[2e44ac7] | 1 | /* |
---|
| 2 | ########################################################## |
---|
| 3 | # # |
---|
| 4 | # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # |
---|
| 5 | # !! !! # |
---|
| 6 | # !! KEEP THIS CODE CONSISTENT WITH KERNELPY.PY !! # |
---|
| 7 | # !! !! # |
---|
| 8 | # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # |
---|
| 9 | # # |
---|
| 10 | ########################################################## |
---|
| 11 | */ |
---|
| 12 | |
---|
[9ee2756] | 13 | // NOTE: the following macros are defined in generate.py: |
---|
| 14 | // |
---|
[6aee3ab] | 15 | // MAX_PD : the maximum number of dispersity loops allowed for this model, |
---|
| 16 | // which will be at most modelinfo.MAX_PD. |
---|
[9ee2756] | 17 | // NUM_PARS : the number of parameters in the parameter table |
---|
| 18 | // NUM_VALUES : the number of values to skip at the start of the |
---|
| 19 | // values array before you get to the dispersity values. |
---|
| 20 | // PARAMETER_TABLE : list of parameter declarations used to create the |
---|
| 21 | // ParameterTable type. |
---|
| 22 | // KERNEL_NAME : model_Iq, model_Iqxy or model_Imagnetic. This code is |
---|
| 23 | // included three times, once for each kernel type. |
---|
[2c108a3] | 24 | // MAGNETIC : defined when the magnetic kernel is being instantiated |
---|
| 25 | // NUM_MAGNETIC : the number of magnetic parameters |
---|
[9ee2756] | 26 | // MAGNETIC_PARS : a comma-separated list of indices to the sld |
---|
| 27 | // parameters in the parameter table. |
---|
[e44432d] | 28 | // CALL_VOLUME(form, shell, table) : assign form and shell values |
---|
[6e7ba14] | 29 | // CALL_EFFECTIVE_RADIUS(type, table) : call the R_eff function |
---|
[9ee2756] | 30 | // CALL_IQ(q, table) : call the Iq function for 1D calcs. |
---|
| 31 | // CALL_IQ_A(q, table) : call the Iq function with |q| for 2D data. |
---|
[c036ddb] | 32 | // CALL_FQ(q, F1, F2, table) : call the Fq function for 1D calcs. |
---|
| 33 | // CALL_FQ_A(q, F1, F2, table) : call the Iq function with |q| for 2D data. |
---|
[9ee2756] | 34 | // CALL_IQ_AC(qa, qc, table) : call the Iqxy function for symmetric shapes |
---|
| 35 | // CALL_IQ_ABC(qa, qc, table) : call the Iqxy function for asymmetric shapes |
---|
[108e70e] | 36 | // CALL_IQ_XY(qx, qy, table) : call the Iqxy function for arbitrary models |
---|
[9ee2756] | 37 | // INVALID(table) : test if the current point is feesible to calculate. This |
---|
| 38 | // will be defined in the kernel definition file. |
---|
[ff10479] | 39 | // PROJECTION : equirectangular=1, sinusoidal=2 |
---|
| 40 | // see explore/jitter.py for definitions. |
---|
[767dca8] | 41 | |
---|
[01c8d9e] | 42 | |
---|
[03cac08] | 43 | #ifndef _PAR_BLOCK_ // protected block so we can include this code twice. |
---|
| 44 | #define _PAR_BLOCK_ |
---|
[2e44ac7] | 45 | |
---|
| 46 | typedef struct { |
---|
[60eab2a] | 47 | #if MAX_PD > 0 |
---|
[2c108a3] | 48 | int32_t pd_par[MAX_PD]; // id of the nth dispersity variable |
---|
| 49 | int32_t pd_length[MAX_PD]; // length of the nth dispersity weight vector |
---|
[0a7e5eb4] | 50 | int32_t pd_offset[MAX_PD]; // offset of pd weights in the value & weight vector |
---|
[5cf3c33] | 51 | int32_t pd_stride[MAX_PD]; // stride to move to the next index at this level |
---|
[60eab2a] | 52 | #endif // MAX_PD > 0 |
---|
[bde38b5] | 53 | int32_t num_eval; // total number of voxels in hypercube |
---|
| 54 | int32_t num_weights; // total length of the weights vector |
---|
[5ff1b03] | 55 | int32_t num_active; // number of non-trivial pd loops |
---|
[8698a0d] | 56 | int32_t theta_par; // id of first orientation variable |
---|
[2e44ac7] | 57 | } ProblemDetails; |
---|
| 58 | |
---|
[bde38b5] | 59 | // Intel HD 4000 needs private arrays to be a multiple of 4 long |
---|
[2e44ac7] | 60 | typedef struct { |
---|
[56547a8] | 61 | PARAMETER_TABLE |
---|
[bde38b5] | 62 | } ParameterTable; |
---|
| 63 | typedef union { |
---|
| 64 | ParameterTable table; |
---|
| 65 | double vector[4*((NUM_PARS+3)/4)]; |
---|
[2e44ac7] | 66 | } ParameterBlock; |
---|
[9eb3632] | 67 | #endif // _PAR_BLOCK_ |
---|
[03cac08] | 68 | |
---|
[9ee2756] | 69 | #if defined(MAGNETIC) && NUM_MAGNETIC > 0 |
---|
| 70 | // ===== Helper functions for magnetism ===== |
---|
[a4280bd] | 71 | |
---|
[32e3c9b] | 72 | // Return value restricted between low and high |
---|
| 73 | static double clip(double value, double low, double high) |
---|
| 74 | { |
---|
[b966a96] | 75 | return (value < low ? low : (value > high ? high : value)); |
---|
[32e3c9b] | 76 | } |
---|
| 77 | |
---|
| 78 | // Compute spin cross sections given in_spin and out_spin |
---|
| 79 | // To convert spin cross sections to sld b: |
---|
| 80 | // uu * (sld - m_sigma_x); |
---|
| 81 | // dd * (sld + m_sigma_x); |
---|
[2c108a3] | 82 | // ud * (m_sigma_y - 1j*m_sigma_z); |
---|
| 83 | // du * (m_sigma_y + 1j*m_sigma_z); |
---|
| 84 | // weights for spin crosssections: dd du real, ud real, uu, du imag, ud imag |
---|
[885753a] | 85 | static void set_spin_weights(double in_spin, double out_spin, double weight[6]) |
---|
[32e3c9b] | 86 | { |
---|
[5e1875c] | 87 | |
---|
| 88 | const double norm; |
---|
[b966a96] | 89 | in_spin = clip(in_spin, 0.0, 1.0); |
---|
| 90 | out_spin = clip(out_spin, 0.0, 1.0); |
---|
[7c35fda] | 91 | // Previous version of this function took the square root of the weights, |
---|
[c036ddb] | 92 | // under the assumption that |
---|
[7c35fda] | 93 | // |
---|
[885753a] | 94 | // w*I(q, rho1, rho2, ...) = I(q, sqrt(w)*rho1, sqrt(w)*rho2, ...) |
