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_f_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__f_f_p_s)8 int ostei_f_f_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__f_f_p_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__f_f_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__f_s_p_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__g_s_p_s = work + (SIMINT_NSHELL_SIMD * 30);
30 double * const INT__h_s_p_s = work + (SIMINT_NSHELL_SIMD * 75);
31 double * const INT__i_s_p_s = work + (SIMINT_NSHELL_SIMD * 138);
32 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*222);
33 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
34 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_s_s = primwork + 8;
35 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 29;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 65;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_p_s = primwork + 115;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 145;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_p_s = primwork + 205;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 250;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_p_s = primwork + 313;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 376;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_p_s = primwork + 432;
44 double * const hrrwork = (double *)(primwork + 516);
45 double * const HRR_INT__f_p_p_s = hrrwork + 0;
46 double * const HRR_INT__f_d_p_s = hrrwork + 90;
47 double * const HRR_INT__g_p_p_s = hrrwork + 270;
48 double * const HRR_INT__g_d_p_s = hrrwork + 405;
49 double * const HRR_INT__h_p_p_s = hrrwork + 675;
50
51
52 // Create constants
53 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
54 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
55 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
56 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
57 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
58 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
59 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
60
61
62 ////////////////////////////////////////
63 // Loop over shells and primitives
64 ////////////////////////////////////////
65
66 real_abcd = 0;
67 istart = 0;
68 for(ab = 0; ab < P.nshell12_clip; ++ab)
69 {
70 const int iend = istart + P.nprim12[ab];
71
72 cd = 0;
73 jstart = 0;
74
75 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
76 {
77 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
78 int jend = jstart;
79 for(i = 0; i < nshellbatch; i++)
80 jend += Q.nprim12[cd+i];
81
82 // Clear the beginning of the workspace (where we are accumulating integrals)
83 memset(work, 0, SIMINT_NSHELL_SIMD * 222 * sizeof(double));
84 abcd = 0;
85
86
87 for(i = istart; i < iend; ++i)
88 {
89 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
90
91 if(check_screen)
92 {
93 // Skip this whole thing if always insignificant
94 if((P.screen[i] * Q.screen_max) < screen_tol)
95 continue;
96 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
97 }
98
99 icd = 0;
100 iprimcd = 0;
101 nprim_icd = Q.nprim12[cd];
102 double * restrict PRIM_PTR_INT__f_s_p_s = INT__f_s_p_s + abcd * 30;
103 double * restrict PRIM_PTR_INT__g_s_p_s = INT__g_s_p_s + abcd * 45;
104 double * restrict PRIM_PTR_INT__h_s_p_s = INT__h_s_p_s + abcd * 63;
105 double * restrict PRIM_PTR_INT__i_s_p_s = INT__i_s_p_s + abcd * 84;
106
107
108
109 // Load these one per loop over i
110 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
111 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
112 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
113
114 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
115
116 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
117 {
118 // calculate the shell offsets
119 // these are the offset from the shell pointed to by cd
120 // for each element
121 int shelloffsets[SIMINT_SIMD_LEN] = {0};
122 int lastoffset = 0;
123 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
124
125 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
126 {
127 // Handle if the first element of the vector is a new shell
128 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
129 {
130 nprim_icd += Q.nprim12[cd + (++icd)];
131 PRIM_PTR_INT__f_s_p_s += 30;
132 PRIM_PTR_INT__g_s_p_s += 45;
133 PRIM_PTR_INT__h_s_p_s += 63;
134 PRIM_PTR_INT__i_s_p_s += 84;
135 }
136 iprimcd++;
137 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
138 {
139 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
140 {
141 shelloffsets[n] = shelloffsets[n-1] + 1;
142 lastoffset++;
143 nprim_icd += Q.nprim12[cd + (++icd)];
144 }
145 else
146 shelloffsets[n] = shelloffsets[n-1];
147 iprimcd++;
148 }
149 }
150 else
151 iprimcd += SIMINT_SIMD_LEN;
152
153 // Do we have to compute this vector (or has it been screened out)?
154 // (not_screened != 0 means we have to do this vector)
155 if(check_screen)
156 {
157 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
158 if(vmax < screen_tol)
159 {
160 PRIM_PTR_INT__f_s_p_s += lastoffset*30;
161 PRIM_PTR_INT__g_s_p_s += lastoffset*45;
162 PRIM_PTR_INT__h_s_p_s += lastoffset*63;
163 PRIM_PTR_INT__i_s_p_s += lastoffset*84;
164 continue;
165 }
166 }
167
168 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
169 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
170 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
171 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
172
173
174 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
175 SIMINT_DBLTYPE PQ[3];
176 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
177 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
178 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
179 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
180 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
181 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
182
183 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
184 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
185 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
186 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
187 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
188 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
189 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
190
191 // NOTE: Minus sign!
