Changes in / [cdd676e:502c7b8] in sasmodels
- Files:
-
- 1 added
- 12 edited
Legend:
- Unmodified
- Added
- Removed
-
explore/beta/sasfit_compare.py
rcdd676e r01c8d9e 3 3 import sys, os 4 4 BETA_DIR = os.path.dirname(os.path.realpath(__file__)) 5 SASMODELS_DIR = os.path.dirname(os.path.dirname(BETA_DIR)) 5 #SASMODELS_DIR = os.path.dirname(os.path.dirname(BETA_DIR)) 6 SASMODELS_DIR = r"C:\Source\sasmodels" 6 7 sys.path.insert(0, SASMODELS_DIR) 7 8 … … 61 62 calculator = DirectModel(data, model,cutoff=0) 62 63 calculator.pars = pars.copy() 63 calculator.pars.setdefault('background', 1e-3)64 calculator.pars.setdefault('background', 0) 64 65 return calculator 65 66 … … 231 232 # = F2/total_volume 232 233 IQD = F2/average_volume*1e-4*volfraction 234 F1 *= 1e-2 # Yun is using sld in 1/A^2, not 1e-6/A^2 235 F2 *= 1e-4 233 236 elif norm == 'yun': 234 237 F1 *= 1e-6 # Yun is using sld in 1/A^2, not 1e-6/A^2 … … 336 339 I = build_model(Pname+"@hardsphere", q) 337 340 Pq = P(**Ppars)*pars.get('volfraction', 1) 338 #Sq = S(**Spars)341 Sq = S(**Spars) 339 342 Iq = I(**Ipars) 340 343 #Iq = Pq*Sq*pars.get('volfraction', 1) 341 Sq = Iq/Pq 342 return Theory(Q=q, F1=None, F2=None, P=Pq, S=Sq, I=Iq, Seff=None, Ibeta=None) 344 #Sq = Iq/Pq 345 #Iq = None#= Sq = None 346 r=I._kernel.results 347 return Theory(Q=q, F1=None, F2=None, P=Pq, S=None, I=None, Seff=r[1], Ibeta=Iq) 343 348 344 349 def compare(title, target, actual, fields='F1 F2 P S I Seff Ibeta'): … … 400 405 radius_equatorial_pd=.1, radius_equatorial_pd_type=pd_type, 401 406 volfraction=0.15, 407 radius_effective=270.7543927018, 402 408 #radius_effective=12.59921049894873, 403 409 ) 404 target = sasmodels_theory(q, model, **pars)410 target = sasmodels_theory(q, model, beta_mode=1, **pars) 405 411 actual = ellipsoid_pe(q, norm='sasview', **pars) 406 412 title = " ".join(("sasmodels", model, pd_type)) … … 450 456 S = data[5] 451 457 Seff = data[6] 452 target = Theory(Q=Q, F1=F1, F2=F2,P=P, S=S, Seff=Seff)458 target = Theory(Q=Q, F1=F1, P=P, S=S, Seff=Seff) 453 459 actual = sphere_r(Q, norm='yun', **pars) 454 460 title = " ".join(("yun", "sphere", "10% dispersion 10% Vf")) … … 462 468 S = data[5] 463 469 Seff = data[6] 464 target = Theory(Q=Q, F1=F1, F2=F2,P=P, S=S, Seff=Seff)470 target = Theory(Q=Q, F1=F1, P=P, S=S, Seff=Seff) 465 471 actual = sphere_r(Q, norm='yun', **pars) 466 472 title = " ".join(("yun", "sphere", "15% dispersion 10% Vf")) -
sasmodels/core.py
r3221de0 r01c8d9e 140 140 used with functions in generate to build the docs or extract model info. 141 141 """ 142 142 143 if "+" in model_string: 143 144 parts = [load_model_info(part) … … 205 206 206 207 numpy_dtype, fast, platform = parse_dtype(model_info, dtype, platform) 207 208 208 source = generate.make_source(model_info) 209 209 if platform == "dll": … … 265 265 # Assign default platform, overriding ocl with dll if OpenCL is unavailable 266 266 # If opencl=False OpenCL is switched off 267 268 267 if platform is None: 269 268 platform = "ocl" … … 291 290 else: 292 291 numpy_dtype = np.dtype(dtype) 293 294 292 # Make sure that the type is supported by opencl, otherwise use dll 295 293 if platform == "ocl": -
sasmodels/data.py
r65fbf7c r01c8d9e 329 329 else: 330 330 dq = resolution * q 331 332 331 data = Data1D(q, Iq, dx=dq, dy=dIq) 333 332 data.filename = "fake data" … … 523 522 else 'log') 524 523 plt.xscale(xscale) 524 525 525 plt.xlabel("$q$/A$^{-1}$") 526 526 plt.yscale(yscale) -
