1 #include "simint/boys/boys.h"
2 #include "simint/ostei/gen/ostei_generated.h"
3 #include "simint/vectorization/vectorization.h"
4 #include <math.h>
5 #include <string.h>
6
7
ostei_f_s_k_f(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__f_s_k_f)8 int ostei_f_s_k_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__f_s_k_f)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__f_s_k_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__f_s_k_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__f_s_l_s = work + (SIMINT_NSHELL_SIMD * 360);
30 double * const INT__f_s_m_s = work + (SIMINT_NSHELL_SIMD * 810);
31 double * const INT__f_s_n_s = work + (SIMINT_NSHELL_SIMD * 1360);
32 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*2020);
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__s_s_n_s = primwork + 1595;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_h_s = primwork + 1859;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_i_s = primwork + 2048;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_k_s = primwork + 2300;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_l_s = primwork + 2624;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_m_s = primwork + 3029;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_n_s = primwork + 3524;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_i_s = primwork + 4118;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_k_s = primwork + 4454;
52 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_l_s = primwork + 4886;
53 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_m_s = primwork + 5426;
54 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_n_s = primwork + 6086;
55 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_k_s = primwork + 6878;
56 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_l_s = primwork + 7238;
57 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_m_s = primwork + 7688;
58 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_n_s = primwork + 8238;
59 double * const hrrwork = (double *)(primwork + 8898);
60 double * const HRR_INT__f_s_k_p = hrrwork + 0;
61 double * const HRR_INT__f_s_k_d = hrrwork + 1080;
62 double * const HRR_INT__f_s_l_p = hrrwork + 3240;
63 double * const HRR_INT__f_s_l_d = hrrwork + 4590;
64 double * const HRR_INT__f_s_m_p = hrrwork + 7290;
65
66
67 // Create constants
68 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
69 const SIMINT_DBLTYPE const_10 = SIMINT_DBLSET1(10);
70 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
71 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
72 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
73 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
74 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
75 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
76 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
77 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
78 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
79
80
81 ////////////////////////////////////////
82 // Loop over shells and primitives
83 ////////////////////////////////////////
84
85 real_abcd = 0;
86 istart = 0;
87 for(ab = 0; ab < P.nshell12_clip; ++ab)
88 {
89 const int iend = istart + P.nprim12[ab];
90
91 cd = 0;
92 jstart = 0;
93
94 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
95 {
96 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
97 int jend = jstart;
98 for(i = 0; i < nshellbatch; i++)
99 jend += Q.nprim12[cd+i];
100
101 // Clear the beginning of the workspace (where we are accumulating integrals)
102 memset(work, 0, SIMINT_NSHELL_SIMD * 2020 * sizeof(double));
103 abcd = 0;
104
105
106 for(i = istart; i < iend; ++i)
107 {
108 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
109
110 if(check_screen)
111 {
112 // Skip this whole thing if always insignificant
113 if((P.screen[i] * Q.screen_max) < screen_tol)
114 continue;
115 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
116 }
117
118 icd = 0;
119 iprimcd = 0;
120 nprim_icd = Q.nprim12[cd];
121 double * restrict PRIM_PTR_INT__f_s_k_s = INT__f_s_k_s + abcd * 360;
122 double * restrict PRIM_PTR_INT__f_s_l_s = INT__f_s_l_s + abcd * 450;
123 double * restrict PRIM_PTR_INT__f_s_m_s = INT__f_s_m_s + abcd * 550;
124 double * restrict PRIM_PTR_INT__f_s_n_s = INT__f_s_n_s + abcd * 660;
125
126
127
128 // Load these one per loop over i
129 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
130 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
131 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
132
133 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
134
135 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
136 {
137 // calculate the shell offsets
138 // these are the offset from the shell pointed to by cd
139 // for each element
140 int shelloffsets[SIMINT_SIMD_LEN] = {0};
141 int lastoffset = 0;
142 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
143
144 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
145 {
146 // Handle if the first element of the vector is a new shell
147 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
148 {
149 nprim_icd += Q.nprim12[cd + (++icd)];
150 PRIM_PTR_INT__f_s_k_s += 360;
151 PRIM_PTR_INT__f_s_l_s += 450;
152 PRIM_PTR_INT__f_s_m_s += 550;
153 PRIM_PTR_INT__f_s_n_s += 660;
154 }
155 iprimcd++;
156 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
157 {
158 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
159 {
160 shelloffsets[n] = shelloffsets[n-1] + 1;
161 lastoffset++;
162 nprim_icd += Q.nprim12[cd + (++icd)];
163 }
164 else
165 shelloffsets[n] = shelloffsets[n-1];
166 iprimcd++;
167 }
168 }
169 else
170 iprimcd += SIMINT_SIMD_LEN;
171
172 // Do we have to compute this vector (or has it been screened out)?
