Changeset e589e9a in sasmodels
- Timestamp:
- Mar 5, 2019 10:36:34 AM (6 years ago)
- Branches:
- master, core_shell_microgels, magnetic_model, ticket-1257-vesicle-product, ticket_1156, ticket_1265_superball, ticket_822_more_unit_tests
- Children:
- 3b6567f, bd91e8f, c11d09f, 9150036
- Parents:
- a8a1d48 (diff), 8064d5e (diff)
Note: this is a merge changeset, the changes displayed below correspond to the merge itself.
Use the (diff) links above to see all the changes relative to each parent. - git-author:
- Paul Butler <butlerpd@…> (03/05/19 10:36:34)
- git-committer:
- GitHub <noreply@…> (03/05/19 10:36:34)
- Files:
-
- 15 edited
- 6 moved
Legend:
- Unmodified
- Added
- Removed
-
sasmodels/jitter.py
r1198f90 r7d97437 15 15 pass 16 16 17 import matplotlib as mpl 17 18 import matplotlib.pyplot as plt 18 19 from matplotlib.widgets import Slider … … 746 747 pass 747 748 748 axcolor = 'lightgoldenrodyellow' 749 # CRUFT: use axisbg instead of facecolor for matplotlib<2 750 facecolor_prop = 'facecolor' if mpl.__version__ > '2' else 'axisbg' 751 props = {facecolor_prop: 'lightgoldenrodyellow'} 749 752 750 753 ## add control widgets to plot 751 axes_theta = plt.axes([0.1, 0.15, 0.45, 0.04], axisbg=axcolor)752 axes_phi = plt.axes([0.1, 0.1, 0.45, 0.04], axisbg=axcolor)753 axes_psi = plt.axes([0.1, 0.05, 0.45, 0.04], axisbg=axcolor)754 axes_theta = plt.axes([0.1, 0.15, 0.45, 0.04], **props) 755 axes_phi = plt.axes([0.1, 0.1, 0.45, 0.04], **props) 756 axes_psi = plt.axes([0.1, 0.05, 0.45, 0.04], **props) 754 757 stheta = Slider(axes_theta, 'Theta', -90, 90, valinit=theta) 755 758 sphi = Slider(axes_phi, 'Phi', -180, 180, valinit=phi) 756 759 spsi = Slider(axes_psi, 'Psi', -180, 180, valinit=psi) 757 760 758 axes_dtheta = plt.axes([0.75, 0.15, 0.15, 0.04], axisbg=axcolor)759 axes_dphi = plt.axes([0.75, 0.1, 0.15, 0.04], axisbg=axcolor)760 axes_dpsi = plt.axes([0.75, 0.05, 0.15, 0.04], axisbg=axcolor)761 axes_dtheta = plt.axes([0.75, 0.15, 0.15, 0.04], **props) 762 axes_dphi = plt.axes([0.75, 0.1, 0.15, 0.04], **props) 763 axes_dpsi = plt.axes([0.75, 0.05, 0.15, 0.04], **props) 761 764 # Note: using ridiculous definition of rectangle distribution, whose width 762 765 # in sasmodels is sqrt(3) times the given width. Divide by sqrt(3) to keep -
sasmodels/kernelcl.py
rf872fd1 r8064d5e 58 58 import time 59 59 60 try: 61 from time import perf_counter as clock 62 except ImportError: # CRUFT: python < 3.3 63 import sys 64 if sys.platform.count("darwin") > 0: 65 from time import time as clock 66 else: 67 from time import clock 68 60 69 import numpy as np # type: ignore 61 62 70 63 71 # Attempt to setup opencl. This may fail if the pyopencl package is not … … 611 619 #Call kernel and retrieve results 612 620 wait_for = None 613 last_nap = time.clock()621 last_nap = clock() 614 622 step = 1000000//self.q_input.nq + 1 615 623 for start in range(0, call_details.num_eval, step): … … 622 630 # Allow other processes to run 623 631 wait_for[0].wait() 624 current_time = time.clock()632 current_time = clock() 625 633 if current_time - last_nap > 0.5: 626 634 time.sleep(0.001) -
sasmodels/resolution.py