sasmodels/generate.py
rd86f0fc r01c8d9e 672 672 673 673 # type in IQXY pattern could be single, float, double, long double, ... 674 _IQXY_PATTERN = re.compile(r"(^|\s)double\s+I(?P<mode>q(a b?c|xy))\s*[(]",674 _IQXY_PATTERN = re.compile(r"(^|\s)double\s+I(?P<mode>q(ac|abc|xy))\s*[(]", 675 675 flags=re.MULTILINE) 676 676 def find_xy_mode(source): … … 701 701 return 'qa' 702 702 703 # type in IQXY pattern could be single, float, double, long double, ... 704 _FQ_PATTERN = re.compile(r"(^|\s)void\s+Fq[(]", flags=re.MULTILINE) 705 def has_Fq(source): 706 for code in source: 707 m = _FQ_PATTERN.search(code) 708 if m is not None: 709 return True 710 return False 703 711 704 712 def _add_source(source, code, path, lineno=1): … … 730 738 # dispersion. Need to be careful that necessary parameters are available 731 739 # for computing volume even if we allow non-disperse volume parameters. 732 733 740 partable = model_info.parameters 734 735 741 # Load templates and user code 736 742 kernel_header = load_template('kernel_header.c') 737 743 kernel_code = load_template('kernel_iq.c') 738 744 user_code = [(f, open(f).read()) for f in model_sources(model_info)] 739 740 745 # Build initial sources 741 746 source = [] … … 743 748 for path, code in user_code: 744 749 _add_source(source, code, path) 745 746 750 if model_info.c_code: 747 751 _add_source(source, model_info.c_code, model_info.filename, … … 789 793 source.append("\\\n".join(p.as_definition() 790 794 for p in partable.kernel_parameters)) 791 792 795 # Define the function calls 793 796 if partable.form_volume_parameters: … … 800 803 call_volume = "#define CALL_VOLUME(v) 1.0" 801 804 source.append(call_volume) 802 803 805 model_refs = _call_pars("_v.", partable.iq_parameters) 804 pars = ",".join(["_q"] + model_refs) 805 call_iq = "#define CALL_IQ(_q, _v) Iq(%s)" % pars 806 #create varaible BETA to turn on and off beta approximation 807 BETA = has_Fq(source) 808 if not BETA: 809 pars = ",".join(["_q"] + model_refs) 810 call_iq = "#define CALL_IQ(_q, _v) Iq(%s)" % pars 811 else: 812 pars = ",".join(["_q", "&_F1", "&_F2",] + model_refs) 813 call_iq = "#define CALL_IQ(_q, _F1, _F2, _v) Fq(%s)" % pars 806 814 if xy_mode == 'qabc': 807 815 pars = ",".join(["_qa", "_qb", "_qc"] + model_refs) … … 831 839 magpars = [k-2 for k, p in enumerate(partable.call_parameters) 832 840 if p.type == 'sld'] 833 834 841 # Fill in definitions for numbers of parameters 842 source.append("#define BETA %d" %(1 if BETA else 0)) 835 843 source.append("#define MAX_PD %s"%partable.max_pd) 836 844 source.append("#define NUM_PARS %d"%partable.npars) … … 839 847 source.append("#define MAGNETIC_PARS %s"%",".join(str(k) for k in magpars)) 840 848 source.append("#define PROJECTION %d"%PROJECTION) 841 842 849 # TODO: allow mixed python/opencl kernels? 843 844 850 ocl = _kernels(kernel_code, call_iq, call_iqxy, clear_iqxy, model_info.name) 845 851 dll = _kernels(kernel_code, call_iq, call_iqxy, clear_iqxy, model_info.name) 852 846 853 result = { 847 854 'dll': '\n'.join(source+dll[0]+dll[1]+dll[2]), 848 855 'opencl': '\n'.join(source+ocl[0]+ocl[1]+ocl[2]), 849 856 } 850 857 #print(result['dll']) 851 858 return result 852 859 … … 1068 1075 model_info = make_model_info(kernel_module) 1069 1076 source = make_source(model_info) 1070 print(source['dll'])1077 #print(source['dll']) 1071 1078 1072 1079 -
