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_k_h_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__k_h_p_s)8 int ostei_k_h_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__k_h_p_s)
13 {
14
15 SIMINT_ASSUME_ALIGN_DBL(work);
16 SIMINT_ASSUME_ALIGN_DBL(INT__k_h_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__k_s_p_s = work + (SIMINT_NSHELL_SIMD * 0);
29 double * const INT__l_s_p_s = work + (SIMINT_NSHELL_SIMD * 108);
30 double * const INT__m_s_p_s = work + (SIMINT_NSHELL_SIMD * 243);
31 double * const INT__n_s_p_s = work + (SIMINT_NSHELL_SIMD * 408);
32 double * const INT__o_s_p_s = work + (SIMINT_NSHELL_SIMD * 606);
33 double * const INT__q_s_p_s = work + (SIMINT_NSHELL_SIMD * 840);
34 SIMINT_DBLTYPE * const primwork = (SIMINT_DBLTYPE *)(work + SIMINT_NSHELL_SIMD*1113);
35 SIMINT_DBLTYPE * const restrict PRIM_INT__s_s_s_s = primwork + 0;
36 SIMINT_DBLTYPE * const restrict PRIM_INT__p_s_s_s = primwork + 14;
37 SIMINT_DBLTYPE * const restrict PRIM_INT__d_s_s_s = primwork + 53;
38 SIMINT_DBLTYPE * const restrict PRIM_INT__f_s_s_s = primwork + 125;
39 SIMINT_DBLTYPE * const restrict PRIM_INT__g_s_s_s = primwork + 235;
40 SIMINT_DBLTYPE * const restrict PRIM_INT__h_s_s_s = primwork + 385;
41 SIMINT_DBLTYPE * const restrict PRIM_INT__i_s_s_s = primwork + 574;
42 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_s_s = primwork + 798;
43 SIMINT_DBLTYPE * const restrict PRIM_INT__k_s_p_s = primwork + 1050;
44 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_s_s = primwork + 1158;
45 SIMINT_DBLTYPE * const restrict PRIM_INT__l_s_p_s = primwork + 1428;
46 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_s_s = primwork + 1563;
47 SIMINT_DBLTYPE * const restrict PRIM_INT__m_s_p_s = primwork + 1838;
48 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_s_s = primwork + 2003;
49 SIMINT_DBLTYPE * const restrict PRIM_INT__n_s_p_s = primwork + 2267;
50 SIMINT_DBLTYPE * const restrict PRIM_INT__o_s_s_s = primwork + 2465;
51 SIMINT_DBLTYPE * const restrict PRIM_INT__o_s_p_s = primwork + 2699;
52 SIMINT_DBLTYPE * const restrict PRIM_INT__q_s_s_s = primwork + 2933;
53 SIMINT_DBLTYPE * const restrict PRIM_INT__q_s_p_s = primwork + 3115;
54 double * const hrrwork = (double *)(primwork + 3388);
55 double * const HRR_INT__k_p_p_s = hrrwork + 0;
56 double * const HRR_INT__k_d_p_s = hrrwork + 324;
57 double * const HRR_INT__k_f_p_s = hrrwork + 972;
58 double * const HRR_INT__k_g_p_s = hrrwork + 2052;
59 double * const HRR_INT__l_p_p_s = hrrwork + 3672;
60 double * const HRR_INT__l_d_p_s = hrrwork + 4077;
61 double * const HRR_INT__l_f_p_s = hrrwork + 4887;
62 double * const HRR_INT__l_g_p_s = hrrwork + 6237;
63 double * const HRR_INT__m_p_p_s = hrrwork + 8262;
64 double * const HRR_INT__m_d_p_s = hrrwork + 8757;
65 double * const HRR_INT__m_f_p_s = hrrwork + 9747;
66 double * const HRR_INT__n_p_p_s = hrrwork + 11397;
67 double * const HRR_INT__n_d_p_s = hrrwork + 11991;
68 double * const HRR_INT__o_p_p_s = hrrwork + 13179;
69
70
71 // Create constants
72 const SIMINT_DBLTYPE const_1 = SIMINT_DBLSET1(1);
73 const SIMINT_DBLTYPE const_10 = SIMINT_DBLSET1(10);
74 const SIMINT_DBLTYPE const_11 = SIMINT_DBLSET1(11);
75 const SIMINT_DBLTYPE const_12 = SIMINT_DBLSET1(12);
76 const SIMINT_DBLTYPE const_2 = SIMINT_DBLSET1(2);
77 const SIMINT_DBLTYPE const_3 = SIMINT_DBLSET1(3);
78 const SIMINT_DBLTYPE const_4 = SIMINT_DBLSET1(4);
79 const SIMINT_DBLTYPE const_5 = SIMINT_DBLSET1(5);