r9e7837a re2592f0 445 445 q = np.sort(q) 446 446 if q_min + 2*MINIMUM_RESOLUTION < q[0]: 447 n_low = np.ceil((q[0]-q_min) / (q[1]-q[0])) if q[1] > q[0] else 15447 n_low = int(np.ceil((q[0]-q_min) / (q[1]-q[0]))) if q[1] > q[0] else 15 448 448 q_low = np.linspace(q_min, q[0], n_low+1)[:-1] 449 449 else: 450 450 q_low = [] 451 451 if q_max - 2*MINIMUM_RESOLUTION > q[-1]: 452 n_high = np.ceil((q_max-q[-1]) / (q[-1]-q[-2])) if q[-1] > q[-2] else 15452 n_high = int(np.ceil((q_max-q[-1]) / (q[-1]-q[-2]))) if q[-1] > q[-2] else 15 453 453 q_high = np.linspace(q[-1], q_max, n_high+1)[1:] 454 454 else: … … 499 499 q_min = q[0]*MINIMUM_ABSOLUTE_Q 500 500 n_low = log_delta_q * (log(q[0])-log(q_min)) 501 q_low = np.logspace(log10(q_min), log10(q[0]), np.ceil(n_low)+1)[:-1]501 q_low = np.logspace(log10(q_min), log10(q[0]), int(np.ceil(n_low))+1)[:-1] 502 502 else: 503 503 q_low = [] 504 504 if q_max > q[-1]: 505 505 n_high = log_delta_q * (log(q_max)-log(q[-1])) 506 q_high = np.logspace(log10(q[-1]), log10(q_max), np.ceil(n_high)+1)[1:]506 q_high = np.logspace(log10(q[-1]), log10(q_max), int(np.ceil(n_high))+1)[1:] 507 507 else: 508 508 q_high = [] -
sasmodels/weights.py
r3d58247 re2592f0 23 23 default = dict(npts=35, width=0, nsigmas=3) 24 24 def __init__(self, npts=None, width=None, nsigmas=None): 25 self.npts = self.default['npts'] if npts is None else npts25 self.npts = self.default['npts'] if npts is None else int(npts) 26 26 self.width = self.default['width'] if width is None else width 27 27 self.nsigmas = self.default['nsigmas'] if nsigmas is None else nsigmas -
README.rst
re30d645 r2a64722 10 10 is available. 11 11 12 Example 12 Install 13 13 ------- 14 15 The easiest way to use sasmodels is from `SasView <http://www.sasview.org/>`_. 16 17 You can also install sasmodels as a standalone package in python. Use 18 `miniconda <https://docs.conda.io/en/latest/miniconda.html>`_ 19 or `anaconda <https://www.anaconda.com/>`_ 20 to create a python environment with the sasmodels dependencies:: 21 22 $ conda create -n sasmodels -c conda-forge numpy scipy matplotlib pyopencl 23 24 The option ``-n sasmodels`` names the environment sasmodels, and the option 25 ``-c conda-forge`` selects the conda-forge package channel because pyopencl 26 is not part of the base anaconda distribution. 27 28 Activate the environment and install sasmodels:: 29 30 $ conda activate sasmodels 31 (sasmodels) $ pip install sasmodels 32 33 Install `bumps <https://github.com/bumps/bumps>`_ if you want to use it to fit 34 your data:: 35 36 (sasmodels) $ pip install bumps 37 38 Usage 39 ----- 40 41 Check that the works:: 42 43 (sasmodels) $ python -m sasmodels.compare cylinder 44 45 To show the orientation explorer:: 46 47 (sasmodels) $ python -m sasmodels.jitter 48 49 Documentation is available online as part of the SasView 50 `fitting perspective <http://www.sasview.org/docs/index.html>`_ 51 as well as separate pages for 52 `individual models <http://www.sasview.org/docs/user/sasgui/perspectives/fitting/models/index.html>`_. 53 Programming details for sasmodels are available in the 54 `developer documentation <http://www.sasview.org/docs/dev/dev.html>`_. 55 56 57 Fitting Example 58 --------------- 14 59 15 60 The example directory contains a radial+tangential data set for an oriented 16 61 rod-like shape. 17 62 18 The data is loaded by sas.dataloader from the sasview package, so sasview 19 is needed to run the example. 63 To load the example data, you will need the SAS data loader from the sasview 64 package. This is not yet available on PyPI, so you will need a copy of the 65 SasView source code to run it. Create a directory somewhere to hold the 66 sasview and sasmodels source code, which we will refer to as $SOURCE. 