1 | """ |
---|
2 | Sasview model constructor. |
---|
3 | |
---|
4 | Given a module defining an OpenCL kernel such as sasmodels.models.cylinder, |
---|
5 | create a sasview model class to run that kernel as follows:: |
---|
6 | |
---|
7 | from sasmodels.sasview_model import load_custom_model |
---|
8 | CylinderModel = load_custom_model('sasmodels/models/cylinder.py') |
---|
9 | """ |
---|
10 | from __future__ import print_function |
---|
11 | |
---|
12 | import math |
---|
13 | from copy import deepcopy |
---|
14 | import collections |
---|
15 | import traceback |
---|
16 | import logging |
---|
17 | from os.path import basename, splitext |
---|
18 | import thread |
---|
19 | |
---|
20 | import numpy as np # type: ignore |
---|
21 | |
---|
22 | from . import core |
---|
23 | from . import custom |
---|
24 | from . import product |
---|
25 | from . import generate |
---|
26 | from . import weights |
---|
27 | from . import modelinfo |
---|
28 | from .details import make_kernel_args, dispersion_mesh |
---|
29 | |
---|
30 | try: |
---|
31 | from typing import Dict, Mapping, Any, Sequence, Tuple, NamedTuple, List, Optional, Union, Callable |
---|
32 | from .modelinfo import ModelInfo, Parameter |
---|
33 | from .kernel import KernelModel |
---|
34 | MultiplicityInfoType = NamedTuple( |
---|
35 | 'MuliplicityInfo', |
---|
36 | [("number", int), ("control", str), ("choices", List[str]), |
---|
37 | ("x_axis_label", str)]) |
---|
38 | SasviewModelType = Callable[[int], "SasviewModel"] |
---|
39 | except ImportError: |
---|
40 | pass |
---|
41 | |
---|
42 | calculation_lock = thread.allocate_lock() |
---|
43 | |
---|
44 | SUPPORT_OLD_STYLE_PLUGINS = True |
---|
45 | |
---|
46 | def _register_old_models(): |
---|
47 | # type: () -> None |
---|
48 | """ |
---|
49 | Place the new models into sasview under the old names. |
---|
50 | |
---|
51 | Monkey patch sas.sascalc.fit as sas.models so that sas.models.pluginmodel |
---|
52 | is available to the plugin modules. |
---|
53 | """ |
---|
54 | import sys |
---|
55 | import sas # needed in order to set sas.models |
---|
56 | import sas.sascalc.fit |
---|
57 | sys.modules['sas.models'] = sas.sascalc.fit |
---|
58 | sas.models = sas.sascalc.fit |
---|
59 | |
---|
60 | import sas.models |
---|
61 | from sasmodels.conversion_table import CONVERSION_TABLE |
---|
62 | for new_name, conversion in CONVERSION_TABLE.get((3,1,2), {}).items(): |
---|
63 | # CoreShellEllipsoidModel => core_shell_ellipsoid:1 |
---|
64 | new_name = new_name.split(':')[0] |
---|
65 | old_name = conversion[0] if len(conversion) < 3 else conversion[2] |
---|
66 | module_attrs = {old_name: find_model(new_name)} |
---|
67 | ConstructedModule = type(old_name, (), module_attrs) |
---|
68 | old_path = 'sas.models.' + old_name |
---|
69 | setattr(sas.models, old_path, ConstructedModule) |
---|
70 | sys.modules[old_path] = ConstructedModule |
---|
71 | |
---|
72 | |
---|
73 | # TODO: separate x_axis_label from multiplicity info |
---|
74 | MultiplicityInfo = collections.namedtuple( |
---|
75 | 'MultiplicityInfo', |
---|
76 | ["number", "control", "choices", "x_axis_label"], |
---|
77 | ) |
---|
78 | |
---|
79 | MODELS = {} |
---|
80 | def find_model(modelname): |
---|
81 | # type: (str) -> SasviewModelType |
---|
82 | """ |
---|
83 | Find a model by name. If the model name ends in py, try loading it from |
---|
84 | custom models, otherwise look for it in the list of builtin models. |
---|
85 | """ |
---|
86 | # TODO: used by sum/product model to load an existing model |
---|
87 | # TODO: doesn't handle custom models properly |
---|
88 | if modelname.endswith('.py'): |
---|
89 | return load_custom_model(modelname) |
---|
90 | elif modelname in MODELS: |
---|
91 | return MODELS[modelname] |
---|
92 | else: |
---|
93 | raise ValueError("unknown model %r"%modelname) |
---|
94 | |
---|
95 | |
---|
96 | # TODO: figure out how to say that the return type is a subclass |
---|
97 | def load_standard_models(): |
---|
98 | # type: () -> List[SasviewModelType] |
---|
99 | """ |
---|
100 | Load and return the list of predefined models. |
---|
101 | |
---|
102 | If there is an error loading a model, then a traceback is logged and the |
---|