sasmodels/kernel_iq.c
r7c35fda r01c8d9e 36 36 // PROJECTION : equirectangular=1, sinusoidal=2 37 37 // see explore/jitter.py for definitions. 38 38 39 39 40 #ifndef _PAR_BLOCK_ // protected block so we can include this code twice. … … 270 271 #endif // _QABC_SECTION 271 272 272 273 273 // ==================== KERNEL CODE ======================== 274 275 274 kernel 276 275 void KERNEL_NAME( … … 339 338 // The code differs slightly between opencl and dll since opencl is only 340 339 // seeing one q value (stored in the variable "this_result") while the dll 341 // version must loop over all q. 340 // version must loop over all q 341 #if BETA 342 double *beta_result = &(result[nq+1]); // = result + nq + 1 343 double weight_norm = (pd_start == 0 ? 0.0 : result[2*nq+1]); 344 #endif 342 345 #ifdef USE_OPENCL 343 346 double pd_norm = (pd_start == 0 ? 0.0 : result[nq]); 344 347 double this_result = (pd_start == 0 ? 0.0 : result[q_index]); 348 #if BETA 349 double this_beta_result = (pd_start == 0 ? 0.0 : result[nq + q_index]; 345 350 #else // !USE_OPENCL 346 double pd_norm = (pd_start == 0 ? 0.0 : result[nq]); 351 double pd_norm = (pd_start == 0 ? 0.0 : result[nq]); 347 352 if (pd_start == 0) { 348 353 #ifdef USE_OPENMP … … 350 355 #endif 351 356 for (int q_index=0; q_index < nq; q_index++) result[q_index] = 0.0; 357 #if BETA 358 for (int q_index=0; q_index < nq; q_index++) beta_result[q_index] = 0.0; 359 #endif 352 360 } 353 361 //if (q_index==0) printf("start %d %g %g\n", pd_start, pd_norm, result[0]); 354 362 #endif // !USE_OPENCL 363 364 365 355 366 356 367 … … 442 453 // inner loop and defines the macros that use them. 443 454 455 444 456 #if defined(CALL_IQ) 445 457 // unoriented 1D 446 458 double qk; 447 #define FETCH_Q() do { qk = q[q_index]; } while (0) 448 #define BUILD_ROTATION() do {} while(0) 449 #define APPLY_ROTATION() do {} while(0) 450 #define CALL_KERNEL() CALL_IQ(qk, local_values.table) 459 #if BETA == 0 460 #define FETCH_Q() do { qk = q[q_index]; } while (0) 461 #define BUILD_ROTATION() do {} while(0) 462 #define APPLY_ROTATION() do {} while(0) 463 #define CALL_KERNEL() CALL_IQ(qk,local_values.table) 464 465 // unoriented 1D Beta 466 #elif BETA == 1 467 double F1, F2; 468 #define FETCH_Q() do { qk = q[q_index]; } while (0) 469 #define BUILD_ROTATION() do {} while(0) 470 #define APPLY_ROTATION() do {} while(0) 471 #define CALL_KERNEL() CALL_IQ(qk,F1,F2,local_values.table) 472 #endif 451 473 452 474 #elif defined(CALL_IQ_A) … … 647 669 pd_norm += weight * CALL_VOLUME(local_values.table); 648 670 BUILD_ROTATION(); 649 671 #if BETA 672 if (weight > cutoff) { 673 weight_norm += weight;} 674 #endif 650 675 #ifndef USE_OPENCL 651 676 // DLL needs to explicitly loop over the q values. … … 693 718 } 694 719 #else // !MAGNETIC 695 const double scattering = CALL_KERNEL(); 720 #if defined(CALL_IQ) && BETA == 1 721 CALL_KERNEL(); 722 const double scatteringF1 = F1; 723 const double scatteringF2 = F2; 724 #else 725 const double scattering = CALL_KERNEL(); 726 #endif 696 727 #endif // !MAGNETIC 697 728 //printf("q_index:%d %g %g %g %g\n", q_index, scattering, weight0); 698 729 699 730 #ifdef USE_OPENCL 700 this_result += weight * scattering; 731 #if