173 // (not_screened != 0 means we have to do this vector)
174 if(check_screen)
175 {
176 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
177 if(vmax < screen_tol)
178 {
179 PRIM_PTR_INT__f_s_k_s += lastoffset*360;
180 PRIM_PTR_INT__f_s_l_s += lastoffset*450;
181 PRIM_PTR_INT__f_s_m_s += lastoffset*550;
182 PRIM_PTR_INT__f_s_n_s += lastoffset*660;
183 continue;
184 }
185 }
186
187 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
188 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
189 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
190 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
191
192
193 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
194 SIMINT_DBLTYPE PQ[3];
195 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
196 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
197 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
198 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
199 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
200 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
201
202 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
203 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
204 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
205 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
206 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
207 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
208 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
209
210 // NOTE: Minus sign!
211 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
212 SIMINT_DBLTYPE aop_PQ[3];
213 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
214 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
215 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
216
217 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
218 SIMINT_DBLTYPE aoq_PQ[3];
219 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
220 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
221 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
222 // Put a minus sign here so we don't have to in RR routines
223 a_over_q = SIMINT_NEG(a_over_q);
224
225
226 //////////////////////////////////////////////
227 // Fjt function section
228 // Maximum v value: 13
229 //////////////////////////////////////////////
230 // The parameter to the Fjt function
231 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
232
233
234 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
235
236
237 boys_F_split(PRIM_INT__s_s_s_s, F_x, 13);
238 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
239 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
240 for(n = 0; n <= 13; n++)
241 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
242
243 //////////////////////////////////////////////
244 // Primitive integrals: Vertical recurrance
245 //////////////////////////////////////////////
246
247 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
248 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
249 const SIMINT_DBLTYPE vrr_const_1_over_2q = one_over_2q;
250 const SIMINT_DBLTYPE vrr_const_2_over_2q = SIMINT_MUL(const_2, one_over_2q);
251 const SIMINT_DBLTYPE vrr_const_3_over_2q = SIMINT_MUL(const_3, one_over_2q);
252 const SIMINT_DBLTYPE vrr_const_4_over_2q = SIMINT_MUL(const_4, one_over_2q);
253 const SIMINT_DBLTYPE vrr_const_5_over_2q = SIMINT_MUL(const_5, one_over_2q);
254 const SIMINT_DBLTYPE vrr_const_6_over_2q = SIMINT_MUL(const_6, one_over_2q);
255 const SIMINT_DBLTYPE vrr_const_7_over_2q = SIMINT_MUL(const_7, one_over_2q);
256 const SIMINT_DBLTYPE vrr_const_8_over_2q = SIMINT_MUL(const_8, one_over_2q);