103 | model is not returned. |
---|
104 | """ |
---|
105 | models = [] |
---|
106 | for name in core.list_models(): |
---|
107 | try: |
---|
108 | MODELS[name] = _make_standard_model(name) |
---|
109 | models.append(MODELS[name]) |
---|
110 | except Exception: |
---|
111 | logging.error(traceback.format_exc()) |
---|
112 | if SUPPORT_OLD_STYLE_PLUGINS: |
---|
113 | _register_old_models() |
---|
114 | |
---|
115 | return models |
---|
116 | |
---|
117 | |
---|
118 | def load_custom_model(path): |
---|
119 | # type: (str) -> SasviewModelType |
---|
120 | """ |
---|
121 | Load a custom model given the model path. |
---|
122 | """ |
---|
123 | kernel_module = custom.load_custom_kernel_module(path) |
---|
124 | try: |
---|
125 | model = kernel_module.Model |
---|
126 | # Old style models do not set the name in the class attributes, so |
---|
127 | # set it here; this name will be overridden when the object is created |
---|
128 | # with an instance variable that has the same value. |
---|
129 | if model.name == "": |
---|
130 | model.name = splitext(basename(path))[0] |
---|
131 | if not hasattr(model, 'filename'): |
---|
132 | model.filename = kernel_module.__file__ |
---|
133 | # For old models, treat .pyc and .py files interchangeably. |
---|
134 | # This is needed because of the Sum|Multi(p1,p2) types of models |
---|
135 | # and the convoluted way in which they are created. |
---|
136 | if model.filename.endswith(".py"): |
---|
137 | logging.info("Loading %s as .pyc", model.filename) |
---|
138 | model.filename = model.filename+'c' |
---|
139 | if not hasattr(model, 'id'): |
---|
140 | model.id = splitext(basename(model.filename))[0] |
---|
141 | except AttributeError: |
---|
142 | model_info = modelinfo.make_model_info(kernel_module) |
---|
143 | model = _make_model_from_info(model_info) |
---|
144 | |
---|
145 | # If a model name already exists and we are loading a different model, |
---|
146 | # use the model file name as the model name. |
---|
147 | if model.name in MODELS and not model.filename == MODELS[model.name].filename: |
---|
148 | _previous_name = model.name |
---|
149 | model.name = model.id |
---|
150 | |
---|
151 | # If the new model name is still in the model list (for instance, |
---|
152 | # if we put a cylinder.py in our plug-in directory), then append |
---|
153 | # an identifier. |
---|
154 | if model.name in MODELS and not model.filename == MODELS[model.name].filename: |
---|
155 | model.name = model.id + '_user' |
---|
156 | logging.info("Model %s already exists: using %s [%s]", _previous_name, model.name, model.filename) |
---|
157 | |
---|
158 | MODELS[model.name] = model |
---|
159 | return model |
---|
160 | |
---|
161 | |
---|
162 | def _make_standard_model(name): |
---|
163 | # type: (str) -> SasviewModelType |
---|
164 | """ |
---|
165 | Load the sasview model defined by *name*. |
---|
166 | |
---|
167 | *name* can be a standard model name or a path to a custom model. |
---|
168 | |
---|
169 | Returns a class that can be used directly as a sasview model. |
---|
170 | """ |
---|
171 | kernel_module = generate.load_kernel_module(name) |
---|
172 | model_info = modelinfo.make_model_info(kernel_module) |
---|
173 | return _make_model_from_info(model_info) |
---|
174 | |
---|
175 | |
---|
176 | def MultiplicationModel(form_factor, structure_factor): |
---|
177 | # type: ("SasviewModel", "SasviewModel") -> "SasviewModel" |
---|
178 | model_info = product.make_product_info(form_factor._model_info, |
---|
179 | structure_factor._model_info) |
---|
180 | ConstructedModel = _make_model_from_info(model_info) |
---|
181 | return ConstructedModel() |
---|
182 | |
---|
183 | def _make_model_from_info(model_info): |
---|
184 | # type: (ModelInfo) -> SasviewModelType |
---|
185 | """ |
---|
186 | Convert *model_info* into a SasView model wrapper. |
---|
187 | """ |
---|
188 | def __init__(self, multiplicity=None): |
---|
189 | SasviewModel.__init__(self, multiplicity=multiplicity) |
---|
190 | attrs = _generate_model_attributes(model_info) |
---|
191 | attrs['__init__'] = __init__ |
---|
192 | attrs['filename'] = model_info.filename |
---|
193 | ConstructedModel = type(model_info.name, (SasviewModel,), attrs) # type: SasviewModelType |