80 const SIMINT_DBLTYPE const_6 = SIMINT_DBLSET1(6);
81 const SIMINT_DBLTYPE const_7 = SIMINT_DBLSET1(7);
82 const SIMINT_DBLTYPE const_8 = SIMINT_DBLSET1(8);
83 const SIMINT_DBLTYPE const_9 = SIMINT_DBLSET1(9);
84 const SIMINT_DBLTYPE one_half = SIMINT_DBLSET1(0.5);
85
86
87 ////////////////////////////////////////
88 // Loop over shells and primitives
89 ////////////////////////////////////////
90
91 real_abcd = 0;
92 istart = 0;
93 for(ab = 0; ab < P.nshell12_clip; ++ab)
94 {
95 const int iend = istart + P.nprim12[ab];
96
97 cd = 0;
98 jstart = 0;
99
100 for(cd = 0; cd < Q.nshell12_clip; cd += SIMINT_NSHELL_SIMD)
101 {
102 const int nshellbatch = ((cd + SIMINT_NSHELL_SIMD) > Q.nshell12_clip) ? Q.nshell12_clip - cd : SIMINT_NSHELL_SIMD;
103 int jend = jstart;
104 for(i = 0; i < nshellbatch; i++)
105 jend += Q.nprim12[cd+i];
106
107 // Clear the beginning of the workspace (where we are accumulating integrals)
108 memset(work, 0, SIMINT_NSHELL_SIMD * 1113 * sizeof(double));
109 abcd = 0;
110
111
112 for(i = istart; i < iend; ++i)
113 {
114 SIMINT_DBLTYPE bra_screen_max; // only used if check_screen
115
116 if(check_screen)
117 {
118 // Skip this whole thing if always insignificant
119 if((P.screen[i] * Q.screen_max) < screen_tol)
120 continue;
121 bra_screen_max = SIMINT_DBLSET1(P.screen[i]);
122 }
123
124 icd = 0;
125 iprimcd = 0;
126 nprim_icd = Q.nprim12[cd];
127 double * restrict PRIM_PTR_INT__k_s_p_s = INT__k_s_p_s + abcd * 108;
128 double * restrict PRIM_PTR_INT__l_s_p_s = INT__l_s_p_s + abcd * 135;
129 double * restrict PRIM_PTR_INT__m_s_p_s = INT__m_s_p_s + abcd * 165;
130 double * restrict PRIM_PTR_INT__n_s_p_s = INT__n_s_p_s + abcd * 198;
131 double * restrict PRIM_PTR_INT__o_s_p_s = INT__o_s_p_s + abcd * 234;
132 double * restrict PRIM_PTR_INT__q_s_p_s = INT__q_s_p_s + abcd * 273;
133
134
135
136 // Load these one per loop over i
137 const SIMINT_DBLTYPE P_alpha = SIMINT_DBLSET1(P.alpha[i]);
138 const SIMINT_DBLTYPE P_prefac = SIMINT_DBLSET1(P.prefac[i]);
139 const SIMINT_DBLTYPE Pxyz[3] = { SIMINT_DBLSET1(P.x[i]), SIMINT_DBLSET1(P.y[i]), SIMINT_DBLSET1(P.z[i]) };
140
141 const SIMINT_DBLTYPE P_PA[3] = { SIMINT_DBLSET1(P.PA_x[i]), SIMINT_DBLSET1(P.PA_y[i]), SIMINT_DBLSET1(P.PA_z[i]) };
142
143 for(j = jstart; j < jend; j += SIMINT_SIMD_LEN)
144 {
145 // calculate the shell offsets
146 // these are the offset from the shell pointed to by cd
147 // for each element
148 int shelloffsets[SIMINT_SIMD_LEN] = {0};
149 int lastoffset = 0;
150 const int nlane = ( ((j + SIMINT_SIMD_LEN) < jend) ? SIMINT_SIMD_LEN : (jend - j));
151
152 if((iprimcd + SIMINT_SIMD_LEN) >= nprim_icd)
153 {
154 // Handle if the first element of the vector is a new shell
155 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
156 {
157 nprim_icd += Q.nprim12[cd + (++icd)];
158 PRIM_PTR_INT__k_s_p_s += 108;
159 PRIM_PTR_INT__l_s_p_s += 135;
160 PRIM_PTR_INT__m_s_p_s += 165;
161 PRIM_PTR_INT__n_s_p_s += 198;
162 PRIM_PTR_INT__o_s_p_s += 234;
163 PRIM_PTR_INT__q_s_p_s += 273;
164 }
165 iprimcd++;
166 for(n = 1; n < SIMINT_SIMD_LEN; ++n)
167 {
168 if(iprimcd >= nprim_icd && ((icd+1) < nshellbatch))
169 {
170 shelloffsets[n] = shelloffsets[n-1] + 1;
171 lastoffset++;
172 nprim_icd += Q.nprim12[cd + (++icd)];
173 }
174 else
175 shelloffsets[n] = shelloffsets[n-1];
176 iprimcd++;
177 }
178 }
179 else
180 iprimcd += SIMINT_SIMD_LEN;
181
182 // Do we have to compute this vector (or has it been screened out)?
