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 | |
---|
18 | import numpy as np # type: ignore |
---|
19 | |
---|
20 | from . import core |
---|
21 | from . import custom |
---|
22 | from . import generate |
---|
23 | from . import weights |
---|
24 | from . import modelinfo |
---|
25 | from . import kernel |
---|
26 | |
---|
27 | try: |
---|
28 | from typing import Dict, Mapping, Any, Sequence, Tuple, NamedTuple, List, Optional, Union, Callable |
---|
29 | from .modelinfo import ModelInfo, Parameter |
---|
30 | from .kernel import KernelModel |
---|
31 | MultiplicityInfoType = NamedTuple( |
---|
32 | 'MuliplicityInfo', |
---|
33 | [("number", int), ("control", str), ("choices", List[str]), |
---|
34 | ("x_axis_label", str)]) |
---|
35 | SasviewModelType = Callable[[int], "SasviewModel"] |
---|
36 | except ImportError: |
---|
37 | pass |
---|
38 | |
---|
39 | # TODO: separate x_axis_label from multiplicity info |
---|
40 | # The profile x-axis label belongs with the profile generating function |
---|
41 | MultiplicityInfo = collections.namedtuple( |
---|
42 | 'MultiplicityInfo', |
---|
43 | ["number", "control", "choices", "x_axis_label"], |
---|
44 | ) |
---|
45 | |
---|
46 | MODELS = {} |
---|
47 | def find_model(modelname): |
---|
48 | # TODO: used by sum/product model to load an existing model |
---|
49 | # TODO: doesn't handle custom models properly |
---|
50 | if modelname.endswith('.py'): |
---|
51 | return load_custom_model(modelname) |
---|
52 | elif modelname in MODELS: |
---|
53 | return MODELS[modelname] |
---|
54 | else: |
---|
55 | raise ValueError("unknown model %r"%modelname) |
---|
56 | |
---|
57 | |
---|
58 | # TODO: figure out how to say that the return type is a subclass |
---|
59 | def load_standard_models(): |
---|
60 | # type: () -> List[SasviewModelType] |
---|
61 | """ |
---|
62 | Load and return the list of predefined models. |
---|
63 | |
---|
64 | If there is an error loading a model, then a traceback is logged and the |
---|
65 | model is not returned. |
---|
66 | """ |
---|
67 | models = [] |
---|
68 | for name in core.list_models(): |
---|
69 | try: |
---|
70 | MODELS[name] = _make_standard_model(name) |
---|
71 | models.append(MODELS[name]) |
---|
72 | except Exception: |
---|
73 | logging.error(traceback.format_exc()) |
---|
74 | return models |
---|
75 | |
---|
76 | |
---|
77 | def load_custom_model(path): |
---|
78 | # type: (str) -> SasviewModelType |
---|
79 | """ |
---|
80 | Load a custom model given the model path. |
---|
81 | """ |
---|
82 | #print("load custom model", path) |
---|
83 | kernel_module = custom.load_custom_kernel_module(path) |
---|
84 | try: |
---|
85 | model = kernel_module.Model |
---|
86 | except AttributeError: |
---|
87 | model_info = modelinfo.make_model_info(kernel_module) |
---|
88 | model = _make_model_from_info(model_info) |
---|
89 | MODELS[model.name] = model |
---|
90 | return model |
---|
91 | |
---|
92 | |
---|
93 | def _make_standard_model(name): |
---|
94 | # type: (str) -> SasviewModelType |
---|
95 | """ |
---|
96 | Load the sasview model defined by *name*. |
---|
97 | |
---|
98 | *name* can be a standard model name or a path to a custom model. |
---|
99 | |
---|
100 | Returns a class that can be used directly as a sasview model. |
---|
101 | """ |
---|
102 | kernel_module = generate.load_kernel_module(name) |
---|
103 | model_info = modelinfo.make_model_info(kernel_module) |
---|
104 | return _make_model_from_info(model_info) |
---|
105 | |
---|
106 | |
---|
107 | def _make_model_from_info(model_info): |
---|