257 const SIMINT_DBLTYPE vrr_const_9_over_2q = SIMINT_MUL(const_9, one_over_2q);
258 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
259 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
260 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
261 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
262 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
263 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
264 const SIMINT_DBLTYPE vrr_const_7_over_2pq = SIMINT_MUL(const_7, one_over_2pq);
265 const SIMINT_DBLTYPE vrr_const_8_over_2pq = SIMINT_MUL(const_8, one_over_2pq);
266 const SIMINT_DBLTYPE vrr_const_9_over_2pq = SIMINT_MUL(const_9, one_over_2pq);
267 const SIMINT_DBLTYPE vrr_const_10_over_2pq = SIMINT_MUL(const_10, one_over_2pq);
268
269
270
271 // Forming PRIM_INT__s_s_p_s[13 * 3];
272 for(n = 0; n < 13; ++n) // loop over orders of auxiliary function
273 {
274
275 PRIM_INT__s_s_p_s[n * 3 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
276 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]);
277
278 PRIM_INT__s_s_p_s[n * 3 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
279 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]);
280
281 PRIM_INT__s_s_p_s[n * 3 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
282 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]);
283
284 }
285
286
287
288 // Forming PRIM_INT__s_s_d_s[12 * 6];
289 for(n = 0; n < 12; ++n) // loop over orders of auxiliary function
290 {
291
292 PRIM_INT__s_s_d_s[n * 6 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_p_s[n * 3 + 0]);
293 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]);
294 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]);
295
296 PRIM_INT__s_s_d_s[n * 6 + 3] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_p_s[n * 3 + 1]);
297 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]);
298 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]);
299
300 PRIM_INT__s_s_d_s[n * 6 + 5] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_p_s[n * 3 + 2]);
301 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]);
302 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]);
303
304 }
305
306
307
308 // Forming PRIM_INT__s_s_f_s[11 * 10];
309 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
310 {
311
312 PRIM_INT__s_s_f_s[n * 10 + 0] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 0]);
313 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]);
314 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]);
315
316 PRIM_INT__s_s_f_s[n * 10 + 1] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 0]);
317 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]);
318
319 PRIM_INT__s_s_f_s[n * 10 + 2] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 0]);
320 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]);
321
322 PRIM_INT__s_s_f_s[n * 10 + 3] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 3]);
323 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]);
324
325 PRIM_INT__s_s_f_s[n * 10 + 5] = SIMINT_MUL(Q_PA[0], PRIM_INT__s_s_d_s[n * 6 + 5]);
326 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]);
327
328 PRIM_INT__s_s_f_s[n * 10 + 6] = SIMINT_MUL(Q_PA[1], PRIM_INT__s_s_d_s[n * 6 + 3]);
329 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]);
330 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]);
331
332 PRIM_INT__s_s_f_s[n * 10 + 7] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 3]);
333 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]);
334
335 PRIM_INT__s_s_f_s[n * 10 + 9] = SIMINT_MUL(Q_PA[2], PRIM_INT__s_s_d_s[n * 6 + 5]);