---|
[7c35fda] | 95 | // |
---|
| 96 | // However, since the weights are applied to the final intensity and |
---|
| 97 | // are not interned inside the I(q) function, we want the full |
---|
[5e1875c] | 98 | // weight and not the square root. Anyway no function will ever use |
---|
| 99 | // set_spin_weights as part of calculating an amplitude, as the weights are |
---|
| 100 | // related to polarisation efficiency of the instrument. The weights serve to |
---|
| 101 | // construct various magnet scattering cross sections, which are linear combinations |
---|
| 102 | // of the spin-resolved cross sections. The polarisation efficiency e_in and e_out |
---|
| 103 | // are parameters ranging from 0.5 (unpolarised) beam to 1 (perfect optics). |
---|
| 104 | // For in_spin or out_spin <0.5 one assumes a CS, where the spin is reversed/flipped |
---|
| 105 | // with respect to the initial supermirror polariser. The actual polarisation efficiency |
---|
| 106 | // in this case is however e_in/out = 1-in/out_spin. |
---|
| 107 | |
---|
| 108 | if (out_spin < 0.5){norm=1-out_spin;} |
---|
| 109 | else{norm=out_spin;} |
---|
| 110 | |
---|
| 111 | |
---|
| 112 | // The norm is needed to make sure that the scattering cross sections are |
---|
| 113 | //correctly weighted, such that the sum of spin-resolved measurements adds up to |
---|
| 114 | // the unpolarised or half-polarised scattering cross section. No intensity weighting |
---|
| 115 | // needed on the incoming polariser side (assuming that a user), has normalised |
---|
| 116 | // to the incoming flux with polariser in for SANSPOl and unpolarised beam, respectively. |
---|
| 117 | |
---|
| 118 | |
---|
| 119 | weight[0] = (1.0-in_spin) * (1.0-out_spin) / norm; // dd |
---|
| 120 | weight[1] = (1.0-in_spin) * out_spin / norm; // du |
---|
| 121 | weight[2] = in_spin * (1.0-out_spin) / norm; // ud |
---|
| 122 | weight[3] = in_spin * out_spin / norm; // uu |
---|
[885753a] | 123 | weight[4] = weight[1]; // du.imag |
---|
| 124 | weight[5] = weight[2]; // ud.imag |
---|
[a4280bd] | 125 | } |
---|
| 126 | |
---|
[9ee2756] | 127 | // Compute the magnetic sld |
---|
[2c108a3] | 128 | static double mag_sld( |
---|
[885753a] | 129 | const unsigned int xs, // 0=dd, 1=du.real, 2=ud.real, 3=uu, 4=du.imag, 5=ud.imag |
---|
[2c108a3] | 130 | const double qx, const double qy, |
---|
| 131 | const double px, const double py, |
---|
| 132 | const double sld, |
---|
| 133 | const double mx, const double my, const double mz |
---|
| 134 | ) |
---|
[a4280bd] | 135 | { |
---|
[2c108a3] | 136 | if (xs < 4) { |
---|
[a4280bd] | 137 | const double perp = qy*mx - qx*my; |
---|
[2c108a3] | 138 | switch (xs) { |
---|
[aadec17] | 139 | default: // keep compiler happy; condition ensures xs in [0,1,2,3] |
---|
[5e1875c] | 140 | case 0: // dd => sld - D M_perpx |
---|
[2c108a3] | 141 | return sld - px*perp; |
---|
[5e1875c] | 142 | case 1: // du.real => -D M_perpy |
---|
[2c108a3] | 143 | return py*perp; |
---|
[5e1875c] | 144 | case 2: // ud.real => -D M_perpy |
---|
[2c108a3] | 145 | return py*perp; |
---|
[5e1875c] | 146 | case 3: // uu => sld + D M_perpx |
---|
[2c108a3] | 147 | return sld + px*perp; |
---|
| 148 | } |
---|
| 149 | } else { |
---|
| 150 | if (xs== 4) { |
---|
[885753a] | 151 | return -mz; // du.imag => +D M_perpz |
---|
[2c108a3] | 152 | } else { // index == 5 |
---|
[885753a] | 153 | return +mz; // ud.imag => -D M_perpz |
---|
[2c108a3] | 154 | } |
---|
| 155 | } |
---|
[32e3c9b] | 156 | } |
---|
[9ee2756] | 157 | |
---|
[2c108a3] | 158 | |
---|
[9ee2756] | 159 | #endif |
---|
| 160 | |
---|
| 161 | // ===== Helper functions for orientation and jitter ===== |
---|
| 162 | |
---|
[8698a0d] | 163 | // To change the definition of the angles, run explore/angles.py, which |
---|
| 164 | // uses sympy to generate the equations. |
---|
| 165 | |
---|
[9ee2756] | 166 | #if !defined(_QAC_SECTION) && defined(CALL_IQ_AC) |
---|
| 167 | #define _QAC_SECTION |
---|
| 168 | |
---|
| 169 | typedef struct { |
---|
| 170 | double R31, R32; |
---|
| 171 | } QACRotation; |
---|
| 172 | |
---|
| 173 | // Fill in the rotation matrix R from the view angles (theta, phi) and the |
---|
| 174 | // jitter angles (dtheta, dphi). This matrix can be applied to all of the |
---|
| 175 | // (qx, qy) points in the image to produce R*[qx,qy]' = [qa,qc]' |
---|
| 176 | static void |
---|
| 177 | qac_rotation( |
---|
| 178 | QACRotation *rotation, |
---|
| 179 | double theta, double phi, |
---|
| 180 | double dtheta, double dphi) |
---|
[8698a0d] | 181 | { |
---|
| 182 | double sin_theta, cos_theta; |
---|
| 183 | double sin_phi, cos_phi; |
---|
| 184 | |
---|
[9ee2756] | 185 | // reverse view matrix |
---|
[8698a0d] | 186 | SINCOS(theta*M_PI_180, sin_theta, cos_theta); |
---|
| 187 | SINCOS(phi*M_PI_180, sin_phi, cos_phi); |
---|
[9ee2756] | 188 | const double V11 = cos_phi*cos_theta; |
---|
| 189 | const double V12 = sin_phi*cos_theta; |
---|
| 190 | const double V21 = -sin_phi; |
---|
| 191 | const double V22 = cos_phi; |
---|
| 192 | const double V31 = sin_theta*cos_phi; |
---|
| 193 | const double V32 = sin_phi*sin_theta; |
---|
[8698a0d] | 194 | |
---|
[9ee2756] | 195 | // reverse jitter matrix |
---|
| 196 | SINCOS(dtheta*M_PI_180, sin_theta, cos_theta); |
---|
| 197 | SINCOS(dphi*M_PI_180, sin_phi, cos_phi); |
---|
| 198 | const double J31 = sin_theta; |
---|
| 199 | const double J32 = -sin_phi*cos_theta; |
---|
| 200 | const double J33 = cos_phi*cos_theta; |
---|
[8698a0d] | 201 | |
---|
[9ee2756] | 202 | // reverse matrix |
---|