108 | # type: (ModelInfo) -> SasviewModelType |
---|
109 | """ |
---|
110 | Convert *model_info* into a SasView model wrapper. |
---|
111 | """ |
---|
112 | def __init__(self, multiplicity=None): |
---|
113 | SasviewModel.__init__(self, multiplicity=multiplicity) |
---|
114 | attrs = _generate_model_attributes(model_info) |
---|
115 | attrs['__init__'] = __init__ |
---|
116 | ConstructedModel = type(model_info.name, (SasviewModel,), attrs) # type: SasviewModelType |
---|
117 | return ConstructedModel |
---|
118 | |
---|
119 | def _generate_model_attributes(model_info): |
---|
120 | # type: (ModelInfo) -> Dict[str, Any] |
---|
121 | """ |
---|
122 | Generate the class attributes for the model. |
---|
123 | |
---|
124 | This should include all the information necessary to query the model |
---|
125 | details so that you do not need to instantiate a model to query it. |
---|
126 | |
---|
127 | All the attributes should be immutable to avoid accidents. |
---|
128 | """ |
---|
129 | |
---|
130 | # TODO: allow model to override axis labels input/output name/unit |
---|
131 | |
---|
132 | # Process multiplicity |
---|
133 | non_fittable = [] # type: List[str] |
---|
134 | xlabel = model_info.profile_axes[0] if model_info.profile is not None else "" |
---|
135 | variants = MultiplicityInfo(0, "", [], xlabel) |
---|
136 | for p in model_info.parameters.kernel_parameters: |
---|
137 | if p.name == model_info.control: |
---|
138 | non_fittable.append(p.name) |
---|
139 | variants = MultiplicityInfo( |
---|
140 | len(p.choices), p.name, p.choices, xlabel |
---|
141 | ) |
---|
142 | break |
---|
143 | elif p.is_control: |
---|
144 | non_fittable.append(p.name) |
---|
145 | variants = MultiplicityInfo( |
---|
146 | int(p.limits[1]), p.name, p.choices, xlabel |
---|
147 | ) |
---|
148 | break |
---|
149 | |
---|
150 | # Organize parameter sets |
---|
151 | orientation_params = [] |
---|
152 | magnetic_params = [] |
---|
153 | fixed = [] |
---|
154 | for p in model_info.parameters.user_parameters(): |
---|
155 | if p.type == 'orientation': |
---|
156 | orientation_params.append(p.name) |
---|
157 | orientation_params.append(p.name+".width") |
---|
158 | fixed.append(p.name+".width") |
---|
159 | if p.type == 'magnetic': |
---|
160 | orientation_params.append(p.name) |
---|
161 | magnetic_params.append(p.name) |
---|
162 | fixed.append(p.name+".width") |
---|
163 | |
---|
164 | # Build class dictionary |
---|
165 | attrs = {} # type: Dict[str, Any] |
---|
166 | attrs['_model_info'] = model_info |
---|
167 | attrs['name'] = model_info.name |
---|
168 | attrs['id'] = model_info.id |
---|
169 | attrs['description'] = model_info.description |
---|
170 | attrs['category'] = model_info.category |
---|
171 | attrs['is_structure_factor'] = model_info.structure_factor |
---|
172 | attrs['is_form_factor'] = model_info.ER is not None |
---|
173 | attrs['is_multiplicity_model'] = variants[0] > 1 |
---|
174 | attrs['multiplicity_info'] = variants |
---|
175 | attrs['orientation_params'] = tuple(orientation_params) |
---|
176 | attrs['magnetic_params'] = tuple(magnetic_params) |
---|
177 | attrs['fixed'] = tuple(fixed) |
---|
178 | attrs['non_fittable'] = tuple(non_fittable) |
---|
179 | |
---|
180 | return attrs |
---|
181 | |
---|
182 | class SasviewModel(object): |
---|
183 | """ |
---|
184 | Sasview wrapper for opencl/ctypes model. |
---|
185 | """ |
---|
186 | # Model parameters for the specific model are set in the class constructor |
---|
187 | # via the _generate_model_attributes function, which subclasses |
---|