336 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]);
337 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]);
338
339 }
340
341
342 VRR_K_s_s_g_s(
343 PRIM_INT__s_s_g_s,
344 PRIM_INT__s_s_f_s,
345 PRIM_INT__s_s_d_s,
346 Q_PA,
347 a_over_q,
348 aoq_PQ,
349 one_over_2q,
350 10);
351
352
353 VRR_K_s_s_h_s(
354 PRIM_INT__s_s_h_s,
355 PRIM_INT__s_s_g_s,
356 PRIM_INT__s_s_f_s,
357 Q_PA,
358 a_over_q,
359 aoq_PQ,
360 one_over_2q,
361 9);
362
363
364 ostei_general_vrr1_K(6, 8,
365 one_over_2q, a_over_q, aoq_PQ, Q_PA,
366 PRIM_INT__s_s_h_s, PRIM_INT__s_s_g_s, PRIM_INT__s_s_i_s);
367
368
369 ostei_general_vrr1_K(7, 7,
370 one_over_2q, a_over_q, aoq_PQ, Q_PA,
371 PRIM_INT__s_s_i_s, PRIM_INT__s_s_h_s, PRIM_INT__s_s_k_s);
372
373
374 ostei_general_vrr_I(1, 0, 7, 0, 3,
375 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
376 PRIM_INT__s_s_k_s, NULL, NULL, PRIM_INT__s_s_i_s, NULL, PRIM_INT__p_s_k_s);
377
378
379 ostei_general_vrr_I(1, 0, 6, 0, 3,
380 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
381 PRIM_INT__s_s_i_s, NULL, NULL, PRIM_INT__s_s_h_s, NULL, PRIM_INT__p_s_i_s);
382
383
384 ostei_general_vrr_I(2, 0, 7, 0, 2,
385 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
386 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);
387
388
389 ostei_general_vrr_I(1, 0, 5, 0, 3,
390 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
391 PRIM_INT__s_s_h_s, NULL, NULL, PRIM_INT__s_s_g_s, NULL, PRIM_INT__p_s_h_s);
392
393
394 ostei_general_vrr_I(2, 0, 6, 0, 2,
395 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
396 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);
397
398
399 ostei_general_vrr_I(3, 0, 7, 0, 1,
400 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
401 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);
402
403
404 ostei_general_vrr1_K(8, 6,
405 one_over_2q, a_over_q, aoq_PQ, Q_PA,
406 PRIM_INT__s_s_k_s, PRIM_INT__s_s_i_s, PRIM_INT__s_s_l_s);
407
408
409 ostei_general_vrr_I(1, 0, 8, 0, 3,
410 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
411 PRIM_INT__s_s_l_s, NULL, NULL, PRIM_INT__s_s_k_s, NULL, PRIM_INT__p_s_l_s);
412
413
414 ostei_general_vrr_I(2, 0, 8, 0, 2,
415 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
416 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);
417
418
419 ostei_general_vrr_I(3, 0, 8, 0, 1,
420 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
421 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);
422
423
424 ostei_general_vrr1_K(9, 5,
425 one_over_2q, a_over_q, aoq_PQ, Q_PA,
426 PRIM_INT__s_s_l_s, PRIM_INT__s_s_k_s, PRIM_INT__s_s_m_s);
427
428
429 ostei_general_vrr_I(1, 0, 9, 0, 3,
430 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
431 PRIM_INT__s_s_m_s, NULL, NULL, PRIM_INT__s_s_l_s, NULL, PRIM_INT__p_s_m_s);
432
433
434 ostei_general_vrr_I(2, 0, 9, 0, 2,
435 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
436 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);
437
438
439 ostei_general_vrr_I(3, 0, 9, 0, 1,
440 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
441 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);
442
443
444 ostei_general_vrr1_K(10, 4,
445 one_over_2q, a_over_q, aoq_PQ, Q_PA,
446 PRIM_INT__s_s_m_s, PRIM_INT__s_s_l_s, PRIM_INT__s_s_n_s);
447
448
449 ostei_general_vrr_I(1, 0, 10, 0, 3,
450 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
451 PRIM_INT__s_s_n_s, NULL, NULL, PRIM_INT__s_s_m_s, NULL, PRIM_INT__p_s_n_s);
452
453
454 ostei_general_vrr_I(2, 0, 10, 0, 2,
455 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