| 203 | rotation->R31 = J31*V11 + J32*V21 + J33*V31; |
---|
| 204 | rotation->R32 = J31*V12 + J32*V22 + J33*V32; |
---|
[8698a0d] | 205 | } |
---|
| 206 | |
---|
[9ee2756] | 207 | // Apply the rotation matrix returned from qac_rotation to the point (qx,qy), |
---|
| 208 | // returning R*[qx,qy]' = [qa,qc]' |
---|
[d86f0fc] | 209 | static void |
---|
[9ee2756] | 210 | qac_apply( |
---|
[ec8d4ac] | 211 | QACRotation *rotation, |
---|
[9ee2756] | 212 | double qx, double qy, |
---|
[70530778] | 213 | double *qab_out, double *qc_out) |
---|
[8698a0d] | 214 | { |
---|
[9ee2756] | 215 | // Indirect calculation of qab, from qab^2 = |q|^2 - qc^2 |
---|
[70530778] | 216 | const double dqc = rotation->R31*qx + rotation->R32*qy; |
---|
| 217 | const double dqab_sq = -dqc*dqc + qx*qx + qy*qy; |
---|
| 218 | //*qab_out = sqrt(fabs(dqab_sq)); |
---|
| 219 | *qab_out = dqab_sq > 0.0 ? sqrt(dqab_sq) : 0.0; |
---|
[9ee2756] | 220 | *qc_out = dqc; |
---|
[8698a0d] | 221 | } |
---|
[9ee2756] | 222 | #endif // _QAC_SECTION |
---|
| 223 | |
---|
| 224 | #if !defined(_QABC_SECTION) && defined(CALL_IQ_ABC) |
---|
| 225 | #define _QABC_SECTION |
---|
| 226 | |
---|
| 227 | typedef struct { |
---|
| 228 | double R11, R12; |
---|
| 229 | double R21, R22; |
---|
| 230 | double R31, R32; |
---|
| 231 | } QABCRotation; |
---|
| 232 | |
---|
| 233 | // Fill in the rotation matrix R from the view angles (theta, phi, psi) and the |
---|
| 234 | // jitter angles (dtheta, dphi, dpsi). This matrix can be applied to all of the |
---|
| 235 | // (qx, qy) points in the image to produce R*[qx,qy]' = [qa,qb,qc]' |
---|
[8698a0d] | 236 | static void |
---|
[9ee2756] | 237 | qabc_rotation( |
---|
| 238 | QABCRotation *rotation, |
---|
[8698a0d] | 239 | double theta, double phi, double psi, |
---|
[9ee2756] | 240 | double dtheta, double dphi, double dpsi) |
---|
[8698a0d] | 241 | { |
---|
| 242 | double sin_theta, cos_theta; |
---|
| 243 | double sin_phi, cos_phi; |
---|
| 244 | double sin_psi, cos_psi; |
---|
| 245 | |
---|
| 246 | // reverse view matrix |
---|
| 247 | SINCOS(theta*M_PI_180, sin_theta, cos_theta); |
---|
| 248 | SINCOS(phi*M_PI_180, sin_phi, cos_phi); |
---|
| 249 | SINCOS(psi*M_PI_180, sin_psi, cos_psi); |
---|
[9ee2756] | 250 | const double V11 = -sin_phi*sin_psi + cos_phi*cos_psi*cos_theta; |
---|
| 251 | const double V12 = sin_phi*cos_psi*cos_theta + sin_psi*cos_phi; |
---|
| 252 | const double V21 = -sin_phi*cos_psi - sin_psi*cos_phi*cos_theta; |
---|
| 253 | const double V22 = -sin_phi*sin_psi*cos_theta + cos_phi*cos_psi; |
---|
[8698a0d] | 254 | const double V31 = sin_theta*cos_phi; |
---|
| 255 | const double V32 = sin_phi*sin_theta; |
---|
| 256 | |
---|
| 257 | // reverse jitter matrix |
---|
[9ee2756] | 258 | SINCOS(dtheta*M_PI_180, sin_theta, cos_theta); |
---|
| 259 | SINCOS(dphi*M_PI_180, sin_phi, cos_phi); |
---|
| 260 | SINCOS(dpsi*M_PI_180, sin_psi, cos_psi); |
---|
[8698a0d] | 261 | const double J11 = cos_psi*cos_theta; |
---|
[9ee2756] | 262 | const double J12 = sin_phi*sin_theta*cos_psi + sin_psi*cos_phi; |
---|
| 263 | const double J13 = sin_phi*sin_psi - sin_theta*cos_phi*cos_psi; |
---|
| 264 | const double J21 = -sin_psi*cos_theta; |
---|
| 265 | const double J22 = -sin_phi*sin_psi*sin_theta + cos_phi*cos_psi; |
---|
| 266 | const double J23 = sin_phi*cos_psi + sin_psi*sin_theta*cos_phi; |
---|
[8698a0d] | 267 | const double J31 = sin_theta; |
---|
| 268 | const double J32 = -sin_phi*cos_theta; |
---|
| 269 | const double J33 = cos_phi*cos_theta; |
---|
| 270 | |
---|
| 271 | // reverse matrix |
---|
[9ee2756] | 272 | rotation->R11 = J11*V11 + J12*V21 + J13*V31; |
---|
| 273 | rotation->R12 = J11*V12 + J12*V22 + J13*V32; |
---|
| 274 | rotation->R21 = J21*V11 + J22*V21 + J23*V31; |
---|
| 275 | rotation->R22 = J21*V12 + J22*V22 + J23*V32; |
---|
| 276 | rotation->R31 = J31*V11 + J32*V21 + J33*V31; |
---|
| 277 | rotation->R32 = J31*V12 + J32*V22 + J33*V32; |
---|
[8698a0d] | 278 | } |
---|
| 279 | |
---|
[9ee2756] | 280 | // Apply the rotation matrix returned from qabc_rotation to the point (qx,qy), |
---|
| 281 | // returning R*[qx,qy]' = [qa,qb,qc]' |
---|
[d86f0fc] | 282 | static void |
---|
[9ee2756] | 283 | qabc_apply( |
---|
[ec8d4ac] | 284 | QABCRotation *rotation, |
---|
[9ee2756] | 285 | double qx, double qy, |
---|
| 286 | double *qa_out, double *qb_out, double *qc_out) |
---|
[8698a0d] | 287 | { |
---|
[ec8d4ac] | 288 | *qa_out = rotation->R11*qx + rotation->R12*qy; |
---|
| 289 | *qb_out = rotation->R21*qx + rotation->R22*qy; |
---|
| 290 | *qc_out = rotation->R31*qx + rotation->R32*qy; |
---|
[8698a0d] | 291 | } |
---|
| 292 | |
---|
[9ee2756] | 293 | #endif // _QABC_SECTION |
---|
| 294 | |
---|
| 295 | // ==================== KERNEL CODE ======================== |
---|
[03cac08] | 296 | kernel |
---|
| 297 | void KERNEL_NAME( |
---|
[07646b6] | 298 | int32_t nq, // number of q values |
---|
| 299 | const int32_t pd_start, // where we are in the dispersity loop |
---|
| 300 | const int32_t pd_stop, // where we are stopping in the dispersity loop |
---|
[74e9b5f] | 301 | pglobal const ProblemDetails *details, |
---|
[07646b6] | 302 | pglobal const double *values, // parameter values and distributions |
---|
| 303 | pglobal const double *q, // nq q values, with padding to boundary |
---|
| 304 | pglobal double *result, // nq+1 return values, again with padding |
---|
| 305 | const double cutoff, // cutoff in the dispersity weight product |
---|
[c57ee9e] | 306 | int32_t effective_radius_type // which effective radius to compute |
---|
[2e44ac7] | 307 | ) |