---|
194 | return ConstructedModel |
---|
195 | |
---|
196 | def _generate_model_attributes(model_info): |
---|
197 | # type: (ModelInfo) -> Dict[str, Any] |
---|
198 | """ |
---|
199 | Generate the class attributes for the model. |
---|
200 | |
---|
201 | This should include all the information necessary to query the model |
---|
202 | details so that you do not need to instantiate a model to query it. |
---|
203 | |
---|
204 | All the attributes should be immutable to avoid accidents. |
---|
205 | """ |
---|
206 | |
---|
207 | # TODO: allow model to override axis labels input/output name/unit |
---|
208 | |
---|
209 | # Process multiplicity |
---|
210 | non_fittable = [] # type: List[str] |
---|
211 | xlabel = model_info.profile_axes[0] if model_info.profile is not None else "" |
---|
212 | variants = MultiplicityInfo(0, "", [], xlabel) |
---|
213 | for p in model_info.parameters.kernel_parameters: |
---|
214 | if p.name == model_info.control: |
---|
215 | non_fittable.append(p.name) |
---|
216 | variants = MultiplicityInfo( |
---|
217 | len(p.choices) if p.choices else int(p.limits[1]), |
---|
218 | p.name, p.choices, xlabel |
---|
219 | ) |
---|
220 | break |
---|
221 | |
---|
222 | # Only a single drop-down list parameter available |
---|
223 | fun_list = [] |
---|
224 | for p in model_info.parameters.kernel_parameters: |
---|
225 | if p.choices: |
---|
226 | fun_list = p.choices |
---|
227 | if p.length > 1: |
---|
228 | non_fittable.extend(p.id+str(k) for k in range(1, p.length+1)) |
---|
229 | break |
---|
230 | |
---|
231 | # Organize parameter sets |
---|
232 | orientation_params = [] |
---|
233 | magnetic_params = [] |
---|
234 | fixed = [] |
---|
235 | for p in model_info.parameters.user_parameters({}, is2d=True): |
---|
236 | if p.type == 'orientation': |
---|
237 | orientation_params.append(p.name) |
---|
238 | orientation_params.append(p.name+".width") |
---|
239 | fixed.append(p.name+".width") |
---|
240 | elif p.type == 'magnetic': |
---|
241 | orientation_params.append(p.name) |
---|
242 | magnetic_params.append(p.name) |
---|
243 | fixed.append(p.name+".width") |
---|
244 | |
---|
245 | |
---|
246 | # Build class dictionary |
---|
247 | attrs = {} # type: Dict[str, Any] |
---|
248 | attrs['_model_info'] = model_info |
---|
249 | attrs['name'] = model_info.name |
---|
250 | attrs['id'] = model_info.id |
---|
251 | attrs['description'] = model_info.description |
---|
252 | attrs['category'] = model_info.category |
---|
253 | attrs['is_structure_factor'] = model_info.structure_factor |
---|
254 | attrs['is_form_factor'] = model_info.ER is not None |
---|
255 | attrs['is_multiplicity_model'] = variants[0] > 1 |
---|
256 | attrs['multiplicity_info'] = variants |
---|
257 | attrs['orientation_params'] = tuple(orientation_params) |
---|
258 | attrs['magnetic_params'] = tuple(magnetic_params) |
---|
259 | attrs['fixed'] = tuple(fixed) |
---|
260 | attrs['non_fittable'] = tuple(non_fittable) |
---|
261 | attrs['fun_list'] = tuple(fun_list) |
---|
262 | |
---|
263 | return attrs |
---|
264 | |
---|
265 | class SasviewModel(object): |
---|
266 | """ |
---|
267 | Sasview wrapper for opencl/ctypes model. |
---|
268 | """ |
---|
269 | # Model parameters for the specific model are set in the class constructor |
---|
270 | # via the _generate_model_attributes function, which subclasses |
---|
271 | # SasviewModel. They are included here for typing and documentation |
---|
272 | # purposes. |
---|
273 | _model = None # type: KernelModel |
---|
274 | _model_info = None # type: ModelInfo |
---|
275 | #: load/save name for the model |
---|
276 | id = None # type: str |
---|
277 | #: display name for the model |
---|
278 | name = None # type: str |
---|
279 | #: short model description |
---|
280 | description = None # type: str |
---|
281 | #: default model category |
---|
282 | category = None # type: str |
---|
283 | |
---|
284 | #: names of the orientation parameters in the order they appear |
---|
285 | orientation_params = None # type: Sequence[str] |
---|
286 | #: names of the magnetic parameters in the order they appear |
---|
287 | magnetic_params = None # type: Sequence[str] |
---|