456 PRIM_INT__p_s_n_s, PRIM_INT__s_s_n_s, NULL, PRIM_INT__p_s_m_s, NULL, PRIM_INT__d_s_n_s);
457
458
459 ostei_general_vrr_I(3, 0, 10, 0, 1,
460 one_over_2p, a_over_p, one_over_2pq, aop_PQ, P_PA,
461 PRIM_INT__d_s_n_s, PRIM_INT__p_s_n_s, NULL, PRIM_INT__d_s_m_s, NULL, PRIM_INT__f_s_n_s);
462
463
464
465
466 ////////////////////////////////////
467 // Accumulate contracted integrals
468 ////////////////////////////////////
469 if(lastoffset == 0)
470 {
471 contract_all(360, PRIM_INT__f_s_k_s, PRIM_PTR_INT__f_s_k_s);
472 contract_all(450, PRIM_INT__f_s_l_s, PRIM_PTR_INT__f_s_l_s);
473 contract_all(550, PRIM_INT__f_s_m_s, PRIM_PTR_INT__f_s_m_s);
474 contract_all(660, PRIM_INT__f_s_n_s, PRIM_PTR_INT__f_s_n_s);
475 }
476 else
477 {
478 contract(360, shelloffsets, PRIM_INT__f_s_k_s, PRIM_PTR_INT__f_s_k_s);
479 contract(450, shelloffsets, PRIM_INT__f_s_l_s, PRIM_PTR_INT__f_s_l_s);
480 contract(550, shelloffsets, PRIM_INT__f_s_m_s, PRIM_PTR_INT__f_s_m_s);
481 contract(660, shelloffsets, PRIM_INT__f_s_n_s, PRIM_PTR_INT__f_s_n_s);
482 PRIM_PTR_INT__f_s_k_s += lastoffset*360;
483 PRIM_PTR_INT__f_s_l_s += lastoffset*450;
484 PRIM_PTR_INT__f_s_m_s += lastoffset*550;
485 PRIM_PTR_INT__f_s_n_s += lastoffset*660;
486 }
487
488 } // close loop over j
489 } // close loop over i
490
491 //Advance to the next batch
492 jstart = SIMINT_SIMD_ROUND(jend);
493
494 //////////////////////////////////////////////
495 // Contracted integrals: Horizontal recurrance
496 //////////////////////////////////////////////
497
498
499
500
501 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
502 {
503 const double hCD[3] = { Q.AB_x[cd+abcd], Q.AB_y[cd+abcd], Q.AB_z[cd+abcd] };
504
505 // set up HRR pointers
506 double const * restrict HRR_INT__f_s_k_s = INT__f_s_k_s + abcd * 360;
507 double const * restrict HRR_INT__f_s_l_s = INT__f_s_l_s + abcd * 450;
508 double const * restrict HRR_INT__f_s_m_s = INT__f_s_m_s + abcd * 550;
509 double const * restrict HRR_INT__f_s_n_s = INT__f_s_n_s + abcd * 660;
510 double * restrict HRR_INT__f_s_k_f = INT__f_s_k_f + real_abcd * 3600;
511
512 // form INT__f_s_k_p
513 ostei_general_hrr_L(3, 0, 7, 1, hCD, HRR_INT__f_s_l_s, HRR_INT__f_s_k_s, HRR_INT__f_s_k_p);
514
515 // form INT__f_s_l_p
516 ostei_general_hrr_L(3, 0, 8, 1, hCD, HRR_INT__f_s_m_s, HRR_INT__f_s_l_s, HRR_INT__f_s_l_p);
517
518 // form INT__f_s_m_p
519 ostei_general_hrr_L(3, 0, 9, 1, hCD, HRR_INT__f_s_n_s, HRR_INT__f_s_m_s, HRR_INT__f_s_m_p);
520
521 // form INT__f_s_k_d
522 ostei_general_hrr_L(3, 0, 7, 2, hCD, HRR_INT__f_s_l_p, HRR_INT__f_s_k_p, HRR_INT__f_s_k_d);
523
524 // form INT__f_s_l_d
525 ostei_general_hrr_L(3, 0, 8, 2, hCD, HRR_INT__f_s_m_p, HRR_INT__f_s_l_p, HRR_INT__f_s_l_d);
526
527 // form INT__f_s_k_f
528 ostei_general_hrr_L(3, 0, 7, 3, hCD, HRR_INT__f_s_l_d, HRR_INT__f_s_k_d, HRR_INT__f_s_k_f);
529
530
531 } // close HRR loop
532
533
534 } // close loop cdbatch
535
536 istart = iend;
537 } // close loop over ab
538
539 return P.nshell12_clip * Q.nshell12_clip;
540 }
541
ostei_s_f_k_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_f_k_f)542 int ostei_s_f_k_f(struct simint_multi_shellpair const P,
543 struct simint_multi_shellpair const Q,
544 double screen_tol,
545 double * const restrict work,
546 double * const restrict INT__s_f_k_f)
547 {
548 double P_AB[3*P.nshell12];
549 struct simint_multi_shellpair P_tmp = P;
550 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
551 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
552 P_tmp.AB_x = P_AB;
553 P_tmp.AB_y = P_AB + P.nshell12;
554 P_tmp.AB_z = P_AB + 2*P.nshell12;
555
556 for(int i = 0; i < P.nshell12; i++)