183 // (not_screened != 0 means we have to do this vector)
184 if(check_screen)
185 {
186 const double vmax = vector_max(SIMINT_MUL(bra_screen_max, SIMINT_DBLLOAD(Q.screen, j)));
187 if(vmax < screen_tol)
188 {
189 PRIM_PTR_INT__k_s_p_s += lastoffset*108;
190 PRIM_PTR_INT__l_s_p_s += lastoffset*135;
191 PRIM_PTR_INT__m_s_p_s += lastoffset*165;
192 PRIM_PTR_INT__n_s_p_s += lastoffset*198;
193 PRIM_PTR_INT__o_s_p_s += lastoffset*234;
194 PRIM_PTR_INT__q_s_p_s += lastoffset*273;
195 continue;
196 }
197 }
198
199 const SIMINT_DBLTYPE Q_alpha = SIMINT_DBLLOAD(Q.alpha, j);
200 const SIMINT_DBLTYPE PQalpha_mul = SIMINT_MUL(P_alpha, Q_alpha);
201 const SIMINT_DBLTYPE PQalpha_sum = SIMINT_ADD(P_alpha, Q_alpha);
202 const SIMINT_DBLTYPE one_over_PQalpha_sum = SIMINT_DIV(const_1, PQalpha_sum);
203
204
205 /* construct R2 = (Px - Qx)**2 + (Py - Qy)**2 + (Pz -Qz)**2 */
206 SIMINT_DBLTYPE PQ[3];
207 PQ[0] = SIMINT_SUB(Pxyz[0], SIMINT_DBLLOAD(Q.x, j));
208 PQ[1] = SIMINT_SUB(Pxyz[1], SIMINT_DBLLOAD(Q.y, j));
209 PQ[2] = SIMINT_SUB(Pxyz[2], SIMINT_DBLLOAD(Q.z, j));
210 SIMINT_DBLTYPE R2 = SIMINT_MUL(PQ[0], PQ[0]);
211 R2 = SIMINT_FMADD(PQ[1], PQ[1], R2);
212 R2 = SIMINT_FMADD(PQ[2], PQ[2], R2);
213
214 const SIMINT_DBLTYPE alpha = SIMINT_MUL(PQalpha_mul, one_over_PQalpha_sum); // alpha from MEST
215 const SIMINT_DBLTYPE one_over_p = SIMINT_DIV(const_1, P_alpha);
216 const SIMINT_DBLTYPE one_over_q = SIMINT_DIV(const_1, Q_alpha);
217 const SIMINT_DBLTYPE one_over_2p = SIMINT_MUL(one_half, one_over_p);
218 const SIMINT_DBLTYPE one_over_2q = SIMINT_MUL(one_half, one_over_q);
219 const SIMINT_DBLTYPE one_over_2pq = SIMINT_MUL(one_half, one_over_PQalpha_sum);
220 const SIMINT_DBLTYPE Q_PA[3] = { SIMINT_DBLLOAD(Q.PA_x, j), SIMINT_DBLLOAD(Q.PA_y, j), SIMINT_DBLLOAD(Q.PA_z, j) };
221
222 // NOTE: Minus sign!