288 | #: names of the fittable parameters |
---|
289 | fixed = None # type: Sequence[str] |
---|
290 | # TODO: the attribute fixed is ill-named |
---|
291 | |
---|
292 | # Axis labels |
---|
293 | input_name = "Q" |
---|
294 | input_unit = "A^{-1}" |
---|
295 | output_name = "Intensity" |
---|
296 | output_unit = "cm^{-1}" |
---|
297 | |
---|
298 | #: default cutoff for polydispersity |
---|
299 | cutoff = 1e-5 |
---|
300 | |
---|
301 | # Note: Use non-mutable values for class attributes to avoid errors |
---|
302 | #: parameters that are not fitted |
---|
303 | non_fittable = () # type: Sequence[str] |
---|
304 | |
---|
305 | #: True if model should appear as a structure factor |
---|
306 | is_structure_factor = False |
---|
307 | #: True if model should appear as a form factor |
---|
308 | is_form_factor = False |
---|
309 | #: True if model has multiplicity |
---|
310 | is_multiplicity_model = False |
---|
311 | #: Mulitplicity information |
---|
312 | multiplicity_info = None # type: MultiplicityInfoType |
---|
313 | |
---|
314 | # Per-instance variables |
---|
315 | #: parameter {name: value} mapping |
---|
316 | params = None # type: Dict[str, float] |
---|
317 | #: values for dispersion width, npts, nsigmas and type |
---|
318 | dispersion = None # type: Dict[str, Any] |
---|
319 | #: units and limits for each parameter |
---|
320 | details = None # type: Dict[str, Sequence[Any]] |
---|
321 | # # actual type is Dict[str, List[str, float, float]] |
---|
322 | #: multiplicity value, or None if no multiplicity on the model |
---|
323 | multiplicity = None # type: Optional[int] |
---|
324 | #: memory for polydispersity array if using ArrayDispersion (used by sasview). |
---|
325 | _persistency_dict = None # type: Dict[str, Tuple[np.ndarray, np.ndarray]] |
---|
326 | |
---|
327 | def __init__(self, multiplicity=None): |
---|
328 | # type: (Optional[int]) -> None |
---|
329 | |
---|
330 | # TODO: _persistency_dict to persistency_dict throughout sasview |
---|
331 | # TODO: refactor multiplicity to encompass variants |
---|
332 | # TODO: dispersion should be a class |
---|
333 | # TODO: refactor multiplicity info |
---|
334 | # TODO: separate profile view from multiplicity |
---|
335 | # The button label, x and y axis labels and scale need to be under |
---|
336 | # the control of the model, not the fit page. Maximum flexibility, |
---|
337 | # the fit page would supply the canvas and the profile could plot |
---|
338 | # how it wants, but this assumes matplotlib. Next level is that |
---|
339 | # we provide some sort of data description including title, labels |
---|
340 | # and lines to plot. |
---|
341 | |
---|
342 | # Get the list of hidden parameters given the mulitplicity |
---|
343 | # Don't include multiplicity in the list of parameters |
---|
344 | self.multiplicity = multiplicity |
---|
345 | if multiplicity is not None: |
---|
346 | hidden = self._model_info.get_hidden_parameters(multiplicity) |
---|
347 | hidden |= set([self.multiplicity_info.control]) |
---|
348 | else: |
---|
349 | hidden = set() |
---|
350 | if self._model_info.structure_factor: |
---|
351 | hidden.add('scale') |
---|
352 | hidden.add('background') |
---|
353 | self._model_info.parameters.defaults['background'] = 0. |
---|
354 | |
---|
355 | self._persistency_dict = {} |
---|
356 | self.params = collections.OrderedDict() |
---|
357 | self.dispersion = collections.OrderedDict() |
---|
358 | self.details = {} |
---|
359 | for p in self._model_info.parameters.user_parameters({}, is2d=True): |
---|
360 | if p.name in hidden: |
---|
361 | continue |
---|
362 | self.params[p.name] = p.default |
---|
363 | self.details[p.id] = [p.units, p.limits[0], p.limits[1]] |
---|
364 | if p.polydisperse: |
---|
365 | self.details[p.id+".width"] = [ |
---|
366 | "", 0.0, 1.0 if p.relative_pd else np.inf |
---|
367 | ] |
---|
368 | self.dispersion[p.name] = { |
---|
369 | 'width': 0, |
---|
370 | 'npts': 35, |
---|
371 | 'nsigmas': 3, |
---|
372 | 'type': 'gaussian', |
---|
373 | } |
---|
374 | |
---|
375 | def __get_state__(self): |
---|
376 | # type: () -> Dict[str, Any] |
---|
377 | state = self.__dict__.copy() |
---|
378 | state.pop('_model') |
---|