192 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
193 SIMINT_DBLTYPE aop_PQ[3];
194 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
195 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
196 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
197
198 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
199 SIMINT_DBLTYPE aoq_PQ[3];
200 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
201 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
202 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
203 // Put a minus sign here so we don't have to in RR routines
204 a_over_q = SIMINT_NEG(a_over_q);
205
206
207 //////////////////////////////////////////////
208 // Fjt function section
209 // Maximum v value: 7
210 //////////////////////////////////////////////
211 // The parameter to the Fjt function
212 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
213
214
215 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
216
217
218 boys_F_split(PRIM_INT__s_s_s_s, F_x, 7);
219 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
220 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
221 for(n = 0; n <= 7; n++)
222 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
223
224 //////////////////////////////////////////////
225 // Primitive integrals: Vertical recurrance
226 //////////////////////////////////////////////
227
228 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
229 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
230 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
231 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
232 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
233 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
234 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
235 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
236 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
237 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
238 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
239
240
241
242 // Forming PRIM_INT__p_s_s_s[7 * 3];
243 for(n = 0; n < 7; ++n) // loop over orders of auxiliary function
244 {
245
246 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
247 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]);
248
249 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
250 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]);
251
252 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
253 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]);
254
255 }
256
257
258
259 // Forming PRIM_INT__d_s_s_s[6 * 6];
260 for(n = 0; n < 6; ++n) // loop over orders of auxiliary function
261 {
262
263 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
264 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]);
265 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]);
266
267 PRIM_INT__d_s_s_s[n * 6 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 0]);
268 PRIM_INT__d_s_s_s[n * 6 + 1] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__d_s_s_s[n * 6 + 1]);
269
270 PRIM_INT__d_s_s_s[n * 6 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 0]);
271 PRIM_INT__d_s_s_s[n * 6 + 2] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__p_s_s_s[(n+1) * 3 + 0], PRIM_INT__d_s_s_s[n * 6 + 2]);
272
273 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
274 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]);
275 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]);
276
277 PRIM_INT__d_s_s_s[n * 6 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 1]);
278 PRIM_INT__d_s_s_s[n * 6 + 4] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__p_s_s_s[(n+1) * 3 + 1], PRIM_INT__d_s_s_s[n * 6 + 4]);
279
280 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
281 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]);
282 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]);
283
284 }
285
286
287
288 // Forming PRIM_INT__f_s_s_s[5 * 10];
289 for(n = 0; n < 5; ++n) // loop over orders of auxiliary function
290 {
291
292 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
293 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]);
294 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]);
295
296 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
297 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]);
298
299 PRIM_INT__f_s_s_s[n * 10 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 0]);
300 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]);
301
302 PRIM_INT__f_s_s_s[n * 10 + 3] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 3]);
303 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]);
304
305 PRIM_INT__f_s_s_s[n * 10 + 4] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 1]);