defined(CALL_IQ)&& BETA == 1 732 this_result += weight * scatteringF2; 733 this_beta_result += weight * scatteringF1; 734 #else 735 this_result += weight * scattering; 736 #endif 701 737 #else // !USE_OPENCL 702 result[q_index] += weight * scattering; 738 #if defined(CALL_IQ)&& BETA == 1 739 result[q_index] += weight * scatteringF2; 740 beta_result[q_index] += weight * scatteringF1; 741 #endif 742 #else 743 result[q_index] += weight * scattering; 744 #endif 703 745 #endif // !USE_OPENCL 704 746 } 705 747 } 706 748 } 707 708 749 // close nested loops 709 750 ++step; … … 728 769 result[q_index] = this_result; 729 770 if (q_index == 0) result[nq] = pd_norm; 771 #if BETA 772 beta_result[q_index] = this_beta_result; 773 #endif 774 if (q_index == 0) result[nq] = pd_norm; 775 730 776 //if (q_index == 0) printf("res: %g/%g\n", result[0], pd_norm); 731 777 #else // !USE_OPENCL 732 778 result[nq] = pd_norm; 779 #if BETA 780 result[2*nq+1] = weight_norm; 781 #endif 733 782 //printf("res: %g/%g\n", result[0], pd_norm); 734 783 #endif // !USE_OPENCL -
sasmodels/kernelcl.py
rd86f0fc r01c8d9e 420 420 if self.program is None: 421 421 compile_program = environment().compile_program 422 with open('model.c','w') as fid: 423 print(self.source['opencl'], file=fid) 422 424 timestamp = generate.ocl_timestamp(self.info) 423 425 self.program = compile_program( … … 539 541 self.dim = '2d' if q_input.is_2d else '1d' 540 542 # plus three for the normalization values 541 self.result = np.empty( q_input.nq+1,dtype)543 self.result = np.empty(2*q_input.nq+2,dtype) 542 544 543 545 # Inputs and outputs for each kernel call … … 555 557 else np.float16 if dtype == generate.F16 556 558 else np.float32) # will never get here, so use np.float32 557 558 def __call__(self, call_details, values, cutoff, magnetic): 559 __call__= Iq 560 561 def Iq(self, call_details, values, cutoff, magnetic): 559 562 # type: (CallDetails, np.ndarray, np.ndarray, float, bool) -> np.ndarray 560 563 context = self.queue.context … … 572 575 ] 573 576 #print("Calling OpenCL") 574 #call_details.show(values)575 # 577 call_details.show(values) 578 #Call kernel and retrieve results 576 579 wait_for = None 577 580 last_nap = time.clock() … … 604 607 return scale*self.result[:self.q_input.nq] + background 605 608 # return self.result[:self.q_input.nq] 609 #NEEDS TO BE FINISHED FOR OPENCL 610 def beta(): 611 return 0 606 612 607 613 def release(self): -
sasmodels/kerneldll.py
r33969b6 r01c8d9e 258 258 # Note: if there is a syntax error then compile raises an error 259 259 # and the source file will not be deleted. 260 os.unlink(filename)261 #print("saving compiled file in %r"%filename)260 #os.unlink(filename) 261 print("saving compiled file in %r"%filename) 262 262 return dll 263 263 … … 371 371 def __init__(self, kernel, model_info, q_input): 372 372 # type: (Callable[[], np.ndarray], ModelInfo, PyInput) -> None 373 #,model_info,q_input) 373 374 self.kernel = kernel 374 375 self.info = model_info … … 376 377 self.dtype = q_input.dtype 377 378 self.dim = '2d' if q_input.is_2d else '1d' 378 self.result = np.empty( q_input.nq+1, q_input.dtype)379 self.result = np.empty(2*q_input.nq+2, q_input.dtype) 379 380 self.real = (np.float32 if self.q_input.dtype == generate.F32 380 381 else np.float64 if self.q_input.dtype == generate.F64 381 382 else np.float128) 382 