188 | # SasviewModel. They are included here for typing and documentation |
---|
189 | # purposes. |
---|
190 | _model = None # type: KernelModel |
---|
191 | _model_info = None # type: ModelInfo |
---|
192 | #: load/save name for the model |
---|
193 | id = None # type: str |
---|
194 | #: display name for the model |
---|
195 | name = None # type: str |
---|
196 | #: short model description |
---|
197 | description = None # type: str |
---|
198 | #: default model category |
---|
199 | category = None # type: str |
---|
200 | |
---|
201 | #: names of the orientation parameters in the order they appear |
---|
202 | orientation_params = None # type: Sequence[str] |
---|
203 | #: names of the magnetic parameters in the order they appear |
---|
204 | magnetic_params = None # type: Sequence[str] |
---|
205 | #: names of the fittable parameters |
---|
206 | fixed = None # type: Sequence[str] |
---|
207 | # TODO: the attribute fixed is ill-named |
---|
208 | |
---|
209 | # Axis labels |
---|
210 | input_name = "Q" |
---|
211 | input_unit = "A^{-1}" |
---|
212 | output_name = "Intensity" |
---|
213 | output_unit = "cm^{-1}" |
---|
214 | |
---|
215 | #: default cutoff for polydispersity |
---|
216 | cutoff = 1e-5 |
---|
217 | |
---|
218 | # Note: Use non-mutable values for class attributes to avoid errors |
---|
219 | #: parameters that are not fitted |
---|
220 | non_fittable = () # type: Sequence[str] |
---|
221 | |
---|
222 | #: True if model should appear as a structure factor |
---|
223 | is_structure_factor = False |
---|
224 | #: True if model should appear as a form factor |
---|
225 | is_form_factor = False |
---|
226 | #: True if model has multiplicity |
---|
227 | is_multiplicity_model = False |
---|
228 | #: Mulitplicity information |
---|
229 | multiplicity_info = None # type: MultiplicityInfoType |
---|
230 | |
---|
231 | # Per-instance variables |
---|
232 | #: parameter {name: value} mapping |
---|
233 | params = None # type: Dict[str, float] |
---|
234 | #: values for dispersion width, npts, nsigmas and type |
---|
235 | dispersion = None # type: Dict[str, Any] |
---|
236 | #: units and limits for each parameter |
---|
237 | details = None # type: Dict[str, Sequence[Any]] |
---|
238 | # # actual type is Dict[str, List[str, float, float]] |
---|
239 | #: multiplicity value, or None if no multiplicity on the model |
---|
240 | multiplicity = None # type: Optional[int] |
---|
241 | #: memory for polydispersity array if using ArrayDispersion (used by sasview). |
---|
242 | _persistency_dict = None # type: Dict[str, Tuple[np.ndarray, np.ndarray]] |
---|
243 | |
---|
244 | def __init__(self, multiplicity=None): |
---|
245 | # type: (Optional[int]) -> None |
---|
246 | |
---|
247 | # TODO: _persistency_dict to persistency_dict throughout sasview |
---|
248 | # TODO: refactor multiplicity to encompass variants |
---|
249 | # TODO: dispersion should be a class |
---|
250 | # TODO: refactor multiplicity info |
---|
251 | # TODO: separate profile view from multiplicity |
---|
252 | # The button label, x and y axis labels and scale need to be under |
---|
253 | # the control of the model, not the fit page. Maximum flexibility, |
---|
254 | # the fit page would supply the canvas and the profile could plot |
---|
255 | # how it wants, but this assumes matplotlib. Next level is that |
---|
256 | # we provide some sort of data description including title, labels |
---|
257 | # and lines to plot. |
---|
258 | |
---|
259 | # Get the list of hidden parameters given the mulitplicity |
---|