20 67 21 To run the example, you need sasview, sasmodels and bumps. Assuming these 22 repositories are installed side by side, change to the sasmodels/example 23 directory and enter:: 68 Use the following to install sasview, and the sasmodels examples:: 24 69 25 PYTHONPATH=..:../../sasview/src ../../bumps/run.py fit.py \ 26 cylinder --preview 70 (sasmodels) $ cd $SOURCE 71 (sasmodels) $ conda install git 72 (sasmodels) $ git clone https://github.com/sasview/sasview.git 73 (sasmodels) $ git clone https://github.com/sasview/sasmodels.git 27 74 28 See bumps documentation for instructions on running the fit. With the 29 python packages installed, e.g., into a virtual environment, then the 30 python path need not be set, and the command would be:: 75 Set the path to the sasview source on your python path within the sasmodels 76 environment. On Windows, this will be:: 31 77 32 bumps fit.py cylinder --preview 78 (sasmodels)> set PYTHONPATH="$SOURCE\sasview\src" 79 (sasmodels)> cd $SOURCE/sasmodels/example 80 (sasmodels)> python -m bumps.cli fit.py cylinder --preview 81 82 On Mac/Linux with the standard shell this will be:: 83 84 (sasmodels) $ export PYTHONPATH="$SOURCE/sasview/src" 85 (sasmodels) $ cd $SOURCE/sasmodels/example 86 (sasmodels) $ bumps fit.py cylinder --preview 33 87 34 88 The fit.py model accepts up to two arguments. The first argument is the … … 38 92 both radial and tangential simultaneously, use the word "both". 39 93 40 Notes 41 ----- 42 43 cylinder.c + cylinder.py is the cylinder model with renamed variables and 44 sld scaled by 1e6 so the numbers are nicer. The model name is "cylinder" 45 46 lamellar.py is an example of a single file model with embedded C code. 94 See `bumps documentation <https://bumps.readthedocs.io/>`_ for detailed 95 instructions on running the fit. 47 96 48 97 |TravisStatus|_ -
explore/beta/sasfit_compare.py
r2a12351b r119073a 505 505 } 506 506 507 Q, IQ = load_sasfit(data_file(' richard_test.txt'))508 Q, IQSD = load_sasfit(data_file(' richard_test2.txt'))509 Q, IQBD = load_sasfit(data_file(' richard_test3.txt'))507 Q, IQ = load_sasfit(data_file('sasfit_sphere_schulz_IQD.txt')) 508 Q, IQSD = load_sasfit(data_file('sasfit_sphere_schulz_IQSD.txt')) 509 Q, IQBD = load_sasfit(data_file('sasfit_sphere_schulz_IQBD.txt')) 510 510 target = Theory(Q=Q, F1=None, F2=None, P=IQ, S=None, I=IQSD, Seff=None, Ibeta=IQBD) 511 511 actual = sphere_r(Q, norm="sasfit", **pars) … … 526 526 } 527 527 528 Q, IQ = load_sasfit(data_file(' richard_test4.txt'))529 Q, IQSD = load_sasfit(data_file(' richard_test5.txt'))530 Q, IQBD = load_sasfit(data_file(' richard_test6.txt'))528 Q, IQ = load_sasfit(data_file('sasfit_ellipsoid_shulz_IQD.txt')) 529 Q, IQSD = load_sasfit(data_file('sasfit_ellipsoid_shulz_IQSD.txt')) 530 Q, IQBD = load_sasfit(data_file('sasfit_ellipsoid_shulz_IQBD.txt')) 531 531 target = Theory(Q=Q, F1=None, F2=None, P=IQ, S=None, I=IQSD, Seff=None, Ibeta=IQBD) 532 532 actual = ellipsoid_pe(Q, norm="sasfit", **pars) -
explore/precision.py