---|
| 308 | { |
---|
[0db7dbd] | 309 | #if defined(USE_GPU) |
---|
[9ee2756] | 310 | // who we are and what element we are working with |
---|
[0db7dbd] | 311 | #if defined(USE_OPENCL) |
---|
[9ee2756] | 312 | const int q_index = get_global_id(0); |
---|
[0db7dbd] | 313 | #else // USE_CUDA |
---|
| 314 | const int q_index = threadIdx.x + blockIdx.x * blockDim.x; |
---|
| 315 | #endif |
---|
[9ee2756] | 316 | if (q_index >= nq) return; |
---|
| 317 | #else |
---|
| 318 | // Define q_index here so that debugging statements can be written to work |
---|
| 319 | // for both OpenCL and DLL using: |
---|
| 320 | // if (q_index == 0) {printf(...);} |
---|
| 321 | int q_index = 0; |
---|
| 322 | #endif |
---|
| 323 | |
---|
[6aee3ab] | 324 | // ** Fill in the local values table ** |
---|
| 325 | // Storage for the current parameter values. |
---|
| 326 | // These will be updated as we walk the dispersity mesh. |
---|
[9eb3632] | 327 | ParameterBlock local_values; |
---|
[6aee3ab] | 328 | // values[0] is scale |
---|
| 329 | // values[1] is background |
---|
| 330 | #ifdef USE_OPENMP |
---|
| 331 | #pragma omp parallel for |
---|
| 332 | #endif |
---|
| 333 | for (int i=0; i < NUM_PARS; i++) { |
---|
| 334 | local_values.vector[i] = values[2+i]; |
---|
| 335 | //if (q_index==0) printf("p%d = %g\n",i, local_values.vector[i]); |
---|
| 336 | } |
---|
| 337 | //if (q_index==0) printf("NUM_VALUES:%d NUM_PARS:%d MAX_PD:%d\n", NUM_VALUES, NUM_PARS, MAX_PD); |
---|
| 338 | //if (q_index==0) printf("start:%d stop:%d\n", pd_start, pd_stop); |
---|
[2e44ac7] | 339 | |
---|
[6aee3ab] | 340 | // ** Precompute magnatism values ** |
---|
[a4280bd] | 341 | #if defined(MAGNETIC) && NUM_MAGNETIC>0 |
---|
[bde38b5] | 342 | // Location of the sld parameters in the parameter vector. |
---|
[9eb3632] | 343 | // These parameters are updated with the effective sld due to magnetism. |
---|
| 344 | const int32_t slds[] = { MAGNETIC_PARS }; |
---|
[32e3c9b] | 345 | |
---|
| 346 | // Interpret polarization cross section. |
---|
[a4280bd] | 347 | // up_frac_i = values[NUM_PARS+2]; |
---|
| 348 | // up_frac_f = values[NUM_PARS+3]; |
---|
| 349 | // up_angle = values[NUM_PARS+4]; |
---|
[9ee2756] | 350 | // TODO: could precompute more magnetism parameters before calling the kernel. |
---|
[885753a] | 351 | double xs_weights[8]; // uu, ud real, du real, dd, ud imag, du imag, fill, fill |
---|
[32e3c9b] | 352 | double cos_mspin, sin_mspin; |
---|
[885753a] | 353 | set_spin_weights(values[NUM_PARS+2], values[NUM_PARS+3], xs_weights); |
---|
[a4280bd] | 354 | SINCOS(-values[NUM_PARS+4]*M_PI_180, sin_mspin, cos_mspin); |
---|
[9eb3632] | 355 | #endif // MAGNETIC |
---|
[3044216] | 356 | |
---|
[6aee3ab] | 357 | // ** Fill in the initial results ** |
---|
[9ee2756] | 358 | // If pd_start is zero that means that we are starting a new calculation, |
---|
| 359 | // and must initialize the result to zero. Otherwise, we are restarting |
---|
| 360 | // the calculation from somewhere in the middle of the dispersity mesh, |
---|
| 361 | // and we update the value rather than reset it. Similarly for the |
---|
| 362 | // normalization factor, which is stored as the final value in the |
---|
| 363 | // results vector (one past the number of q values). |
---|
| 364 | // |
---|
| 365 | // The code differs slightly between opencl and dll since opencl is only |
---|
[07646b6] | 366 | // seeing one q value (stored in the variable "this_F2") while the dll |
---|
[c036ddb] | 367 | // version must loop over all q. |
---|
[07646b6] | 368 | #if defined(CALL_FQ) |
---|
| 369 | double weight_norm = (pd_start == 0 ? 0.0 : result[2*nq]); |
---|
| 370 | double weighted_form = (pd_start == 0 ? 0.0 : result[2*nq+1]); |
---|
| 371 | double weighted_shell = (pd_start == 0 ? 0.0 : result[2*nq+2]); |
---|
| 372 | double weighted_radius = (pd_start == 0 ? 0.0 : result[2*nq+3]); |
---|
| 373 | #else |
---|
| 374 | double weight_norm = (pd_start == 0 ? 0.0 : result[nq]); |
---|
| 375 | double weighted_form = (pd_start == 0 ? 0.0 : result[nq+1]); |
---|
| 376 | double weighted_shell = (pd_start == 0 ? 0.0 : result[nq+2]); |
---|
| 377 | double weighted_radius = (pd_start == 0 ? 0.0 : result[nq+3]); |
---|
| 378 | #endif |
---|
[0db7dbd] | 379 | #if defined(USE_GPU) |
---|
[e44432d] | 380 | #if defined(CALL_FQ) |
---|
[6e7ba14] | 381 | double this_F2 = (pd_start == 0 ? 0.0 : result[2*q_index+0]); |
---|
| 382 | double this_F1 = (pd_start == 0 ? 0.0 : result[2*q_index+1]); |
---|
[c036ddb] | 383 | #else |
---|
[07646b6] | 384 | double this_F2 = (pd_start == 0 ? 0.0 : result[q_index]); |
---|
[c036ddb] | 385 | #endif |
---|
[0db7dbd] | 386 | #else // !USE_GPU |
---|
[9ee2756] | 387 | if (pd_start == 0) { |
---|
| 388 | #ifdef USE_OPENMP |
---|
| 389 | #pragma omp parallel for |
---|
| 390 | #endif |
---|
[e44432d] | 391 | #if defined(CALL_FQ) |
---|
[6e7ba14] | 392 | // 2*nq for F^2,F pairs |
---|
| 393 | for (int q_index=0; q_index < 2*nq; q_index++) result[q_index] = 0.0; |
---|
[c036ddb] | 394 | #else |
---|
| 395 | for (int q_index=0; q_index < nq; q_index++) result[q_index] = 0.0; |
---|
[01c8d9e] | 396 | #endif |
---|
[9ee2756] | 397 | } |
---|
[6aee3ab] | 398 | //if (q_index==0) printf("start %d %g %g\n", pd_start, pd_norm, result[0]); |
---|
[0db7dbd] | 399 | #endif // !USE_GPU |
---|
[9ee2756] | 400 | |
---|
| 401 | |
---|
| 402 | // ====== macros to set up the parts of the loop ======= |
---|
| 403 | /* |
---|
| 404 | Based on the level of the loop, uses C preprocessor magic to construct |
---|