223 const SIMINT_DBLTYPE a_over_p = SIMINT_MUL(SIMINT_NEG(alpha), one_over_p);
224 SIMINT_DBLTYPE aop_PQ[3];
225 aop_PQ[0] = SIMINT_MUL(a_over_p, PQ[0]);
226 aop_PQ[1] = SIMINT_MUL(a_over_p, PQ[1]);
227 aop_PQ[2] = SIMINT_MUL(a_over_p, PQ[2]);
228
229 SIMINT_DBLTYPE a_over_q = SIMINT_MUL(alpha, one_over_q);
230 SIMINT_DBLTYPE aoq_PQ[3];
231 aoq_PQ[0] = SIMINT_MUL(a_over_q, PQ[0]);
232 aoq_PQ[1] = SIMINT_MUL(a_over_q, PQ[1]);
233 aoq_PQ[2] = SIMINT_MUL(a_over_q, PQ[2]);
234 // Put a minus sign here so we don't have to in RR routines
235 a_over_q = SIMINT_NEG(a_over_q);
236
237
238 //////////////////////////////////////////////
239 // Fjt function section
240 // Maximum v value: 13
241 //////////////////////////////////////////////
242 // The parameter to the Fjt function
243 const SIMINT_DBLTYPE F_x = SIMINT_MUL(R2, alpha);
244
245
246 const SIMINT_DBLTYPE Q_prefac = mask_load(nlane, Q.prefac + j);
247
248
249 boys_F_split(PRIM_INT__s_s_s_s, F_x, 13);
250 SIMINT_DBLTYPE prefac = SIMINT_SQRT(one_over_PQalpha_sum);
251 prefac = SIMINT_MUL(SIMINT_MUL(P_prefac, Q_prefac), prefac);
252 for(n = 0; n <= 13; n++)
253 PRIM_INT__s_s_s_s[n] = SIMINT_MUL(PRIM_INT__s_s_s_s[n], prefac);
254
255 //////////////////////////////////////////////
256 // Primitive integrals: Vertical recurrance
257 //////////////////////////////////////////////
258
259 const SIMINT_DBLTYPE vrr_const_1_over_2p = one_over_2p;
260 const SIMINT_DBLTYPE vrr_const_2_over_2p = SIMINT_MUL(const_2, one_over_2p);
261 const SIMINT_DBLTYPE vrr_const_3_over_2p = SIMINT_MUL(const_3, one_over_2p);
262 const SIMINT_DBLTYPE vrr_const_4_over_2p = SIMINT_MUL(const_4, one_over_2p);
263 const SIMINT_DBLTYPE vrr_const_5_over_2p = SIMINT_MUL(const_5, one_over_2p);
264 const SIMINT_DBLTYPE vrr_const_6_over_2p = SIMINT_MUL(const_6, one_over_2p);
265 const SIMINT_DBLTYPE vrr_const_7_over_2p = SIMINT_MUL(const_7, one_over_2p);
266 const SIMINT_DBLTYPE vrr_const_8_over_2p = SIMINT_MUL(const_8, one_over_2p);
267 const SIMINT_DBLTYPE vrr_const_9_over_2p = SIMINT_MUL(const_9, one_over_2p);
268 const SIMINT_DBLTYPE vrr_const_10_over_2p = SIMINT_MUL(const_10, one_over_2p);
269 const SIMINT_DBLTYPE vrr_const_11_over_2p = SIMINT_MUL(const_11, one_over_2p);
270 const SIMINT_DBLTYPE vrr_const_1_over_2pq = one_over_2pq;
271 const SIMINT_DBLTYPE vrr_const_2_over_2pq = SIMINT_MUL(const_2, one_over_2pq);
272 const SIMINT_DBLTYPE vrr_const_3_over_2pq = SIMINT_MUL(const_3, one_over_2pq);
273 const SIMINT_DBLTYPE vrr_const_4_over_2pq = SIMINT_MUL(const_4, one_over_2pq);
274 const SIMINT_DBLTYPE vrr_const_5_over_2pq = SIMINT_MUL(const_5, one_over_2pq);
275 const SIMINT_DBLTYPE vrr_const_6_over_2pq = SIMINT_MUL(const_6, one_over_2pq);
276 const SIMINT_DBLTYPE vrr_const_7_over_2pq = SIMINT_MUL(const_7, one_over_2pq);
277 const SIMINT_DBLTYPE vrr_const_8_over_2pq = SIMINT_MUL(const_8, one_over_2pq);
278 const SIMINT_DBLTYPE vrr_const_9_over_2pq = SIMINT_MUL(const_9, one_over_2pq);
279 const SIMINT_DBLTYPE vrr_const_10_over_2pq = SIMINT_MUL(const_10, one_over_2pq);
280 const SIMINT_DBLTYPE vrr_const_11_over_2pq = SIMINT_MUL(const_11, one_over_2pq);
281 const SIMINT_DBLTYPE vrr_const_12_over_2pq = SIMINT_MUL(const_12, one_over_2pq);
282
283
284
285 // Forming PRIM_INT__p_s_s_s[13 * 3];