383 383 def __call__(self, call_details, values, cutoff, magnetic):384 def Iq(self, call_details, values, cutoff, magnetic): 384 385 # type: (CallDetails, np.ndarray, np.ndarray, float, bool) -> np.ndarray 385 386 self._call_kernel(call_details, values, cutoff, magnetic) 387 #print("returned",self.q_input.q, self.result) 388 pd_norm = self.result[self.q_input.nq] 389 scale = values[0]/(pd_norm if pd_norm != 0.0 else 1.0) 390 background = values[1] 391 #print("scale",scale,background) 392 return scale*self.result[:self.q_input.nq] + background 393 __call__ = Iq 394 395 def beta(self, call_details, values, cutoff, magnetic): 396 # type: (CallDetails, np.ndarray, np.ndarray, float, bool) -> np.ndarray 397 self._call_kernel(call_details, values, cutoff, magnetic) 398 w_norm = self.result[2*self.q_input.nq + 1] 399 pd_norm = self.result[self.q_input.nq] 400 if w_norm == 0.: 401 w_norm = 1. 402 F2 = self.result[:self.q_input.nq]/w_norm 403 F1 = self.result[self.q_input.nq+1:2*self.q_input.nq+1]/w_norm 404 volume_avg = pd_norm/w_norm 405 return F1, F2, volume_avg 406 407 def _call_kernel(self, call_details, values, cutoff, magnetic): 386 408 kernel = self.kernel[1 if magnetic else 0] 387 409 args = [ … … 390 412 None, # pd_stop pd_stride[MAX_PD] 391 413 call_details.buffer.ctypes.data, # problem 392 values.ctypes.data, # pars393 self.q_input.q.ctypes.data, # q414 values.ctypes.data, # pars 415 self.q_input.q.ctypes.data, # q 394 416 self.result.ctypes.data, # results 395 417 self.real(cutoff), # cutoff 396 418 ] 397 #print("Calling DLL") 419 #print(self.beta_result.ctypes.data) 420 # print("Calling DLL line 397") 421 # print("values", values) 398 422 #call_details.show(values) 399 423 step = 100 … … 403 427 kernel(*args) # type: ignore 404 428 405 #print("returned",self.q_input.q, self.result)406 pd_norm = self.result[self.q_input.nq]407 scale = values[0]/(pd_norm if pd_norm != 0.0 else 1.0)408 background = values[1]409 #print("scale",scale,background)410 return scale*self.result[:self.q_input.nq] + background411 412 429 def release(self): 413 430 # type: () -> None -
sasmodels/modelinfo.py
r95498a3 r01c8d9e 163 163 parameter.length = length 164 164 parameter.length_control = control 165 166 165 return parameter 167 166 … … 418 417 # scale and background are implicit parameters 419 418 COMMON = [Parameter(*p) for p in COMMON_PARAMETERS] 420 421 419 def __init__(self, parameters): 422 420 # type: (List[Parameter]) -> None 423 421 self.kernel_parameters = parameters 424 422 self._set_vector_lengths() 425 426 423 self.npars = sum(p.length for p in self.kernel_parameters) 427 424 self.nmagnetic = sum(p.length for p in self.kernel_parameters … … 430 427 if self.nmagnetic: 431 428 self.nvalues += 3 + 3*self.nmagnetic 432 433 429 self.call_parameters = self._get_call_parameters() 434 430 self.defaults = self._get_defaults() … … 444 440 self.form_volume_parameters = [p for p in self.kernel_parameters 445 441 if p.type == 'volume'] 446 447 442 # Theta offset 448 443 offset = 0 … … 466 461 self.magnetism_index = [k for k, p in enumerate(self.call_parameters) 467 462 if p.id.startswith('M0:')] 468 469 463 self.pd_1d = set(p.name for p in self.call_parameters 470 464 if p.polydisperse and p.type not in ('orientation', 'magnetic')) … … 770 764 # Custom sum/multi models 771 765 return kernel_module.model_info 766 772 767 info = ModelInfo() 773 768 #print("make parameter table", kernel_module.parameters) … … 822 817 info.lineno = {} 823 818 _find_source_lines(info, kernel_module) 824 825 819 return info 826 820 -