| 405 | level-specific looping variables, including these from loop level 3: |
---|
| 406 | |
---|
| 407 | int n3 : length of loop for mesh level 3 |
---|
| 408 | int i3 : current position in the loop for level 3, which is calculated |
---|
| 409 | from a combination of pd_start, pd_stride[3] and pd_length[3]. |
---|
| 410 | int p3 : is the index into the parameter table for mesh level 3 |
---|
| 411 | double v3[] : pointer into dispersity array to values for loop 3 |
---|
| 412 | double w3[] : pointer into dispersity array to weights for loop 3 |
---|
| 413 | double weight3 : the product of weights from levels 3 and up, computed |
---|
| 414 | as weight5*weight4*w3[i3]. Note that we need an outermost |
---|
| 415 | value weight5 set to 1.0 for this to work properly. |
---|
| 416 | |
---|
| 417 | After expansion, the loop struction will look like the following: |
---|
| 418 | |
---|
| 419 | // --- PD_INIT(4) --- |
---|
| 420 | const int n4 = pd_length[4]; |
---|
| 421 | const int p4 = pd_par[4]; |
---|
[74e9b5f] | 422 | pglobal const double *v4 = pd_value + pd_offset[4]; |
---|
| 423 | pglobal const double *w4 = pd_weight + pd_offset[4]; |
---|
[9ee2756] | 424 | int i4 = (pd_start/pd_stride[4])%n4; // position in level 4 at pd_start |
---|
| 425 | |
---|
| 426 | // --- PD_INIT(3) --- |
---|
| 427 | const int n3 = pd_length[3]; |
---|
| 428 | ... |
---|
| 429 | int i3 = (pd_start/pd_stride[3])%n3; // position in level 3 at pd_start |
---|
| 430 | |
---|
| 431 | PD_INIT(2) |
---|
| 432 | PD_INIT(1) |
---|
| 433 | PD_INIT(0) |
---|
| 434 | |
---|
| 435 | // --- PD_OUTERMOST_WEIGHT(5) --- |
---|
| 436 | const double weight5 = 1.0; |
---|
| 437 | |
---|
| 438 | // --- PD_OPEN(4,5) --- |
---|
| 439 | while (i4 < n4) { |
---|
| 440 | parameter[p4] = v4[i4]; // set the value for pd parameter 4 at this mesh point |
---|
| 441 | const double weight4 = w4[i4] * weight5; |
---|
| 442 | |
---|
| 443 | // from PD_OPEN(3,4) |
---|
| 444 | while (i3 < n3) { |
---|
| 445 | parameter[p3] = v3[i3]; // set the value for pd parameter 3 at this mesh point |
---|
| 446 | const double weight3 = w3[i3] * weight4; |
---|
| 447 | |
---|
| 448 | PD_OPEN(3,2) |
---|
| 449 | PD_OPEN(2,1) |
---|
| 450 | PD_OPEN(0,1) |
---|
| 451 | |
---|
[6aee3ab] | 452 | // ... main loop body ... |
---|
| 453 | APPLY_PROJECTION // convert jitter values to spherical coords |
---|
| 454 | BUILD_ROTATION // construct the rotation matrix qxy => qabc |
---|
| 455 | for each q |
---|
| 456 | FETCH_Q // set qx,qy from the q input vector |
---|
| 457 | APPLY_ROTATION // convert qx,qy to qa,qb,qc |
---|
[07646b6] | 458 | CALL_KERNEL // F2 = Iqxy(qa, qb, qc, p1, p2, ...) |
---|
[9ee2756] | 459 | |
---|
| 460 | ++step; // increment counter representing position in dispersity mesh |
---|
| 461 | |
---|
| 462 | PD_CLOSE(0) |
---|
| 463 | PD_CLOSE(1) |
---|
| 464 | PD_CLOSE(2) |
---|
| 465 | |
---|
| 466 | // --- PD_CLOSE(3) --- |
---|
| 467 | if (step >= pd_stop) break; |
---|
| 468 | ++i3; |
---|
| 469 | } |
---|
| 470 | i3 = 0; // reset loop counter for next round through the loop |
---|
| 471 | |
---|
| 472 | // --- PD_CLOSE(4) --- |
---|
| 473 | if (step >= pd_stop) break; |
---|
| 474 | ++i4; |
---|
[2e44ac7] | 475 | } |
---|
[9ee2756] | 476 | i4 = 0; // reset loop counter even though no more rounds through the loop |
---|
| 477 | |
---|
| 478 | */ |
---|
| 479 | |
---|
| 480 | |
---|
[6aee3ab] | 481 | // ** prepare inner loops ** |
---|
[9ee2756] | 482 | |
---|
| 483 | // Depending on the shape type (radial, axial, triaxial), the variables |
---|
[6aee3ab] | 484 | // and calling parameters in the loop body will be slightly different. |
---|
| 485 | // Macros capture the differences in one spot so the rest of the code |
---|
| 486 | // is easier to read. The code below both declares variables for the |
---|
| 487 | // inner loop and defines the macros that use them. |
---|
[ff10479] | 488 | |
---|
[01c8d9e] | 489 | |
---|
[c036ddb] | 490 | #if defined(CALL_FQ) // COMPUTE_F1_F2 is true |
---|
| 491 | // unoriented 1D returning <F> and <F^2> |
---|
[12f4c19] | 492 | // Note that F1 and F2 are returned from CALL_FQ by reference, and the |
---|
| 493 | // user of the CALL_KERNEL macro below is assuming that F1 and F2 are defined. |
---|
[9ee2756] | 494 | double qk; |
---|
[c036ddb] | 495 | double F1, F2; |
---|
| 496 | #define FETCH_Q() do { qk = q[q_index]; } while (0) |
---|
| 497 | #define BUILD_ROTATION() do {} while(0) |
---|
| 498 | #define APPLY_ROTATION() do {} while(0) |
---|
| 499 | #define CALL_KERNEL() CALL_FQ(qk,F1,F2,local_values.table) |
---|
| 500 | |
---|
| 501 | #elif defined(CALL_FQ_A) |
---|
| 502 | // unoriented 2D return <F> and <F^2> |
---|
[12f4c19] | 503 | // Note that the CALL_FQ_A macro is computing _F1_slot and _F2_slot by |
---|
| 504 | // reference then returning _F2_slot. We are calling them _F1_slot and |
---|
| 505 | // _F2_slot here so they don't conflict with _F1 and _F2 in the macro |
---|
| 506 | // expansion, or with the use of F2 = CALL_KERNEL() when it is used below. |
---|
[c036ddb] | 507 | double qx, qy; |
---|
[12f4c19] | 508 | double _F1_slot, _F2_slot; |
---|
[c036ddb] | 509 | #define FETCH_Q() do { qx = q[2*q_index]; qy = q[2*q_index+1]; } while (0) |
---|
| 510 | #define BUILD_ROTATION() do {} while(0) |
---|
| 511 | #define APPLY_ROTATION() do {} while(0) |
---|
[12f4c19] | 512 | #define CALL_KERNEL() CALL_FQ_A(sqrt(qx*qx+qy*qy),_F1_slot,_F2_slot,local_values.table) |
---|
[c036ddb] | 513 | |
---|
| 514 | #elif defined(CALL_IQ) |
---|
| 515 | // unoriented 1D return <F^2> |
---|
| 516 | double qk; |