286 for(n = 0; n < 13; ++n) // loop over orders of auxiliary function
287 {
288
289 PRIM_INT__p_s_s_s[n * 3 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__s_s_s_s[n * 1 + 0]);
290 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]);
291
292 PRIM_INT__p_s_s_s[n * 3 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__s_s_s_s[n * 1 + 0]);
293 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]);
294
295 PRIM_INT__p_s_s_s[n * 3 + 2] = SIMINT_MUL(P_PA[2], PRIM_INT__s_s_s_s[n * 1 + 0]);
296 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]);
297
298 }
299
300
301
302 // Forming PRIM_INT__d_s_s_s[12 * 6];
303 for(n = 0; n < 12; ++n) // loop over orders of auxiliary function
304 {
305
306 PRIM_INT__d_s_s_s[n * 6 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__p_s_s_s[n * 3 + 0]);
307 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]);
308 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]);
309
310 PRIM_INT__d_s_s_s[n * 6 + 3] = SIMINT_MUL(P_PA[1], PRIM_INT__p_s_s_s[n * 3 + 1]);
311 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]);
312 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]);
313
314 PRIM_INT__d_s_s_s[n * 6 + 5] = SIMINT_MUL(P_PA[2], PRIM_INT__p_s_s_s[n * 3 + 2]);
315 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]);
316 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]);
317
318 }
319
320
321
322 // Forming PRIM_INT__f_s_s_s[11 * 10];
323 for(n = 0; n < 11; ++n) // loop over orders of auxiliary function
324 {
325
326 PRIM_INT__f_s_s_s[n * 10 + 0] = SIMINT_MUL(P_PA[0], PRIM_INT__d_s_s_s[n * 6 + 0]);
327 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]);
328 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]);
329
330 PRIM_INT__f_s_s_s[n * 10 + 1] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 0]);
331 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]);
332
333 PRIM_INT__f_s_s_s[n * 10 + 6] = SIMINT_MUL(P_PA[1], PRIM_INT__d_s_s_s[n * 6 + 3]);
334 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]);
335 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]);
336
337 PRIM_INT__f_s_s_s[n * 10 + 9] = SIMINT_MUL(P_PA[2], PRIM_INT__d_s_s_s[n * 6 + 5]);
338 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]);
339 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]);
340
341 }
342
343
344 VRR_I_g_s_s_s(
345 PRIM_INT__g_s_s_s,
346 PRIM_INT__f_s_s_s,
347 PRIM_INT__d_s_s_s,
348 P_PA,
349 a_over_p,
350 aop_PQ,
351 one_over_2p,
352 10);
353
354
355 VRR_I_h_s_s_s(
356 PRIM_INT__h_s_s_s,
357 PRIM_INT__g_s_s_s,
358 PRIM_INT__f_s_s_s,
359 P_PA,
360 a_over_p,
361 aop_PQ,
362 one_over_2p,
363 9);
364
365
366 ostei_general_vrr1_I(6, 8,
367 one_over_2p, a_over_p, aop_PQ, P_PA,
368 PRIM_INT__h_s_s_s, PRIM_INT__g_s_s_s, PRIM_INT__i_s_s_s);
369
370
371 ostei_general_vrr1_I(7, 7,
372 one_over_2p, a_over_p, aop_PQ, P_PA,
373 PRIM_INT__i_s_s_s, PRIM_INT__h_s_s_s, PRIM_INT__k_s_s_s);
374
375
376 ostei_general_vrr_K(7, 0, 1, 0, 1,
377 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
378 PRIM_INT__k_s_s_s, NULL, NULL, PRIM_INT__i_s_s_s, NULL, PRIM_INT__k_s_p_s);
379
380
381 ostei_general_vrr1_I(8, 6,
382 one_over_2p, a_over_p, aop_PQ, P_PA,
383 PRIM_INT__k_s_s_s, PRIM_INT__i_s_s_s, PRIM_INT__l_s_s_s);