379 | # May need to reload model info on set state since it has pointers |
---|
380 | # to python implementations of Iq, etc. |
---|
381 | #state.pop('_model_info') |
---|
382 | return state |
---|
383 | |
---|
384 | def __set_state__(self, state): |
---|
385 | # type: (Dict[str, Any]) -> None |
---|
386 | self.__dict__ = state |
---|
387 | self._model = None |
---|
388 | |
---|
389 | def __str__(self): |
---|
390 | # type: () -> str |
---|
391 | """ |
---|
392 | :return: string representation |
---|
393 | """ |
---|
394 | return self.name |
---|
395 | |
---|
396 | def is_fittable(self, par_name): |
---|
397 | # type: (str) -> bool |
---|
398 | """ |
---|
399 | Check if a given parameter is fittable or not |
---|
400 | |
---|
401 | :param par_name: the parameter name to check |
---|
402 | """ |
---|
403 | return par_name in self.fixed |
---|
404 | #For the future |
---|
405 | #return self.params[str(par_name)].is_fittable() |
---|
406 | |
---|
407 | |
---|
408 | def getProfile(self): |
---|
409 | # type: () -> (np.ndarray, np.ndarray) |
---|
410 | """ |
---|
411 | Get SLD profile |
---|
412 | |
---|
413 | : return: (z, beta) where z is a list of depth of the transition points |
---|
414 | beta is a list of the corresponding SLD values |
---|
415 | """ |
---|
416 | args = {} # type: Dict[str, Any] |
---|
417 | for p in self._model_info.parameters.kernel_parameters: |
---|
418 | if p.id == self.multiplicity_info.control: |
---|
419 | value = float(self.multiplicity) |
---|
420 | elif p.length == 1: |
---|
421 | value = self.params.get(p.id, np.NaN) |
---|
422 | else: |
---|
423 | value = np.array([self.params.get(p.id+str(k), np.NaN) |
---|
424 | for k in range(1, p.length+1)]) |
---|
425 | args[p.id] = value |
---|
426 | |
---|
427 | x, y = self._model_info.profile(**args) |
---|
428 | return x, 1e-6*y |
---|
429 | |
---|
430 | def setParam(self, name, value): |
---|
431 | # type: (str, float) -> None |
---|
432 | """ |
---|
433 | Set the value of a model parameter |
---|
434 | |
---|
435 | :param name: name of the parameter |
---|
436 | :param value: value of the parameter |
---|
437 | |
---|
438 | """ |
---|
439 | # Look for dispersion parameters |
---|
440 | toks = name.split('.') |
---|
441 | if len(toks) == 2: |
---|
442 | for item in self.dispersion.keys(): |
---|
443 | if item == toks[0]: |
---|
444 | for par in self.dispersion[item]: |
---|
445 | if par == toks[1]: |
---|
446 | self.dispersion[item][par] = value |
---|
447 | return |
---|
448 | else: |
---|
449 | # Look for standard parameter |
---|
450 | for item in self.params.keys(): |
---|
451 | if item == name: |
---|
452 | self.params[item] = value |
---|
453 | return |
---|
454 | |
---|
455 | raise ValueError("Model does not contain parameter %s" % name) |
---|
456 | |
---|
457 | def getParam(self, name): |
---|
458 | # type: (str) -> float |
---|
459 | """ |
---|
460 | Set the value of a model parameter |
---|
461 | |
---|
462 | :param name: name of the parameter |
---|
463 | |
---|
464 | """ |
---|
465 | # Look for dispersion parameters |
---|
466 | toks = name.split('.') |
---|
467 | if len(toks) == 2: |
---|
468 | for item in self.dispersion.keys(): |
---|
469 | if item == toks[0]: |
---|
470 | for par in self.dispersion[item]: |
---|
471 | if par == toks[1]: |
---|
472 | return self.dispersion[item][par] |
---|
473 | else: |
---|
474 | # Look for standard parameter |
---|
475 | for item in self.params.keys(): |
---|
476 | if item == name: |
---|
477 | return self.params[item] |
---|
478 | |
---|
479 | raise ValueError("Model does not contain parameter %s" % name) |
---|
480 | |
---|
481 | def getParamList(self): |
---|
482 | # type: () -> Sequence[str] |
---|
483 | """ |
---|
484 | Return a list of all available parameters for the model |
---|
485 | """ |
---|
486 | param_list = list(self.params.keys()) |
---|
487 | # WARNING: Extending the list with the dispersion parameters |
---|
488 | param_list.extend(self.getDispParamList()) |
---|
489 | return param_list |
---|
490 | |
---|
491 | def getDispParamList(self): |
---|
492 | # type: () -> Sequence[str] |
---|
493 | """ |
---|
494 | Return a list of polydispersity parameters for the model |
---|
495 | """ |
---|