---|
| 517 | #define FETCH_Q() do { qk = q[q_index]; } while (0) |
---|
| 518 | #define BUILD_ROTATION() do {} while(0) |
---|
| 519 | #define APPLY_ROTATION() do {} while(0) |
---|
| 520 | #define CALL_KERNEL() CALL_IQ(qk,local_values.table) |
---|
[9ee2756] | 521 | |
---|
| 522 | #elif defined(CALL_IQ_A) |
---|
| 523 | // unoriented 2D |
---|
| 524 | double qx, qy; |
---|
[2c108a3] | 525 | #define FETCH_Q() do { qx = q[2*q_index]; qy = q[2*q_index+1]; } while (0) |
---|
| 526 | #define BUILD_ROTATION() do {} while(0) |
---|
| 527 | #define APPLY_ROTATION() do {} while(0) |
---|
| 528 | #define CALL_KERNEL() CALL_IQ_A(sqrt(qx*qx+qy*qy), local_values.table) |
---|
[9ee2756] | 529 | |
---|
| 530 | #elif defined(CALL_IQ_AC) |
---|
| 531 | // oriented symmetric 2D |
---|
| 532 | double qx, qy; |
---|
[2c108a3] | 533 | #define FETCH_Q() do { qx = q[2*q_index]; qy = q[2*q_index+1]; } while (0) |
---|
[9ee2756] | 534 | double qa, qc; |
---|
| 535 | QACRotation rotation; |
---|
[ff10479] | 536 | // theta, phi, dtheta, dphi are defined below in projection to avoid repeated code. |
---|
[767dca8] | 537 | #define BUILD_ROTATION() qac_rotation(&rotation, theta, phi, dtheta, dphi); |
---|
[ec8d4ac] | 538 | #define APPLY_ROTATION() qac_apply(&rotation, qx, qy, &qa, &qc) |
---|
[2c108a3] | 539 | #define CALL_KERNEL() CALL_IQ_AC(qa, qc, local_values.table) |
---|
[9ee2756] | 540 | |
---|
| 541 | #elif defined(CALL_IQ_ABC) |
---|
| 542 | // oriented asymmetric 2D |
---|
| 543 | double qx, qy; |
---|
[2c108a3] | 544 | #define FETCH_Q() do { qx = q[2*q_index]; qy = q[2*q_index+1]; } while (0) |
---|
[9ee2756] | 545 | double qa, qb, qc; |
---|
| 546 | QABCRotation rotation; |
---|
[ff10479] | 547 | // theta, phi, dtheta, dphi are defined below in projection to avoid repeated code. |
---|
| 548 | // psi and dpsi are only for IQ_ABC, so they are processed here. |
---|
| 549 | const double psi = values[details->theta_par+4]; |
---|
[6aee3ab] | 550 | local_values.table.psi = 0.; |
---|
[ff10479] | 551 | #define BUILD_ROTATION() qabc_rotation(&rotation, theta, phi, psi, dtheta, dphi, local_values.table.psi) |
---|
[ec8d4ac] | 552 | #define APPLY_ROTATION() qabc_apply(&rotation, qx, qy, &qa, &qb, &qc) |
---|
[ff10479] | 553 | #define CALL_KERNEL() CALL_IQ_ABC(qa, qb, qc, local_values.table) |
---|
[c036ddb] | 554 | |
---|
[108e70e] | 555 | #elif defined(CALL_IQ_XY) |
---|
| 556 | // direct call to qx,qy calculator |
---|
| 557 | double qx, qy; |
---|
| 558 | #define FETCH_Q() do { qx = q[2*q_index]; qy = q[2*q_index+1]; } while (0) |
---|
| 559 | #define BUILD_ROTATION() do {} while(0) |
---|
| 560 | #define APPLY_ROTATION() do {} while(0) |
---|
| 561 | #define CALL_KERNEL() CALL_IQ_XY(qx, qy, local_values.table) |
---|
[ff10479] | 562 | #endif |
---|
| 563 | |
---|
[2a7e20e] | 564 | // Define APPLY_PROJECTION depending on model symmetries. We do this outside |
---|
| 565 | // the previous if block so that we don't need to repeat the identical |
---|
| 566 | // logic in the IQ_AC and IQ_ABC branches. This will become more important |
---|
| 567 | // if we implement more projections, or more complicated projections. |
---|
[c036ddb] | 568 | #if defined(CALL_IQ) || defined(CALL_IQ_A) || defined(CALL_FQ) || defined(CALL_FQ_A) |
---|
| 569 | // no orientation |
---|
[ff10479] | 570 | #define APPLY_PROJECTION() const double weight=weight0 |
---|
[2a7e20e] | 571 | #elif defined(CALL_IQ_XY) // pass orientation to the model |
---|
[108e70e] | 572 | // CRUFT: support oriented model which define Iqxy rather than Iqac or Iqabc |
---|
| 573 | // Need to plug the values for the orientation angles back into parameter |
---|
| 574 | // table in case they were overridden by the orientation offset. This |
---|
| 575 | // means that orientation dispersity will not work for these models, but |
---|
| 576 | // it was broken anyway, so no matter. Still want to provide Iqxy in case |
---|
| 577 | // the user model wants full control of orientation/magnetism. |
---|
| 578 | #if defined(HAVE_PSI) |
---|
| 579 | const double theta = values[details->theta_par+2]; |
---|
| 580 | const double phi = values[details->theta_par+3]; |
---|
| 581 | const double psi = values[details->theta_par+4]; |
---|
| 582 | double weight; |
---|
| 583 | #define APPLY_PROJECTION() do { \ |
---|
| 584 | local_values.table.theta = theta; \ |
---|
| 585 | local_values.table.phi = phi; \ |
---|
| 586 | local_values.table.psi = psi; \ |
---|
| 587 | weight=weight0; \ |
---|
| 588 | } while (0) |
---|
| 589 | #elif defined(HAVE_THETA) |
---|
| 590 | const double theta = values[details->theta_par+2]; |
---|
| 591 | const double phi = values[details->theta_par+3]; |
---|
| 592 | double weight; |
---|
| 593 | #define APPLY_PROJECTION() do { \ |
---|
| 594 | local_values.table.theta = theta; \ |
---|
| 595 | local_values.table.phi = phi; \ |
---|
| 596 | weight=weight0; \ |
---|
| 597 | } while (0) |
---|
| 598 | #else |
---|
| 599 | #define APPLY_PROJECTION() const double weight=weight0 |
---|
| 600 | #endif |
---|
[2a7e20e] | 601 | #else // apply jitter and view before calling the model |
---|
[9ee2756] | 602 | // Grab the "view" angles (theta, phi, psi) from the initial parameter table. |
---|
| 603 | const double theta = values[details->theta_par+2]; |
---|
| 604 | const double phi = values[details->theta_par+3]; |
---|
[6aee3ab] | 605 | // Make sure jitter angle defaults to zero if there is no jitter distribution |
---|
| 606 | local_values.table.theta = 0.; |
---|
| 607 | local_values.table.phi = 0.; |
---|