384
385
386 ostei_general_vrr_K(8, 0, 1, 0, 1,
387 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
388 PRIM_INT__l_s_s_s, NULL, NULL, PRIM_INT__k_s_s_s, NULL, PRIM_INT__l_s_p_s);
389
390
391 ostei_general_vrr1_I(9, 5,
392 one_over_2p, a_over_p, aop_PQ, P_PA,
393 PRIM_INT__l_s_s_s, PRIM_INT__k_s_s_s, PRIM_INT__m_s_s_s);
394
395
396 ostei_general_vrr_K(9, 0, 1, 0, 1,
397 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
398 PRIM_INT__m_s_s_s, NULL, NULL, PRIM_INT__l_s_s_s, NULL, PRIM_INT__m_s_p_s);
399
400
401 ostei_general_vrr1_I(10, 4,
402 one_over_2p, a_over_p, aop_PQ, P_PA,
403 PRIM_INT__m_s_s_s, PRIM_INT__l_s_s_s, PRIM_INT__n_s_s_s);
404
405
406 ostei_general_vrr_K(10, 0, 1, 0, 1,
407 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
408 PRIM_INT__n_s_s_s, NULL, NULL, PRIM_INT__m_s_s_s, NULL, PRIM_INT__n_s_p_s);
409
410
411 ostei_general_vrr1_I(11, 3,
412 one_over_2p, a_over_p, aop_PQ, P_PA,
413 PRIM_INT__n_s_s_s, PRIM_INT__m_s_s_s, PRIM_INT__o_s_s_s);
414
415
416 ostei_general_vrr_K(11, 0, 1, 0, 1,
417 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
418 PRIM_INT__o_s_s_s, NULL, NULL, PRIM_INT__n_s_s_s, NULL, PRIM_INT__o_s_p_s);
419
420
421 ostei_general_vrr1_I(12, 2,
422 one_over_2p, a_over_p, aop_PQ, P_PA,
423 PRIM_INT__o_s_s_s, PRIM_INT__n_s_s_s, PRIM_INT__q_s_s_s);
424
425
426 ostei_general_vrr_K(12, 0, 1, 0, 1,
427 one_over_2q, a_over_q, one_over_2pq, aoq_PQ, Q_PA,
428 PRIM_INT__q_s_s_s, NULL, NULL, PRIM_INT__o_s_s_s, NULL, PRIM_INT__q_s_p_s);
429
430
431
432
433 ////////////////////////////////////
434 // Accumulate contracted integrals
435 ////////////////////////////////////
436 if(lastoffset == 0)
437 {
438 contract_all(108, PRIM_INT__k_s_p_s, PRIM_PTR_INT__k_s_p_s);
439 contract_all(135, PRIM_INT__l_s_p_s, PRIM_PTR_INT__l_s_p_s);
440 contract_all(165, PRIM_INT__m_s_p_s, PRIM_PTR_INT__m_s_p_s);
441 contract_all(198, PRIM_INT__n_s_p_s, PRIM_PTR_INT__n_s_p_s);
442 contract_all(234, PRIM_INT__o_s_p_s, PRIM_PTR_INT__o_s_p_s);
443 contract_all(273, PRIM_INT__q_s_p_s, PRIM_PTR_INT__q_s_p_s);
444 }
445 else
446 {
447 contract(108, shelloffsets, PRIM_INT__k_s_p_s, PRIM_PTR_INT__k_s_p_s);
448 contract(135, shelloffsets, PRIM_INT__l_s_p_s, PRIM_PTR_INT__l_s_p_s);
449 contract(165, shelloffsets, PRIM_INT__m_s_p_s, PRIM_PTR_INT__m_s_p_s);
450 contract(198, shelloffsets, PRIM_INT__n_s_p_s, PRIM_PTR_INT__n_s_p_s);
451 contract(234, shelloffsets, PRIM_INT__o_s_p_s, PRIM_PTR_INT__o_s_p_s);
452 contract(273, shelloffsets, PRIM_INT__q_s_p_s, PRIM_PTR_INT__q_s_p_s);
453 PRIM_PTR_INT__k_s_p_s += lastoffset*108;
454 PRIM_PTR_INT__l_s_p_s += lastoffset*135;
455 PRIM_PTR_INT__m_s_p_s += lastoffset*165;
456 PRIM_PTR_INT__n_s_p_s += lastoffset*198;
457 PRIM_PTR_INT__o_s_p_s += lastoffset*234;
458 PRIM_PTR_INT__q_s_p_s += lastoffset*273;
459 }
460
461 } // close loop over j
462 } // close loop over i
463
464 //Advance to the next batch
465 jstart = SIMINT_SIMD_ROUND(jend);
466
467 //////////////////////////////////////////////
468 // Contracted integrals: Horizontal recurrance
469 //////////////////////////////////////////////
470
471
472 const double hAB[3] = { P.AB_x[ab], P.AB_y[ab], P.AB_z[ab] };
473
474
475 for(abcd = 0; abcd < nshellbatch; ++abcd, ++real_abcd)