raa8c6e0 rcd28947 207 207 return model_info 208 208 209 # Hack to allow second parameter A in two parameter functions 209 # Hack to allow second parameter A in the gammainc and gammaincc functions. 210 # Create a 2-D variant of the precision test if we need to handle other two 211 # parameter functions. 210 212 A = 1 211 213 def parse_extra_pars(): 214 """ 215 Parse the command line looking for the second parameter "A=..." for the 216 gammainc/gammaincc functions. 217 """ 212 218 global A 213 219 … … 333 339 ) 334 340 add_function( 341 # Note: "a" is given as A=... on the command line via parse_extra_pars 335 342 name="gammainc(x)", 336 343 mp_function=lambda x, a=A: mp.gammainc(a, a=0, b=x)/mp.gamma(a), … … 339 346 ) 340 347 add_function( 348 # Note: "a" is given as A=... on the command line via parse_extra_pars 341 349 name="gammaincc(x)", 342 350 mp_function=lambda x, a=A: mp.gammainc(a, a=x, b=mp.inf)/mp.gamma(a), -
sasmodels/compare.py
r07646b6 rc1799d3 1152 1152 'rel_err' : True, 1153 1153 'explore' : False, 1154 'use_demo' : True,1154 'use_demo' : False, 1155 1155 'zero' : False, 1156 1156 'html' : False, -
sasmodels/direct_model.py
r5024a56 rc1799d3 332 332 333 333 # Need to pull background out of resolution for multiple scattering 334 background = pars.get('background', DEFAULT_BACKGROUND) 334 default_background = self._model.info.parameters.common_parameters[1].default 335 background = pars.get('background', default_background) 335 336 pars = pars.copy() 336 337 pars['background'] = 0. -
sasmodels/generate.py
r39a06c9 ra8a1d48 703 703 """ 704 704 for code in source: 705 m = _FQ_PATTERN.search(code) 706 if m is not None: 705 if _FQ_PATTERN.search(code) is not None: 707 706 return True 708 707 return False … … 712 711 # type: (List[str]) -> bool 713 712 """ 714 Return True if C source defines " void Fq(".713 Return True if C source defines "double shell_volume(". 715 714 """ 716 715 for code in source: 717 m = _SHELL_VOLUME_PATTERN.search(code) 718 if m is not None: 716 if _SHELL_VOLUME_PATTERN.search(code) is not None: 719 717 return True 720 718 return False … … 1008 1006 pars = model_info.parameters.kernel_parameters 1009 1007 else: 1010 pars = model_info.parameters.COMMON + model_info.parameters.kernel_parameters 1008 pars = (model_info.parameters.common_parameters 1009 + model_info.parameters.kernel_parameters) 1011 1010 partable = make_partable(pars) 1012 1011 subst = dict(id=model_info.id.replace('_', '-'), -
sasmodels/kernel.py
re44432d rcd28947 133 133 nout = 2 if self.info.have_Fq and self.dim == '1d' else 1 134 134 total_weight = self.result[nout*self.q_input.nq + 0] 135 # Note: total_weight = sum(weight > cutoff), with cutoff >= 0, so it 136 # is okay to test directly against zero. If weight is zero then I(q), 137 # etc. must also be zero. 135 138 if total_weight == 0.: 136 139 total_weight = 1. -
sasmodels/modelinfo.py
r39a06c9 rc1799d3 404 404 parameters counted as n individual parameters p1, p2, ... 405 405 406 * *common_parameters* is the list of common parameters, with a unique 407 copy for each model so that structure factors can have a default 408 background of 0.0. 409 406 410 * *call_parameters* is the complete list of parameters to the kernel, 407 411 including scale and background, with vector parameters recorded as … … 416 420 parameters don't use vector notation, and instead use p1, p2, ... 417 421 """ 418 # scale and background are implicit parameters419 COMMON = [Parameter(*p) for p in COMMON_PARAMETERS]420 421 422 def __init__(self, parameters): 422 423 # type: (List[Parameter]) -> None 424 425 # scale and background are implicit parameters 426 # Need them to be unique to each model in case they have different 427 # properties, such as default=0.0 for structure factor backgrounds. 