[ff10479] | 608 | // The "jitter" angles (dtheta, dphi, dpsi) are stored with the |
---|
| 609 | // dispersity values and copied to the local parameter table as |
---|
| 610 | // we go through the mesh. |
---|
| 611 | double dtheta, dphi, weight; |
---|
[2a7e20e] | 612 | #if PROJECTION == 1 // equirectangular |
---|
[ff10479] | 613 | #define APPLY_PROJECTION() do { \ |
---|
[767dca8] | 614 | dtheta = local_values.table.theta; \ |
---|
| 615 | dphi = local_values.table.phi; \ |
---|
[ff10479] | 616 | weight = fabs(cos(dtheta*M_PI_180)) * weight0; \ |
---|
[767dca8] | 617 | } while (0) |
---|
[2a7e20e] | 618 | #elif PROJECTION == 2 // sinusoidal |
---|
[ff10479] | 619 | #define APPLY_PROJECTION() do { \ |
---|
[767dca8] | 620 | dtheta = local_values.table.theta; \ |
---|
| 621 | dphi = local_values.table.phi; \ |
---|
[ff10479] | 622 | weight = weight0; \ |
---|
| 623 | if (dtheta != 90.0) dphi /= cos(dtheta*M_PI_180); \ |
---|
| 624 | else if (dphi != 0.0) weight = 0.; \ |
---|
| 625 | if (fabs(dphi) >= 180.) weight = 0.; \ |
---|
[767dca8] | 626 | } while (0) |
---|
| 627 | #endif |
---|
[2a7e20e] | 628 | #endif // done defining APPLY_PROJECTION |
---|
[9ee2756] | 629 | |
---|
[6aee3ab] | 630 | // ** define looping macros ** |
---|
| 631 | |
---|
| 632 | // Define looping variables |
---|
| 633 | #define PD_INIT(_LOOP) \ |
---|
| 634 | const int n##_LOOP = details->pd_length[_LOOP]; \ |
---|
| 635 | const int p##_LOOP = details->pd_par[_LOOP]; \ |
---|
[74e9b5f] | 636 | pglobal const double *v##_LOOP = pd_value + details->pd_offset[_LOOP]; \ |
---|
| 637 | pglobal const double *w##_LOOP = pd_weight + details->pd_offset[_LOOP]; \ |
---|
[6aee3ab] | 638 | int i##_LOOP = (pd_start/details->pd_stride[_LOOP])%n##_LOOP; |
---|
| 639 | |
---|
| 640 | // Jump into the middle of the dispersity loop |
---|
| 641 | #define PD_OPEN(_LOOP,_OUTER) \ |
---|
| 642 | while (i##_LOOP < n##_LOOP) { \ |
---|
| 643 | local_values.vector[p##_LOOP] = v##_LOOP[i##_LOOP]; \ |
---|
| 644 | const double weight##_LOOP = w##_LOOP[i##_LOOP] * weight##_OUTER; |
---|
| 645 | |
---|
| 646 | // create the variable "weight#=1.0" where # is the outermost level+1 (=MAX_PD). |
---|
| 647 | #define _PD_OUTERMOST_WEIGHT(_n) const double weight##_n = 1.0; |
---|
| 648 | #define PD_OUTERMOST_WEIGHT(_n) _PD_OUTERMOST_WEIGHT(_n) |
---|
| 649 | |
---|
| 650 | // Close out the loop |
---|
| 651 | #define PD_CLOSE(_LOOP) \ |
---|
| 652 | if (step >= pd_stop) break; \ |
---|
| 653 | ++i##_LOOP; \ |
---|
| 654 | } \ |
---|
| 655 | i##_LOOP = 0; |
---|
[9eb3632] | 656 | |
---|
[9ee2756] | 657 | // ====== construct the loops ======= |
---|
| 658 | |
---|
| 659 | // Pointers to the start of the dispersity and weight vectors, if needed. |
---|
[7b7da6b] | 660 | #if MAX_PD>0 |
---|
[74e9b5f] | 661 | pglobal const double *pd_value = values + NUM_VALUES; |
---|
| 662 | pglobal const double *pd_weight = pd_value + details->num_weights; |
---|
[7b7da6b] | 663 | #endif |
---|
[9eb3632] | 664 | |
---|
[9ee2756] | 665 | // The variable "step" is the current position in the dispersity loop. |
---|
| 666 | // It will be incremented each time a new point in the mesh is accumulated, |
---|
| 667 | // and used to test whether we have reached pd_stop. |
---|
| 668 | int step = pd_start; |
---|
| 669 | |
---|
[6aee3ab] | 670 | // *** define loops for each of 0, 1, 2, ..., modelinfo.MAX_PD-1 *** |
---|
| 671 | |
---|
[9ee2756] | 672 | // define looping variables |
---|
[9eb3632] | 673 | #if MAX_PD>4 |
---|
[9ee2756] | 674 | PD_INIT(4) |
---|
[9eb3632] | 675 | #endif |
---|
| 676 | #if MAX_PD>3 |
---|
[9ee2756] | 677 | PD_INIT(3) |
---|
[9eb3632] | 678 | #endif |
---|
| 679 | #if MAX_PD>2 |
---|
[9ee2756] | 680 | PD_INIT(2) |
---|
[9eb3632] | 681 | #endif |
---|
| 682 | #if MAX_PD>1 |
---|
[9ee2756] | 683 | PD_INIT(1) |
---|
[9eb3632] | 684 | #endif |
---|
| 685 | #if MAX_PD>0 |
---|
[9ee2756] | 686 | PD_INIT(0) |
---|
[9eb3632] | 687 | #endif |
---|
[2e44ac7] | 688 | |
---|
[9ee2756] | 689 | // open nested loops |
---|
| 690 | PD_OUTERMOST_WEIGHT(MAX_PD) |
---|
[9eb3632] | 691 | #if MAX_PD>4 |
---|
[9ee2756] | 692 | PD_OPEN(4,5) |
---|
[9eb3632] | 693 | #endif |
---|
| 694 | #if MAX_PD>3 |
---|
[9ee2756] | 695 | PD_OPEN(3,4) |
---|
[9eb3632] | 696 | #endif |
---|
| 697 | #if MAX_PD>2 |
---|
[9ee2756] | 698 | PD_OPEN(2,3) |
---|
[9eb3632] | 699 | #endif |
---|
| 700 | #if MAX_PD>1 |
---|
[9ee2756] | 701 | PD_OPEN(1,2) |
---|
[9eb3632] | 702 | #endif |
---|
| 703 | #if MAX_PD>0 |
---|
[9ee2756] | 704 | PD_OPEN(0,1) |
---|
[9eb3632] | 705 | #endif |
---|
[5ff1b03] | 706 | |
---|
[9ee2756] | 707 | //if (q_index==0) {printf("step:%d of %d, pars:",step,pd_stop); for (int i=0; i < NUM_PARS; i++) printf("p%d=%g ",i, local_values.vector[i]); printf("\n");} |
---|
| 708 | |
---|
| 709 | // ====== loop body ======= |
---|
| 710 | #ifdef INVALID |
---|
| 711 | if (!INVALID(local_values.table)) |
---|
| 712 | #endif |
---|
| 713 | { |
---|
[ff10479] | 714 | APPLY_PROJECTION(); |
---|
| 715 | |
---|
[9ee2756] | 716 | // Accumulate I(q) |
---|
| 717 | // Note: weight==0 must always be excluded |
---|
[767dca8] | 718 | if (weight > cutoff) { |
---|
[e44432d] | 719 | double form, shell; |
---|
| 720 | CALL_VOLUME(form, shell, local_values.table); |
---|
[c036ddb] | 721 | weight_norm += weight; |
---|
[e44432d] | 722 | weighted_form += weight * form; |
---|
| 723 | weighted_shell += weight * shell; |
---|
[6e7ba14] | 724 | if (effective_radius_type != 0) { |
---|
| 725 | weighted_radius += weight * CALL_EFFECTIVE_RADIUS(effective_radius_type, local_values.table); |