496 | # TODO: fix test so that parameter order doesn't matter |
---|
497 | ret = ['%s.%s' % (p_name, ext) |
---|
498 | for p_name in self.dispersion.keys() |
---|
499 | for ext in ('npts', 'nsigmas', 'width')] |
---|
500 | #print(ret) |
---|
501 | return ret |
---|
502 | |
---|
503 | def clone(self): |
---|
504 | # type: () -> "SasviewModel" |
---|
505 | """ Return a identical copy of self """ |
---|
506 | return deepcopy(self) |
---|
507 | |
---|
508 | def run(self, x=0.0): |
---|
509 | # type: (Union[float, (float, float), List[float]]) -> float |
---|
510 | """ |
---|
511 | Evaluate the model |
---|
512 | |
---|
513 | :param x: input q, or [q,phi] |
---|
514 | |
---|
515 | :return: scattering function P(q) |
---|
516 | |
---|
517 | **DEPRECATED**: use calculate_Iq instead |
---|
518 | """ |
---|
519 | if isinstance(x, (list, tuple)): |
---|
520 | # pylint: disable=unpacking-non-sequence |
---|
521 | q, phi = x |
---|
522 | return self.calculate_Iq([q*math.cos(phi)], [q*math.sin(phi)])[0] |
---|
523 | else: |
---|
524 | return self.calculate_Iq([x])[0] |
---|
525 | |
---|
526 | |
---|
527 | def runXY(self, x=0.0): |
---|
528 | # type: (Union[float, (float, float), List[float]]) -> float |
---|
529 | """ |
---|
530 | Evaluate the model in cartesian coordinates |
---|
531 | |
---|
532 | :param x: input q, or [qx, qy] |
---|
533 | |
---|
534 | :return: scattering function P(q) |
---|
535 | |
---|
536 | **DEPRECATED**: use calculate_Iq instead |
---|
537 | """ |
---|
538 | if isinstance(x, (list, tuple)): |
---|
539 | return self.calculate_Iq([x[0]], [x[1]])[0] |
---|
540 | else: |
---|
541 | return self.calculate_Iq([x])[0] |
---|
542 | |
---|
543 | def evalDistribution(self, qdist): |
---|
544 | # type: (Union[np.ndarray, Tuple[np.ndarray, np.ndarray], List[np.ndarray]]) -> np.ndarray |
---|
545 | r""" |
---|
546 | Evaluate a distribution of q-values. |
---|
547 | |
---|
548 | :param qdist: array of q or a list of arrays [qx,qy] |
---|
549 | |
---|
550 | * For 1D, a numpy array is expected as input |
---|
551 | |
---|
552 | :: |
---|
553 | |
---|
554 | evalDistribution(q) |
---|
555 | |
---|
556 | where *q* is a numpy array. |
---|
557 | |
---|
558 | * For 2D, a list of *[qx,qy]* is expected with 1D arrays as input |
---|
559 | |
---|
560 | :: |
---|
561 | |
---|
562 | qx = [ qx[0], qx[1], qx[2], ....] |
---|
563 | qy = [ qy[0], qy[1], qy[2], ....] |
---|
564 | |
---|
565 | If the model is 1D only, then |
---|
566 | |
---|
567 | .. math:: |
---|
568 | |
---|
569 | q = \sqrt{q_x^2+q_y^2} |
---|
570 | |
---|
571 | """ |
---|
572 | if isinstance(qdist, (list, tuple)): |
---|
573 | # Check whether we have a list of ndarrays [qx,qy] |
---|
574 | qx, qy = qdist |
---|
575 | if not self._model_info.parameters.has_2d: |
---|
576 | return self.calculate_Iq(np.sqrt(qx ** 2 + qy ** 2)) |
---|
577 | else: |
---|
578 | return self.calculate_Iq(qx, qy) |
---|
579 | |
---|
580 | elif isinstance(qdist, np.ndarray): |
---|
581 | # We have a simple 1D distribution of q-values |
---|
582 | return self.calculate_Iq(qdist) |
---|
583 | |
---|
584 | else: |
---|
585 | raise TypeError("evalDistribution expects q or [qx, qy], not %r" |
---|
586 | % type(qdist)) |
---|
587 | |
---|
588 | def get_composition_models(self): |
---|
589 | """ |
---|
590 | Returns usable models that compose this model |
---|
591 | """ |
---|
592 | s_model = None |
---|
593 | p_model = None |
---|
594 | if hasattr(self._model_info, "composition") \ |
---|
595 | and self._model_info.composition is not None: |
---|
596 | p_model = _make_model_from_info(self._model_info.composition[1][0])() |
---|
597 | s_model = _make_model_from_info(self._model_info.composition[1][1])() |
---|
598 | return p_model, s_model |
---|
599 | |
---|
600 | def calculate_Iq(self, qx, qy=None): |
---|
601 | # type: (Sequence[float], Optional[Sequence[float]]) -> np.ndarray |
---|
602 | """ |
---|
603 | Calculate Iq for one set of q with the current parameters. |
---|
604 | |
---|
605 | If the model is 1D, use *q*. If 2D, use *qx*, *qy*. |
---|
606 | |
---|
607 | This should NOT be used for fitting since it copies the *q* vectors |
---|
608 | to the card for each evaluation. |
---|
609 | """ |
---|