sasmodels/models/ellipsoid.c
r108e70e r01c8d9e 20 20 // i(h) = int_0^1 Phi^2(h a sqrt(1 + u^2(v^2-1)) du 21 21 const double v_square_minus_one = square(radius_polar/radius_equatorial) - 1.0; 22 22 23 23 // translate a point in [-1,1] to a point in [0, 1] 24 24 // const double u = GAUSS_Z[i]*(upper-lower)/2 + (upper+lower)/2; … … 36 36 const double s = (sld - sld_solvent) * form_volume(radius_polar, radius_equatorial); 37 37 return 1.0e-4 * s * s * form; 38 } 39 40 static void 41 Fq(double q, 42 double *F1, 43 double *F2, 44 double sld, 45 double sld_solvent, 46 double radius_polar, 47 double radius_equatorial) 48 { 49 // Using ratio v = Rp/Re, we can implement the form given in Guinier (1955) 50 // i(h) = int_0^pi/2 Phi^2(h a sqrt(cos^2 + v^2 sin^2) cos dT 51 // = int_0^pi/2 Phi^2(h a sqrt((1-sin^2) + v^2 sin^2) cos dT 52 // = int_0^pi/2 Phi^2(h a sqrt(1 + sin^2(v^2-1)) cos dT 53 // u-substitution of 54 // u = sin, du = cos dT 55 // i(h) = int_0^1 Phi^2(h a sqrt(1 + u^2(v^2-1)) du 56 const double v_square_minus_one = square(radius_polar/radius_equatorial) - 1.0; 57 // translate a point in [-1,1] to a point in [0, 1] 58 // const double u = GAUSS_Z[i]*(upper-lower)/2 + (upper+lower)/2; 59 const double zm = 0.5; 60 const double zb = 0.5; 61 double total_F2 = 0.0; 62 double total_F1 = 0.0; 63 for (int i=0;i<GAUSS_N;i++) { 64 const double u = GAUSS_Z[i]*zm + zb; 65 const double r = radius_equatorial*sqrt(1.0 + u*u*v_square_minus_one); 66 const double f = sas_3j1x_x(q*r); 67 total_F2 += GAUSS_W[i] * f * f; 68 total_F1 += GAUSS_W[i] * f; 69 } 70 // translate dx in [-1,1] to dx in [lower,upper] 71 const double form_squared_avg = total_F2*zm; 72 const double form_avg = total_F1*zm; 73 const double s = (sld - sld_solvent) * form_volume(radius_polar, radius_equatorial); 74 *F2 = 1e-4 * s * s * form_squared_avg; 75 *F1 = 1e-2 * s * form_avg; 38 76 } 77 78 79 80 81 39 82 40 83 static double -
sasmodels/models/lib/sphere_form.c
r925ad6e r01c8d9e 13 13 return 1.0e-4*square(contrast * fq); 14 14 } 15 -
sasmodels/models/sphere.py
ref07e95 r01c8d9e 67 67 ] 68 68 69 source = ["lib/sas_3j1x_x.c", "lib/sphere_form.c" ]69 source = ["lib/sas_3j1x_x.c", "lib/sphere_form.c", "sphere.c"] 70 70 71 # No volume normalization despite having a volume parameter72 # This should perhaps be volume normalized?73 form_volume = """74 return sphere_volume(radius);75 """76 77 Iq = """78 return sphere_form(q, radius, sld, sld_solvent);79 """80 71 81 72 def ER(radius): -
sasmodels/product.py
r2d81cfe r01c8d9e 16 16 import numpy as np # type: ignore 17 17 18 from .modelinfo import ParameterTable, ModelInfo 18 from .modelinfo import ParameterTable, ModelInfo, Parameter 19 19 from .kernel import KernelModel, Kernel 20 20 from .details import make_details, dispersion_mesh … … 74 74 translate_name = dict((old.id, new.id) for old, new 75 75 in zip(s_pars.kernel_parameters[1:], s_list)) 76 combined_pars = p_pars.kernel_parameters + s_list 76 beta_parameter = Parameter("beta_mode", "", 0, [["P*S"],["P*(1+beta*(S-1))"], "", "Structure factor dispersion calculation mode"]) 77 combined_pars = p_pars.kernel_parameters + s_list + [beta_parameter] 77 78 parameters = ParameterTable(combined_pars) 78 79 parameters.max_pd = p_pars.max_pd + s_pars.max_pd … … 151 152 #: Structure factor modelling interaction between particles. 