428 self.common_parameters = [Parameter(*p) for p in COMMON_PARAMETERS] 429 423 430 self.kernel_parameters = parameters 424 431 self._set_vector_lengths() … … 468 475 if p.polydisperse and p.type not in ('orientation', 'magnetic')) 469 476 self.pd_2d = set(p.name for p in self.call_parameters if p.polydisperse) 477 478 def set_zero_background(self): 479 """ 480 Set the default background to zero for this model. This is done for 481 structure factor models. 482 """ 483 # type: () -> None 484 # Make sure background is the second common parameter. 485 assert self.common_parameters[1].id == "background" 486 self.common_parameters[1].default = 0.0 487 self.defaults = self._get_defaults() 470 488 471 489 def check_angles(self): … … 567 585 def _get_call_parameters(self): 568 586 # type: () -> List[Parameter] 569 full_list = self. COMMON[:]587 full_list = self.common_parameters[:] 570 588 for p in self.kernel_parameters: 571 589 if p.length == 1: … … 670 688 671 689 # Gather the user parameters in order 672 result = control + self. COMMON690 result = control + self.common_parameters 673 691 for p in self.kernel_parameters: 674 692 if not is2d and p.type in ('orientation', 'magnetic'): … … 770 788 771 789 info = ModelInfo() 790 791 # Build the parameter table 772 792 #print("make parameter table", kernel_module.parameters) 773 793 parameters = make_parameter_table(getattr(kernel_module, 'parameters', [])) 794 795 # background defaults to zero for structure factor models 796 structure_factor = getattr(kernel_module, 'structure_factor', False) 797 if structure_factor: 798 parameters.set_zero_background() 799 800 # TODO: remove demo parameters 801 # The plots in the docs are generated from the model default values. 802 # Sascomp set parameters from the command line, and so doesn't need 803 # demo values for testing. 774 804 demo = expand_pars(parameters, getattr(kernel_module, 'demo', None)) 805 775 806 filename = abspath(kernel_module.__file__).replace('.pyc', '.py') 776 807 kernel_id = splitext(basename(filename))[0] -
sasmodels/models/hardsphere.py
r304c775 rc1799d3 162 162 return pars 163 163 164 demo = dict(radius_effective=200, volfraction=0.2,165 radius_effective_pd=0.1, radius_effective_pd_n=40)166 164 # Q=0.001 is in the Taylor series, low Q part, so add Q=0.1, 167 165 # assuming double precision sasview is correct -
sasmodels/models/pearl_necklace.c
r99658f6 r9b5fd42 40 40 const double si = sas_sinx_x(q*A_s); 41 41 const double omsi = 1.0 - si; 42 const double pow_si = pow (si, num_pearls);42 const double pow_si = pown(si, num_pearls); 43 43 44 44 // form factor for num_pearls … … 81 81 radius_from_volume(double radius, double edge_sep, double thick_string, double fp_num_pearls) 82 82 { 83 const int num_pearls = (int) fp_num_pearls +0.5;84 83 const double vol_tot = form_volume(radius, edge_sep, thick_string, fp_num_pearls); 85 84 return cbrt(vol_tot/M_4PI_3); -
sasmodels/sasview_model.py