610 | ## uncomment the following when trying to debug the uncoordinated calls |
---|
611 | ## to calculate_Iq |
---|
612 | #if calculation_lock.locked(): |
---|
613 | # logging.info("calculation waiting for another thread to complete") |
---|
614 | # logging.info("\n".join(traceback.format_stack())) |
---|
615 | |
---|
616 | with calculation_lock: |
---|
617 | return self._calculate_Iq(qx, qy) |
---|
618 | |
---|
619 | def _calculate_Iq(self, qx, qy=None): |
---|
620 | #core.HAVE_OPENCL = False |
---|
621 | if self._model is None: |
---|
622 | self._model = core.build_model(self._model_info) |
---|
623 | if qy is not None: |
---|
624 | q_vectors = [np.asarray(qx), np.asarray(qy)] |
---|
625 | else: |
---|
626 | q_vectors = [np.asarray(qx)] |
---|
627 | calculator = self._model.make_kernel(q_vectors) |
---|
628 | parameters = self._model_info.parameters |
---|
629 | pairs = [self._get_weights(p) for p in parameters.call_parameters] |
---|
630 | #weights.plot_weights(self._model_info, pairs) |
---|
631 | call_details, values, is_magnetic = make_kernel_args(calculator, pairs) |
---|
632 | #call_details.show() |
---|
633 | #print("pairs", pairs) |
---|
634 | #print("params", self.params) |
---|
635 | #print("values", values) |
---|
636 | #print("is_mag", is_magnetic) |
---|
637 | result = calculator(call_details, values, cutoff=self.cutoff, |
---|
638 | magnetic=is_magnetic) |
---|
639 | calculator.release() |
---|
640 | self._model.release() |
---|
641 | return result |
---|
642 | |
---|
643 | def calculate_ER(self): |
---|
644 | # type: () -> float |
---|
645 | """ |
---|
646 | Calculate the effective radius for P(q)*S(q) |
---|
647 | |
---|
648 | :return: the value of the effective radius |
---|
649 | """ |
---|
650 | if self._model_info.ER is None: |
---|
651 | return 1.0 |
---|
652 | else: |
---|
653 | value, weight = self._dispersion_mesh() |
---|
654 | fv = self._model_info.ER(*value) |
---|
655 | #print(values[0].shape, weights.shape, fv.shape) |
---|
656 | return np.sum(weight * fv) / np.sum(weight) |
---|
657 | |
---|
658 | def calculate_VR(self): |
---|
659 | # type: () -> float |
---|
660 | """ |
---|
661 | Calculate the volf ratio for P(q)*S(q) |
---|
662 | |
---|
663 | :return: the value of the volf ratio |
---|
664 | """ |
---|
665 | if self._model_info.VR is None: |
---|
666 | return 1.0 |
---|
667 | else: |
---|
668 | value, weight = self._dispersion_mesh() |
---|
669 | whole, part = self._model_info.VR(*value) |
---|
670 | return np.sum(weight * part) / np.sum(weight * whole) |
---|
671 | |
---|
672 | def set_dispersion(self, parameter, dispersion): |
---|
673 | # type: (str, weights.Dispersion) -> Dict[str, Any] |
---|
674 | """ |
---|
675 | Set the dispersion object for a model parameter |
---|
676 | |
---|
677 | :param parameter: name of the parameter [string] |
---|
678 | :param dispersion: dispersion object of type Dispersion |
---|
679 | """ |
---|
680 | if parameter in self.params: |
---|
681 | # TODO: Store the disperser object directly in the model. |
---|
682 | # The current method of relying on the sasview GUI to |
---|
683 | # remember them is kind of funky. |
---|
684 | # Note: can't seem to get disperser parameters from sasview |
---|
685 | # (1) Could create a sasview model that has not yet been |
---|
686 | # converted, assign the disperser to one of its polydisperse |
---|
687 | # parameters, then retrieve the disperser parameters from the |
---|
688 | # sasview model. |
---|
689 | # (2) Could write a disperser parameter retriever in sasview. |
---|
690 | # (3) Could modify sasview to use sasmodels.weights dispersers. |
---|
691 | # For now, rely on the fact that the sasview only ever uses |
---|
692 | # new dispersers in the set_dispersion call and create a new |
---|
693 | # one instead of trying to assign parameters. |
---|
694 | self.dispersion[parameter] = dispersion.get_pars() |
---|
695 | else: |
---|
696 | raise ValueError("%r is not a dispersity or orientation parameter") |
---|
697 | |
---|
698 | def _dispersion_mesh(self): |
---|
699 | # type: () -> List[Tuple[np.ndarray, np.ndarray]] |
---|
700 | """ |
---|
701 | Create a mesh grid of dispersion parameters and weights. |
---|
702 | |
---|