---|
| 726 | } |
---|
[2c108a3] | 727 | BUILD_ROTATION(); |
---|
[c036ddb] | 728 | |
---|
[0db7dbd] | 729 | #if !defined(USE_GPU) |
---|
[9ee2756] | 730 | // DLL needs to explicitly loop over the q values. |
---|
| 731 | #ifdef USE_OPENMP |
---|
| 732 | #pragma omp parallel for |
---|
| 733 | #endif |
---|
| 734 | for (q_index=0; q_index<nq; q_index++) |
---|
[0db7dbd] | 735 | #endif // !USE_GPU |
---|
[9ee2756] | 736 | { |
---|
| 737 | |
---|
[2c108a3] | 738 | FETCH_Q(); |
---|
| 739 | APPLY_ROTATION(); |
---|
[9ee2756] | 740 | |
---|
| 741 | // ======= COMPUTE SCATTERING ========== |
---|
| 742 | #if defined(MAGNETIC) && NUM_MAGNETIC > 0 |
---|
[2c108a3] | 743 | // Compute the scattering from the magnetic cross sections. |
---|
[07646b6] | 744 | double F2 = 0.0; |
---|
[2c108a3] | 745 | const double qsq = qx*qx + qy*qy; |
---|
| 746 | if (qsq > 1.e-16) { |
---|
| 747 | // TODO: what is the magnetic scattering at q=0 |
---|
| 748 | const double px = (qy*cos_mspin + qx*sin_mspin)/qsq; |
---|
| 749 | const double py = (qy*sin_mspin - qx*cos_mspin)/qsq; |
---|
| 750 | |
---|
| 751 | // loop over uu, ud real, du real, dd, ud imag, du imag |
---|
[aadec17] | 752 | for (unsigned int xs=0; xs<6; xs++) { |
---|
[885753a] | 753 | const double xs_weight = xs_weights[xs]; |
---|
[2c108a3] | 754 | if (xs_weight > 1.e-8) { |
---|
| 755 | // Since the cross section weight is significant, set the slds |
---|
| 756 | // to the effective slds for this cross section, call the |
---|
| 757 | // kernel, and add according to weight. |
---|
| 758 | for (int sk=0; sk<NUM_MAGNETIC; sk++) { |
---|
| 759 | const int32_t mag_index = NUM_PARS+5 + 3*sk; |
---|
| 760 | const int32_t sld_index = slds[sk]; |
---|
| 761 | const double mx = values[mag_index]; |
---|
| 762 | const double my = values[mag_index+1]; |
---|
| 763 | const double mz = values[mag_index+2]; |
---|
| 764 | local_values.vector[sld_index] = |
---|
| 765 | mag_sld(xs, qx, qy, px, py, values[sld_index+2], mx, my, mz); |
---|
[885753a] | 766 | //if (q_index==0) printf("%d: (qx,qy)=(%g,%g) xs=%d sld%d=%g p=(%g,%g) m=(%g,%g,%g)\n", |
---|
[dc6f601] | 767 | // q_index, qx, qy, xs, sk, local_values.vector[sld_index], px, py, mx, my, mz); |
---|
[2c108a3] | 768 | } |
---|
[07646b6] | 769 | F2 += xs_weight * CALL_KERNEL(); |
---|
[a4280bd] | 770 | } |
---|
[9eb3632] | 771 | } |
---|
[32e3c9b] | 772 | } |
---|
[9ee2756] | 773 | #else // !MAGNETIC |
---|
[e44432d] | 774 | #if defined(CALL_FQ) |
---|
[c036ddb] | 775 | CALL_KERNEL(); // sets F1 and F2 by reference |
---|
[01c8d9e] | 776 | #else |
---|
[07646b6] | 777 | const double F2 = CALL_KERNEL(); |
---|
[01c8d9e] | 778 | #endif |
---|
[9ee2756] | 779 | #endif // !MAGNETIC |
---|
[07646b6] | 780 | //printf("q_index:%d %g %g %g %g\n", q_index, F2, weight0); |
---|
[9ee2756] | 781 | |
---|
[0db7dbd] | 782 | #if defined(USE_GPU) |
---|
[e44432d] | 783 | #if defined(CALL_FQ) |
---|
[c036ddb] | 784 | this_F2 += weight * F2; |
---|
[6e7ba14] | 785 | this_F1 += weight * F1; |
---|
[c036ddb] | 786 | #else |
---|
[07646b6] | 787 | this_F2 += weight * F2; |
---|
[01c8d9e] | 788 | #endif |
---|
[9ee2756] | 789 | #else // !USE_OPENCL |
---|
[e44432d] | 790 | #if defined(CALL_FQ) |
---|
[6e7ba14] | 791 | result[2*q_index+0] += weight * F2; |
---|
| 792 | result[2*q_index+1] += weight * F1; |
---|
[c036ddb] | 793 | #else |
---|
[07646b6] | 794 | result[q_index] += weight * F2; |
---|
[01c8d9e] | 795 | #endif |
---|
[9ee2756] | 796 | #endif // !USE_OPENCL |
---|
[3044216] | 797 | } |
---|
[03cac08] | 798 | } |
---|
[2e44ac7] | 799 | } |
---|
[9ee2756] | 800 | // close nested loops |
---|
| 801 | ++step; |
---|
| 802 | #if MAX_PD>0 |
---|
| 803 | PD_CLOSE(0) |
---|
[9eb3632] | 804 | #endif |
---|
| 805 | #if MAX_PD>1 |
---|
[9ee2756] | 806 | PD_CLOSE(1) |
---|
[9eb3632] | 807 | #endif |
---|
| 808 | #if MAX_PD>2 |
---|
[9ee2756] | 809 | PD_CLOSE(2) |
---|
[9eb3632] | 810 | #endif |
---|
| 811 | #if MAX_PD>3 |
---|
[9ee2756] | 812 | PD_CLOSE(3) |
---|
[9eb3632] | 813 | #endif |
---|
| 814 | #if MAX_PD>4 |
---|
[9ee2756] | 815 | PD_CLOSE(4) |
---|
[9eb3632] | 816 | #endif |
---|
[f2f67a6] | 817 | |
---|
[07646b6] | 818 | // Remember the results and the updated norm. |
---|
[0db7dbd] | 819 | #if defined(USE_GPU) |
---|
[e44432d] | 820 | #if defined(CALL_FQ) |
---|
[07646b6] | 821 | result[2*q_index+0] = this_F2; |
---|
| 822 | result[2*q_index+1] = this_F1; |
---|
[c036ddb] | 823 | #else |
---|
[07646b6] | 824 | result[q_index] = this_F2; |
---|
[01c8d9e] | 825 | #endif |
---|
[07646b6] | 826 | if (q_index == 0) |
---|
| 827 | #endif |
---|
| 828 | { |
---|
| 829 | #if defined(CALL_FQ) |
---|
[6e7ba14] | 830 | result[2*nq] = weight_norm; |
---|
[e44432d] | 831 | result[2*nq+1] = weighted_form; |
---|
| 832 | result[2*nq+2] = weighted_shell; |
---|
| 833 | result[2*nq+3] = weighted_radius; |
---|
[07646b6] | 834 | #else |
---|
[6e7ba14] | 835 | result[nq] = weight_norm; |
---|
[e44432d] | 836 | result[nq+1] = weighted_form; |
---|
| 837 | result[nq+2] = weighted_shell; |
---|
| 838 | result[nq+3] = weighted_radius; |
---|
[07646b6] | 839 | #endif |
---|
| 840 | } |
---|
[2c108a3] | 841 | |
---|
[6aee3ab] | 842 | // ** clear the macros in preparation for the next kernel ** |
---|
[2c108a3] | 843 | #undef PD_INIT |
---|
| 844 | #undef PD_OPEN |
---|
| 845 | #undef PD_CLOSE |
---|
| 846 | #undef FETCH_Q |
---|
[ff10479] | 847 | #undef APPLY_PROJECTION |
---|
[2c108a3] | 848 | #undef BUILD_ROTATION |
---|
| 849 | #undef APPLY_ROTATION |
---|
| 850 | #undef CALL_KERNEL |
---|
[2e44ac7] | 851 | } |
---|