306 PRIM_INT__f_s_s_s[n * 10 + 4] = SIMINT_FMADD( aop_PQ[2], PRIM_INT__d_s_s_s[(n+1) * 6 + 1], PRIM_INT__f_s_s_s[n * 10 + 4]);
307
308 PRIM_INT__f_s_s_s[n * 10 + 5] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 5]);
309 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]);
310
311 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
312 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]);
313 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]);
314
315 PRIM_INT__f_s_s_s[n * 10 + 7] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 3]);
316 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]);
317
318 PRIM_INT__f_s_s_s[n * 10 + 8] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 5]);
319 PRIM_INT__f_s_s_s[n * 10 + 8] = SIMINT_FMADD( aop_PQ[1], PRIM_INT__d_s_s_s[(n+1) * 6 + 5], PRIM_INT__f_s_s_s[n * 10 + 8]);
320
321 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
322 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]);
323 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]);
324
325 }
326
327
328 VRR_K_f_s_p_s(
329 PRIM_INT__f_s_p_s,
330 PRIM_INT__f_s_s_s,
331 PRIM_INT__d_s_s_s,
332 Q_PA,
333 aoq_PQ,
334 one_over_2pq,
335 1);
336
337
338 VRR_I_g_s_s_s(
339 PRIM_INT__g_s_s_s,
340 PRIM_INT__f_s_s_s,
341 PRIM_INT__d_s_s_s,
342 P_PA,
343 a_over_p,
344 aop_PQ,
345 one_over_2p,
346 4);
347
348
349 VRR_K_g_s_p_s(
350 PRIM_INT__g_s_p_s,
351 PRIM_INT__g_s_s_s,
352 PRIM_INT__f_s_s_s,
353 Q_PA,
354 aoq_PQ,
355 one_over_2pq,
356 1);
357
358
359 VRR_I_h_s_s_s(
360 PRIM_INT__h_s_s_s,
361 PRIM_INT__g_s_s_s,
362 PRIM_INT__f_s_s_s,
363 P_PA,
364 a_over_p,
365 aop_PQ,
366 one_over_2p,
367 3);
368
369
370 ostei_general_vrr_K(5, 0, 1, 0, 1,
371 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
372 PRIM_INT__h_s_s_s, NULL, NULL, PRIM_INT__g_s_s_s, NULL, PRIM_INT__h_s_p_s);
373
374
375 ostei_general_vrr1_I(6, 2,
376 one_over_2p, a_over_p, aop_PQ, P_PA,
377 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
378
379
380 ostei_general_vrr_K(6, 0, 1, 0, 1,
381 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
382 PRIM_INT__i_s_s_s, NULL, NULL, PRIM_INT__h_s_s_s, NULL, PRIM_INT__i_s_p_s);
383
384
385
386
387 ////////////////////////////////////
388 // Accumulate contracted integrals
389 ////////////////////////////////////
390 if(lastoffset == 0)
391 {
392 contract_all(30, PRIM_INT__f_s_p_s, PRIM_PTR_INT__f_s_p_s);
393 contract_all(45, PRIM_INT__g_s_p_s, PRIM_PTR_INT__g_s_p_s);
394 contract_all(63, PRIM_INT__h_s_p_s, PRIM_PTR_INT__h_s_p_s);
395 contract_all(84, PRIM_INT__i_s_p_s, PRIM_PTR_INT__i_s_p_s);
396 }
397 else
398 {
399 contract(30, shelloffsets, PRIM_INT__f_s_p_s, PRIM_PTR_INT__f_s_p_s);
400 contract(45, shelloffsets, PRIM_INT__g_s_p_s, PRIM_PTR_INT__g_s_p_s);
401 contract(63, shelloffsets, PRIM_INT__h_s_p_s, PRIM_PTR_INT__h_s_p_s);
402 contract(84, shelloffsets, PRIM_INT__i_s_p_s, PRIM_PTR_INT__i_s_p_s);
403 PRIM_PTR_INT__f_s_p_s += lastoffset*30;
404 PRIM_PTR_INT__g_s_p_s += lastoffset*45;
405 PRIM_PTR_INT__h_s_p_s += lastoffset*63;
406 PRIM_PTR_INT__i_s_p_s += lastoffset*84;
407 }
408
409 } // close loop over j
410 } // close loop over i
411
412 //Advance to the next batch
413 jstart = SIMINT_SIMD_ROUND(jend);
414
415 //////////////////////////////////////////////
416 // Contracted integrals: Horizontal recurrance
417 //////////////////////////////////////////////
418
419
420 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
421
422
423 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
424 {
425
426 // set up HRR pointers
427 double const * restrict HRR_INT__f_s_p_s = INT__f_s_p_s + abcd * 30;
428 double const * restrict HRR_INT__g_s_p_s = INT__g_s_p_s + abcd * 45;
429 double const * restrict HRR_INT__h_s_p_s = INT__h_s_p_s + abcd * 63;
430 double const * restrict HRR_INT__i_s_p_s = INT__i_s_p_s + abcd * 84;
431 double * restrict HRR_INT__f_f_p_s = INT__f_f_p_s + real_abcd * 300;
432
433 // form INT__f_p_p_s
434 HRR_J_f_p(
435 HRR_INT__f_p_p_s,
436 HRR_INT__f_s_p_s,
437 HRR_INT__g_s_p_s,
438 hAB, 3);
439
440 // form INT__g_p_p_s
441 HRR_J_g_p(
442 HRR_INT__g_p_p_s,
443 HRR_INT__g_s_p_s,
444 HRR_INT__h_s_p_s,
445 hAB, 3);
446
447 // form INT__h_p_p_s
448 ostei_general_hrr_J(5, 1, 1, 0, hAB, HRR_INT__i_s_p_s, HRR_INT__h_s_p_s, HRR_INT__h_p_p_s);
449
450 // form INT__f_d_p_s
451 HRR_J_f_d(
452 HRR_INT__f_d_p_s,
453 HRR_INT__f_p_p_s,
454 HRR_INT__g_p_p_s,
455 hAB, 3);
456
457 // form INT__g_d_p_s
458 ostei_general_hrr_J(4, 2, 1, 0, hAB, HRR_INT__h_p_p_s, HRR_INT__g_p_p_s, HRR_INT__g_d_p_s);
459
460 // form INT__f_f_p_s
461 ostei_general_hrr_J(3, 3, 1, 0, hAB, HRR_INT__g_d_p_s, HRR_INT__f_d_p_s, HRR_INT__f_f_p_s);
462
463
464 } // close HRR loop
465
466
467 } // close loop cdbatch
468
469 istart = iend;
470 } // close loop over ab
471
472 return P.nshell12_clip * Q.nshell12_clip;
473 }
474
475