476 {
477
478 // set up HRR pointers
479 double const * restrict HRR_INT__k_s_p_s = INT__k_s_p_s + abcd * 108;
480 double const * restrict HRR_INT__l_s_p_s = INT__l_s_p_s + abcd * 135;
481 double const * restrict HRR_INT__m_s_p_s = INT__m_s_p_s + abcd * 165;
482 double const * restrict HRR_INT__n_s_p_s = INT__n_s_p_s + abcd * 198;
483 double const * restrict HRR_INT__o_s_p_s = INT__o_s_p_s + abcd * 234;
484 double const * restrict HRR_INT__q_s_p_s = INT__q_s_p_s + abcd * 273;
485 double * restrict HRR_INT__k_h_p_s = INT__k_h_p_s + real_abcd * 2268;
486
487 // form INT__k_p_p_s
488 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);
489
490 // form INT__l_p_p_s
491 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);
492
493 // form INT__m_p_p_s
494 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);
495
496 // form INT__n_p_p_s
497 ostei_general_hrr_J(10, 1, 1, 0, hAB, HRR_INT__o_s_p_s, HRR_INT__n_s_p_s, HRR_INT__n_p_p_s);
498
499 // form INT__o_p_p_s
500 ostei_general_hrr_J(11, 1, 1, 0, hAB, HRR_INT__q_s_p_s, HRR_INT__o_s_p_s, HRR_INT__o_p_p_s);
501
502 // form INT__k_d_p_s
503 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);
504
505 // form INT__l_d_p_s
506 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);
507
508 // form INT__m_d_p_s
509 ostei_general_hrr_J(9, 2, 1, 0, hAB, HRR_INT__n_p_p_s, HRR_INT__m_p_p_s, HRR_INT__m_d_p_s);
510
511 // form INT__n_d_p_s
512 ostei_general_hrr_J(10, 2, 1, 0, hAB, HRR_INT__o_p_p_s, HRR_INT__n_p_p_s, HRR_INT__n_d_p_s);
513
514 // form INT__k_f_p_s
515 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);
516
517 // form INT__l_f_p_s
518 ostei_general_hrr_J(8, 3, 1, 0, hAB, HRR_INT__m_d_p_s, HRR_INT__l_d_p_s, HRR_INT__l_f_p_s);
519
520 // form INT__m_f_p_s
521 ostei_general_hrr_J(9, 3, 1, 0, hAB, HRR_INT__n_d_p_s, HRR_INT__m_d_p_s, HRR_INT__m_f_p_s);
522
523 // form INT__k_g_p_s
524 ostei_general_hrr_J(7, 4, 1, 0, hAB, HRR_INT__l_f_p_s, HRR_INT__k_f_p_s, HRR_INT__k_g_p_s);
525
526 // form INT__l_g_p_s
527 ostei_general_hrr_J(8, 4, 1, 0, hAB, HRR_INT__m_f_p_s, HRR_INT__l_f_p_s, HRR_INT__l_g_p_s);
528
529 // form INT__k_h_p_s
530 ostei_general_hrr_J(7, 5, 1, 0, hAB, HRR_INT__l_g_p_s, HRR_INT__k_g_p_s, HRR_INT__k_h_p_s);
531
532
533 } // close HRR loop
534
535
536 } // close loop cdbatch
537
538 istart = iend;
539 } // close loop over ab
540
541 return P.nshell12_clip * Q.nshell12_clip;
542 }
543
ostei_h_k_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__h_k_p_s)544 int ostei_h_k_p_s(struct simint_multi_shellpair const P,
545 struct simint_multi_shellpair const Q,
546 double screen_tol,
547 double * const restrict work,
548 double * const restrict INT__h_k_p_s)
549 {
550 double P_AB[3*P.nshell12];
551 struct simint_multi_shellpair P_tmp = P;
552 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
553 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
554 P_tmp.AB_x = P_AB;
555 P_tmp.AB_y = P_AB + P.nshell12;
556 P_tmp.AB_z = P_AB + 2*P.nshell12;
557
558 for(int i = 0; i < P.nshell12; i++)
559 {
560 P_tmp.AB_x[i] = -P.AB_x[i];
561 P_tmp.AB_y[i] = -P.AB_y[i];
562 P_tmp.AB_z[i] = -P.AB_z[i];
563 }
564
565 int ret = ostei_k_h_p_s(P_tmp, Q, screen_tol, work, INT__h_k_p_s);
566 double buffer[2268] SIMINT_ALIGN_ARRAY_DBL;