557 {
558 P_tmp.AB_x[i] = -P.AB_x[i];
559 P_tmp.AB_y[i] = -P.AB_y[i];
560 P_tmp.AB_z[i] = -P.AB_z[i];
561 }
562
563 int ret = ostei_f_s_k_f(P_tmp, Q, screen_tol, work, INT__s_f_k_f);
564 double buffer[3600] SIMINT_ALIGN_ARRAY_DBL;
565
566 for(int q = 0; q < ret; q++)
567 {
568 int idx = 0;
569 for(int a = 0; a < 1; ++a)
570 for(int b = 0; b < 10; ++b)
571 for(int c = 0; c < 36; ++c)
572 for(int d = 0; d < 10; ++d)
573 buffer[idx++] = INT__s_f_k_f[q*3600+b*360+a*360+c*10+d];
574
575 memcpy(INT__s_f_k_f+q*3600, buffer, 3600*sizeof(double));
576 }
577
578 return ret;
579 }
580
ostei_f_s_f_k(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__f_s_f_k)581 int ostei_f_s_f_k(struct simint_multi_shellpair const P,
582 struct simint_multi_shellpair const Q,
583 double screen_tol,
584 double * const restrict work,
585 double * const restrict INT__f_s_f_k)
586 {
587 double Q_AB[3*Q.nshell12];
588 struct simint_multi_shellpair Q_tmp = Q;
589 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
590 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
591 Q_tmp.AB_x = Q_AB;
592 Q_tmp.AB_y = Q_AB + Q.nshell12;
593 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
594
595 for(int i = 0; i < Q.nshell12; i++)
596 {
597 Q_tmp.AB_x[i] = -Q.AB_x[i];
598 Q_tmp.AB_y[i] = -Q.AB_y[i];
599 Q_tmp.AB_z[i] = -Q.AB_z[i];
600 }
601
602 int ret = ostei_f_s_k_f(P, Q_tmp, screen_tol, work, INT__f_s_f_k);
603 double buffer[3600] SIMINT_ALIGN_ARRAY_DBL;
604
605 for(int q = 0; q < ret; q++)
606 {
607 int idx = 0;
608 for(int a = 0; a < 10; ++a)
609 for(int b = 0; b < 1; ++b)
610 for(int c = 0; c < 10; ++c)
611 for(int d = 0; d < 36; ++d)
612 buffer[idx++] = INT__f_s_f_k[q*3600+a*360+b*360+d*10+c];
613
614 memcpy(INT__f_s_f_k+q*3600, buffer, 3600*sizeof(double));
615 }
616
617 return ret;
618 }
619
ostei_s_f_f_k(struct simint_multi_shellpair const P,struct simint_multi_shellpair const Q,double screen_tol,double * const restrict work,double * const restrict INT__s_f_f_k)620 int ostei_s_f_f_k(struct simint_multi_shellpair const P,
621 struct simint_multi_shellpair const Q,
622 double screen_tol,
623 double * const restrict work,
624 double * const restrict INT__s_f_f_k)
625 {
626 double P_AB[3*P.nshell12];
627 struct simint_multi_shellpair P_tmp = P;
628 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
629 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
630 P_tmp.AB_x = P_AB;
631 P_tmp.AB_y = P_AB + P.nshell12;
632 P_tmp.AB_z = P_AB + 2*P.nshell12;
633
634 for(int i = 0; i < P.nshell12; i++)
635 {
636 P_tmp.AB_x[i] = -P.AB_x[i];
637 P_tmp.AB_y[i] = -P.AB_y[i];
638 P_tmp.AB_z[i] = -P.AB_z[i];
639 }
640
641 double Q_AB[3*Q.nshell12];
642 struct simint_multi_shellpair Q_tmp = Q;
643 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
644 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
645 Q_tmp.AB_x = Q_AB;
646 Q_tmp.AB_y = Q_AB + Q.nshell12;
647 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
648
649 for(int i = 0; i < Q.nshell12; i++)
650 {
651 Q_tmp.AB_x[i] = -Q.AB_x[i];
652 Q_tmp.AB_y[i] = -Q.AB_y[i];
653 Q_tmp.AB_z[i] = -Q.AB_z[i];
654 }
655
656 int ret = ostei_f_s_k_f(P_tmp, Q_tmp, screen_tol, work, INT__s_f_f_k);
657 double buffer[3600] SIMINT_ALIGN_ARRAY_DBL;
658
659 for(int q = 0; q < ret; q++)
660 {
661 int idx = 0;
662 for(int a = 0; a < 1; ++a)
663 for(int b = 0; b < 10; ++b)
664 for(int c = 0; c < 10; ++c)
665 for(int d = 0; d < 36; ++d)
666 buffer[idx++] = INT__s_f_f_k[q*3600+b*360+a*360+d*10+c];
667
668 memcpy(INT__s_f_f_k+q*3600, buffer, 3600*sizeof(double));
669 }
670
671 return ret;
672 }
673
674