260 | # Don't include multiplicity in the list of parameters |
---|
261 | self.multiplicity = multiplicity |
---|
262 | if multiplicity is not None: |
---|
263 | hidden = self._model_info.get_hidden_parameters(multiplicity) |
---|
264 | hidden |= set([self.multiplicity_info.control]) |
---|
265 | else: |
---|
266 | hidden = set() |
---|
267 | |
---|
268 | self._persistency_dict = {} |
---|
269 | self.params = collections.OrderedDict() |
---|
270 | self.dispersion = collections.OrderedDict() |
---|
271 | self.details = {} |
---|
272 | for p in self._model_info.parameters.user_parameters(): |
---|
273 | if p.name in hidden: |
---|
274 | continue |
---|
275 | self.params[p.name] = p.default |
---|
276 | self.details[p.id] = [p.units, p.limits[0], p.limits[1]] |
---|
277 | if p.polydisperse: |
---|
278 | self.details[p.id+".width"] = [ |
---|
279 | "", 0.0, 1.0 if p.relative_pd else np.inf |
---|
280 | ] |
---|
281 | self.dispersion[p.name] = { |
---|
282 | 'width': 0, |
---|
283 | 'npts': 35, |
---|
284 | 'nsigmas': 3, |
---|
285 | 'type': 'gaussian', |
---|
286 | } |
---|
287 | |
---|
288 | def __get_state__(self): |
---|
289 | # type: () -> Dict[str, Any] |
---|
290 | state = self.__dict__.copy() |
---|
291 | state.pop('_model') |
---|
292 | # May need to reload model info on set state since it has pointers |
---|
293 | # to python implementations of Iq, etc. |
---|
294 | #state.pop('_model_info') |
---|
295 | return state |
---|
296 | |
---|
297 | def __set_state__(self, state): |
---|
298 | # type: (Dict[str, Any]) -> None |
---|
299 | self.__dict__ = state |
---|
300 | self._model = None |
---|
301 | |
---|
302 | def __str__(self): |
---|
303 | # type: () -> str |
---|
304 | """ |
---|
305 | :return: string representation |
---|
306 | """ |
---|
307 | return self.name |
---|
308 | |
---|
309 | def is_fittable(self, par_name): |
---|
310 | # type: (str) -> bool |
---|
311 | """ |
---|
312 | Check if a given parameter is fittable or not |
---|
313 | |
---|
314 | :param par_name: the parameter name to check |
---|
315 | """ |
---|
316 | return par_name in self.fixed |
---|
317 | #For the future |
---|
318 | #return self.params[str(par_name)].is_fittable() |
---|
319 | |
---|
320 | |
---|
321 | def getProfile(self): |
---|
322 | # type: () -> (np.ndarray, np.ndarray) |
---|
323 | """ |
---|
324 | Get SLD profile |
---|
325 | |
---|
326 | : return: (z, beta) where z is a list of depth of the transition points |
---|
327 | beta is a list of the corresponding SLD values |
---|
328 | """ |
---|
329 | args = [] # type: List[Union[float, np.ndarray]] |
---|
330 | for p in self._model_info.parameters.kernel_parameters: |
---|
331 | if p.id == self.multiplicity_info.control: |
---|
332 | args.append(float(self.multiplicity)) |
---|
333 | elif p.length == 1: |
---|
334 | args.append(self.params.get(p.id, np.NaN)) |
---|
335 | else: |
---|
336 | args.append([self.params.get(p.id+str(k), np.NaN) |
---|
337 | for k in range(1,p.length+1)]) |
---|
338 | return self._model_info.profile(*args) |
---|
339 | |
---|
340 | def setParam(self, name, value): |
---|
341 | # type: (str, float) -> None |
---|
342 | """ |
---|
343 | Set the value of a model parameter |
---|
344 | |
---|
345 | :param name: name of the parameter |
---|
346 | :param value: value of the parameter |
---|
347 | |
---|
348 | """ |
---|
349 | # Look for dispersion parameters |
---|
350 | toks = name.split('.') |
---|
351 | if len(toks) == 2: |
---|
352 | for item in self.dispersion.keys(): |
---|
353 | if item == toks[0]: |
---|
354 | for par in self.dispersion[item]: |
