_x[i];
711 Q_tmp.AB_y[i] = -Q.AB_y[i];
712 Q_tmp.AB_z[i] = -Q.AB_z[i];
713 }
714
715 int ret = ostei_g_s_i_f(P_tmp, Q_tmp, screen_tol, work, INT__s_g_f_i);
716 double buffer[4200] SIMINT_ALIGN_ARRAY_DBL;
717
718 for(int q = 0; q < ret; q++)
719 {
720 int idx = 0;
721 for(int a = 0; a < 1; ++a)
722 for(int b = 0; b < 15; ++b)
723 for(int c = 0; c < 10; ++c)
724 for(int d = 0; d < 28; ++d)
725 buffer[idx++] = INT__s_g_f_i[q*4200+b*280+a*280+d*10+c];
726
727 memcpy(INT__s_g_f_i+q*4200, buffer, 4200*sizeof(double));
728 }
729
730 return ret;
731 }
732
733