r5024a56 ra8a1d48 382 382 hidden.add('scale') 383 383 hidden.add('background') 384 self._model_info.parameters.defaults['background'] = 0.385 384 386 385 # Update the parameter lists to exclude any hidden parameters … … 695 694 return self._calculate_Iq(qx, qy) 696 695 697 def _calculate_Iq(self, qx, qy=None , Fq=False, effective_radius_type=1):696 def _calculate_Iq(self, qx, qy=None): 698 697 if self._model is None: 699 698 self._model = core.build_model(self._model_info) … … 715 714 #print("values", values) 716 715 #print("is_mag", is_magnetic) 717 if Fq:718 result = calculator.Fq(call_details, values, cutoff=self.cutoff,719 magnetic=is_magnetic,720 effective_radius_type=effective_radius_type)721 716 result = calculator(call_details, values, cutoff=self.cutoff, 722 717 magnetic=is_magnetic) … … 736 731 Calculate the effective radius for P(q)*S(q) 737 732 733 *mode* is the R_eff type, which defaults to 1 to match the ER 734 calculation for sasview models from version 3.x. 735 738 736 :return: the value of the effective radius 739 737 """ 740 Fq = self._calculate_Iq([0.1], True, mode) 741 return Fq[2] 738 # ER and VR are only needed for old multiplication models, based on 739 # sas.sascalc.fit.MultiplicationModel. Fail for now. If we want to 740 # continue supporting them then add some test cases so that the code 741 # is exercised. We can access ER/VR using the kernel Fq function by 742 # extending _calculate_Iq so that it calls: 743 # if er_mode > 0: 744 # res = calculator.Fq(call_details, values, cutoff=self.cutoff, 745 # magnetic=False, effective_radius_type=mode) 746 # R_eff, form_shell_ratio = res[2], res[4] 747 # return R_eff, form_shell_ratio 748 # Then use the following in calculate_ER: 749 # ER, VR = self._calculate_Iq(q=[0.1], er_mode=mode) 750 # return ER 751 # Similarly, for calculate_VR: 752 # ER, VR = self._calculate_Iq(q=[0.1], er_mode=1) 753 # return VR 754 # Obviously a combined calculate_ER_VR method would be better, but 755 # we only need them to support very old models, so ignore the 2x 756 # performance hit. 757 raise NotImplementedError("ER function is no longer available.") 742 758 743 759 def calculate_VR(self): … … 748 764 :return: the value of the form:shell volume ratio 749 765 """ 750 Fq = self._calculate_Iq([0.1], True, mode)751 r eturn Fq[4]766 # See comments in calculate_ER. 767 raise NotImplementedError("VR function is no longer available.") 752 768 753 769 def set_dispersion(self, parameter, dispersion): … … 914 930 CylinderModel().evalDistribution([0.1, 0.1]) 915 931 932 def test_structure_factor_background(): 933 # type: () -> None 934 """ 935 Check that sasview model and direct model match, with background=0. 936 """ 937 from .data import empty_data1D 938 from .core import load_model_info, build_model 939 from .direct_model import DirectModel 940 941 model_name = "hardsphere" 942 q = [0.0] 943 944 sasview_model = _make_standard_model(model_name)() 945 sasview_value = sasview_model.evalDistribution(np.array(q))[0] 946 947 data = empty_data1D(q) 948 model_info = load_model_info(model_name) 949 model = build_model(model_info) 950 direct_model = DirectModel(data, model) 951 direct_value_zero_background = direct_model(background=0.0) 952 953 assert sasview_value == direct_value_zero_background 954 955 # Additionally check that direct value background defaults to zero 956 direct_value_default = direct_model() 957 assert sasview_value == direct_value_default 958 959 916 960 def magnetic_demo(): 917 961 Model = _make_standard_model('sphere') … … 934 978 #print("rpa:", test_rpa()) 935 979 #test_empty_distribution() 980 #test_structure_factor_background()
Note: See TracChangeset
for help on using the changeset viewer.