152 153 self.S = S 154 153 155 #: Model precision. This is not really relevant, since it is the 154 156 #: individual P and S models that control the effective dtype, … … 168 170 # in opencl; or both in opencl, but one in single precision and the 169 171 # other in double precision). 172 170 173 p_kernel = self.P.make_kernel(q_vectors) 171 174 s_kernel = self.S.make_kernel(q_vectors) … … 193 196 # type: (CallDetails, np.ndarray, float, bool) -> np.ndarray 194 197 p_info, s_info = self.info.composition[1] 195 196 198 # if there are magnetic parameters, they will only be on the 197 199 # form factor P, not the structure factor S. … … 205 207 nweights = call_details.num_weights 206 208 weights = values[nvalues:nvalues + 2*nweights] 207 208 209 # Construct the calling parameters for P. 209 210 p_npars = p_info.parameters.npars … … 218 219 p_values.append([0.]*spacer) 219 220 p_values = np.hstack(p_values).astype(self.p_kernel.dtype) 220 221 221 # Call ER and VR for P since these are needed for S. 222 222 p_er, p_vr = calc_er_vr(p_info, p_details, p_values) 223 223 s_vr = (volfrac/p_vr if p_vr != 0. else volfrac) 224 224 #print("volfrac:%g p_er:%g p_vr:%g s_vr:%g"%(volfrac,p_er,p_vr,s_vr)) 225 226 225 # Construct the calling parameters for S. 227 226 # The effective radius is not in the combined parameter list, so … … 253 252 s_values.append([0.]*spacer) 254 253 s_values = np.hstack(s_values).astype(self.s_kernel.dtype) 255 254 # beta mode is the first parameter after the structure factor pars 255 beta_index = 2+p_npars+s_npars 256 beta_mode = values[beta_index] 256 257 # Call the kernels 257 p_result = self.p_kernel(p_details, p_values, cutoff, magnetic) 258 s_result = self.s_kernel(s_details, s_values, cutoff, False) 259 258 if beta_mode: # beta: 259 F1, F2, volume_avg = self.p_kernel.beta(p_details, p_values, cutoff, magnetic) 260 else: 261 p_result = self.p_kernel.Iq(p_details, p_values, cutoff, magnetic) 262 s_result = self.s_kernel.Iq(s_details, s_values, cutoff, False) 260 263 #print("p_npars",p_npars,s_npars,p_er,s_vr,values[2+p_npars+1:2+p_npars+s_npars]) 261 264 #call_details.show(values) … … 265 268 #s_details.show(s_values) 266 269 #print("=>", s_result) 267 268 # remember the parts for plotting later269 self.results = [p_result, s_result]270 271 270 #import pylab as plt 272 271 #plt.subplot(211); plt.loglog(self.p_kernel.q_input.q, p_result, '-') 273 272 #plt.subplot(212); plt.loglog(self.s_kernel.q_input.q, s_result, '-') 274 273 #plt.figure() 275 276 return values[0]*(p_result*s_result) + values[1] 274 if beta_mode:#beta 275 beta_factor = F1**2/F2 276 Sq_eff = 1+beta_factor*(s_result - 1) 277 self.results = [F2, Sq_eff,F1,s_result] 278 final_result = volfrac*values[0]*(F2 + (F1**2)*(s_result - 1))/volume_avg+values[1] 279 #final_result = volfrac * values[0] * F2 * Sq_eff / volume_avg + values[1] 280 else: 281 # remember the parts for plotting later 282 self.results = [p_result, s_result] 283 final_result = values[0]*(p_result*s_result) + values[1] 284 return final_result 277 285 278 286 def release(self):
Note: See TracChangeset
for help on using the changeset viewer.