703 | Returns [p1,p2,...],w where pj is a vector of values for parameter j |
---|
704 | and w is a vector containing the products for weights for each |
---|
705 | parameter set in the vector. |
---|
706 | """ |
---|
707 | pars = [self._get_weights(p) |
---|
708 | for p in self._model_info.parameters.call_parameters |
---|
709 | if p.type == 'volume'] |
---|
710 | return dispersion_mesh(self._model_info, pars) |
---|
711 | |
---|
712 | def _get_weights(self, par): |
---|
713 | # type: (Parameter) -> Tuple[np.ndarray, np.ndarray] |
---|
714 | """ |
---|
715 | Return dispersion weights for parameter |
---|
716 | """ |
---|
717 | if par.name not in self.params: |
---|
718 | if par.name == self.multiplicity_info.control: |
---|
719 | return [self.multiplicity], [1.0] |
---|
720 | else: |
---|
721 | # For hidden parameters use the default value. |
---|
722 | value = self._model_info.parameters.defaults.get(par.name, np.NaN) |
---|
723 | return [value], [1.0] |
---|
724 | elif par.polydisperse: |
---|
725 | dis = self.dispersion[par.name] |
---|
726 | if dis['type'] == 'array': |
---|
727 | value, weight = dis['values'], dis['weights'] |
---|
728 | else: |
---|
729 | value, weight = weights.get_weights( |
---|
730 | dis['type'], dis['npts'], dis['width'], dis['nsigmas'], |
---|
731 | self.params[par.name], par.limits, par.relative_pd) |
---|
732 | return value, weight / np.sum(weight) |
---|
733 | else: |
---|
734 | return [self.params[par.name]], [1.0] |
---|
735 | |
---|
736 | def test_cylinder(): |
---|
737 | # type: () -> float |
---|
738 | """ |
---|
739 | Test that the cylinder model runs, returning the value at [0.1,0.1]. |
---|
740 | """ |
---|
741 | Cylinder = _make_standard_model('cylinder') |
---|
742 | cylinder = Cylinder() |
---|
743 | return cylinder.evalDistribution([0.1, 0.1]) |
---|
744 | |
---|
745 | def test_structure_factor(): |
---|
746 | # type: () -> float |
---|
747 | """ |
---|
748 | Test that 2-D hardsphere model runs and doesn't produce NaN. |
---|
749 | """ |
---|
750 | Model = _make_standard_model('hardsphere') |
---|
751 | model = Model() |
---|
752 | value = model.evalDistribution([0.1, 0.1]) |
---|
753 | if np.isnan(value): |
---|
754 | raise ValueError("hardsphere returns null") |
---|
755 | |
---|
756 | def test_rpa(): |
---|
757 | # type: () -> float |
---|
758 | """ |
---|
759 | Test that the 2-D RPA model runs |
---|
760 | """ |
---|
761 | RPA = _make_standard_model('rpa') |
---|
762 | rpa = RPA(3) |
---|
763 | return rpa.evalDistribution([0.1, 0.1]) |
---|
764 | |
---|
765 | def test_empty_distribution(): |
---|
766 | # type: () -> None |
---|
767 | """ |
---|
768 | Make sure that sasmodels returns NaN when there are no polydispersity points |
---|
769 | """ |
---|
770 | Cylinder = _make_standard_model('cylinder') |
---|
771 | cylinder = Cylinder() |
---|
772 | cylinder.setParam('radius', -1.0) |
---|
773 | cylinder.setParam('background', 0.) |
---|
774 | Iq = cylinder.evalDistribution(np.asarray([0.1])) |
---|
775 | assert np.isnan(Iq[0]), "empty distribution fails" |
---|
776 | |
---|
777 | def test_model_list(): |
---|
778 | # type: () -> None |
---|
779 | """ |
---|
780 | Make sure that all models build as sasview models |
---|
781 | """ |
---|
782 | from .exception import annotate_exception |
---|
783 | for name in core.list_models(): |
---|
784 | try: |
---|
785 | _make_standard_model(name) |
---|
786 | except: |
---|
787 | annotate_exception("when loading "+name) |
---|
788 | raise |
---|
789 | |
---|
790 | def test_old_name(): |
---|
791 | # type: () -> None |
---|
792 | """ |
---|
793 | Load and run cylinder model from sas.models.CylinderModel |
---|
794 | """ |
---|
795 | if not SUPPORT_OLD_STYLE_PLUGINS: |
---|
796 | return |
---|
797 | try: |
---|
798 | # if sasview is not on the path then don't try to test it |
---|
799 | import sas |
---|
800 | except ImportError: |
---|
801 | return |
---|
802 | load_standard_models() |
---|
803 | from sas.models.CylinderModel import CylinderModel |
---|
804 | CylinderModel().evalDistribution([0.1, 0.1]) |
---|
805 | |
---|
806 | if __name__ == "__main__": |
---|
807 | print("cylinder(0.1,0.1)=%g"%test_cylinder()) |
---|
808 | #test_empty_distribution() |
---|