567
568 for(int q = 0; q < ret; q++)
569 {
570 int idx = 0;
571 for(int a = 0; a < 21; ++a)
572 for(int b = 0; b < 36; ++b)
573 for(int c = 0; c < 3; ++c)
574 for(int d = 0; d < 1; ++d)
575 buffer[idx++] = INT__h_k_p_s[q*2268+b*63+a*3+c*1+d];
576
577 memcpy(INT__h_k_p_s+q*2268, buffer, 2268*sizeof(double));
578 }
579
580 return ret;
581 }
582
ostei_k_h_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__k_h_s_p)583 int ostei_k_h_s_p(struct simint_multi_shellpair const P,
584 struct simint_multi_shellpair const Q,
585 double screen_tol,
586 double * const restrict work,
587 double * const restrict INT__k_h_s_p)
588 {
589 double Q_AB[3*Q.nshell12];
590 struct simint_multi_shellpair Q_tmp = Q;
591 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
592 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
593 Q_tmp.AB_x = Q_AB;
594 Q_tmp.AB_y = Q_AB + Q.nshell12;
595 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
596
597 for(int i = 0; i < Q.nshell12; i++)
598 {
599 Q_tmp.AB_x[i] = -Q.AB_x[i];
600 Q_tmp.AB_y[i] = -Q.AB_y[i];
601 Q_tmp.AB_z[i] = -Q.AB_z[i];
602 }
603
604 int ret = ostei_k_h_p_s(P, Q_tmp, screen_tol, work, INT__k_h_s_p);
605 double buffer[2268] SIMINT_ALIGN_ARRAY_DBL;
606
607 for(int q = 0; q < ret; q++)
608 {
609 int idx = 0;
610 for(int a = 0; a < 36; ++a)
611 for(int b = 0; b < 21; ++b)
612 for(int c = 0; c < 1; ++c)
613 for(int d = 0; d < 3; ++d)
614 buffer[idx++] = INT__k_h_s_p[q*2268+a*63+b*3+d*1+c];
615
616 memcpy(INT__k_h_s_p+q*2268, buffer, 2268*sizeof(double));
617 }
618
619 return ret;
620 }
621
ostei_h_k_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__h_k_s_p)622 int ostei_h_k_s_p(struct simint_multi_shellpair const P,
623 struct simint_multi_shellpair const Q,
624 double screen_tol,
625 double * const restrict work,
626 double * const restrict INT__h_k_s_p)
627 {
628 double P_AB[3*P.nshell12];
629 struct simint_multi_shellpair P_tmp = P;
630 P_tmp.PA_x = P.PB_x; P_tmp.PA_y = P.PB_y; P_tmp.PA_z = P.PB_z;
631 P_tmp.PB_x = P.PA_x; P_tmp.PB_y = P.PA_y; P_tmp.PB_z = P.PA_z;
632 P_tmp.AB_x = P_AB;
633 P_tmp.AB_y = P_AB + P.nshell12;
634 P_tmp.AB_z = P_AB + 2*P.nshell12;
635
636 for(int i = 0; i < P.nshell12; i++)
637 {
638 P_tmp.AB_x[i] = -P.AB_x[i];
639 P_tmp.AB_y[i] = -P.AB_y[i];
640 P_tmp.AB_z[i] = -P.AB_z[i];
641 }
642
643 double Q_AB[3*Q.nshell12];
644 struct simint_multi_shellpair Q_tmp = Q;
645 Q_tmp.PA_x = Q.PB_x; Q_tmp.PA_y = Q.PB_y; Q_tmp.PA_z = Q.PB_z;
646 Q_tmp.PB_x = Q.PA_x; Q_tmp.PB_y = Q.PA_y; Q_tmp.PB_z = Q.PA_z;
647 Q_tmp.AB_x = Q_AB;
648 Q_tmp.AB_y = Q_AB + Q.nshell12;
649 Q_tmp.AB_z = Q_AB + 2*Q.nshell12;
650
651 for(int i = 0; i < Q.nshell12; i++)
652 {
653 Q_tmp.AB_x[i] = -Q.AB_x[i];
654 Q_tmp.AB_y[i] = -Q.AB_y[i];
655 Q_tmp.AB_z[i] = -Q.AB_z[i];
656 }
657
658 int ret = ostei_k_h_p_s(P_tmp, Q_tmp, screen_tol, work, INT__h_k_s_p);
659 double buffer[2268] SIMINT_ALIGN_ARRAY_DBL;
660
661 for(int q = 0; q < ret; q++)
662 {
663 int idx = 0;
664 for(int a = 0; a < 21; ++a)
665 for(int b = 0; b < 36; ++b)
666 for(int c = 0; c < 1; ++c)
667 for(int d = 0; d < 3; ++d)
668 buffer[idx++] = INT__h_k_s_p[q*2268+b*63+a*3+d*1+c];
669
670 memcpy(INT__h_k_s_p+q*2268, buffer, 2268*sizeof(double));
671 }
672
673 return ret;
674 }
675
676