---|
355 | if par == toks[1]: |
---|
356 | self.dispersion[item][par] = value |
---|
357 | return |
---|
358 | else: |
---|
359 | # Look for standard parameter |
---|
360 | for item in self.params.keys(): |
---|
361 | if item == name: |
---|
362 | self.params[item] = value |
---|
363 | return |
---|
364 | |
---|
365 | raise ValueError("Model does not contain parameter %s" % name) |
---|
366 | |
---|
367 | def getParam(self, name): |
---|
368 | # type: (str) -> float |
---|
369 | """ |
---|
370 | Set the value of a model parameter |
---|
371 | |
---|
372 | :param name: name of the parameter |
---|
373 | |
---|
374 | """ |
---|
375 | # Look for dispersion parameters |
---|
376 | toks = name.split('.') |
---|
377 | if len(toks) == 2: |
---|
378 | for item in self.dispersion.keys(): |
---|
379 | if item == toks[0]: |
---|
380 | for par in self.dispersion[item]: |
---|
381 | if par == toks[1]: |
---|
382 | return self.dispersion[item][par] |
---|
383 | else: |
---|
384 | # Look for standard parameter |
---|
385 | for item in self.params.keys(): |
---|
386 | if item == name: |
---|
387 | return self.params[item] |
---|
388 | |
---|
389 | raise ValueError("Model does not contain parameter %s" % name) |
---|
390 | |
---|
391 | def getParamList(self): |
---|
392 | # type: () -> Sequence[str] |
---|
393 | """ |
---|
394 | Return a list of all available parameters for the model |
---|
395 | """ |
---|
396 | param_list = list(self.params.keys()) |
---|
397 | # WARNING: Extending the list with the dispersion parameters |
---|
398 | param_list.extend(self.getDispParamList()) |
---|
399 | return param_list |
---|
400 | |
---|
401 | def getDispParamList(self): |
---|
402 | # type: () -> Sequence[str] |
---|
403 | """ |
---|
404 | Return a list of polydispersity parameters for the model |
---|
405 | """ |
---|
406 | # TODO: fix test so that parameter order doesn't matter |
---|
407 | ret = ['%s.%s' % (p.name, ext) |
---|
408 | for p in self._model_info.parameters.user_parameters() |
---|
409 | for ext in ('npts', 'nsigmas', 'width') |
---|
410 | if p.polydisperse] |
---|
411 | #print(ret) |
---|
412 | return ret |
---|
413 | |
---|
414 | def clone(self): |
---|
415 | # type: () -> "SasviewModel" |
---|
416 | """ Return a identical copy of self """ |
---|
417 | return deepcopy(self) |
---|
418 | |
---|
419 | def run(self, x=0.0): |
---|
420 | # type: (Union[float, (float, float), List[float]]) -> float |
---|
421 | """ |
---|
422 | Evaluate the model |
---|
423 | |
---|
424 | :param x: input q, or [q,phi] |
---|
425 | |
---|
426 | :return: scattering function P(q) |
---|
427 | |
---|
428 | **DEPRECATED**: use calculate_Iq instead |
---|
429 | """ |
---|
430 | if isinstance(x, (list, tuple)): |
---|
431 | # pylint: disable=unpacking-non-sequence |
---|
432 | q, phi = x |
---|
433 | return self.calculate_Iq([q*math.cos(phi)], [q*math.sin(phi)])[0] |
---|
434 | else: |
---|
435 | return self.calculate_Iq([x])[0] |
---|
436 | |
---|
437 | |
---|
438 | def runXY(self, x=0.0): |
---|
439 | # type: (Union[float, (float, float), List[float]]) -> float |
---|
440 | """ |
---|
441 | Evaluate the model in cartesian coordinates |
---|
442 | |
---|
443 | :param x: input q, or [qx, qy] |
---|
444 | |
---|
445 | :return: scattering function P(q) |
---|
446 | |
---|
447 | **DEPRECATED**: use calculate_Iq instead |
---|
448 | """ |
---|
449 | if isinstance(x, (list, tuple)): |
---|
450 | return self.calculate_Iq([x[0]], [x[1]])[0] |
---|
451 | else: |
---|
452 | return self.calculate_Iq([x])[0] |
---|
453 | |
---|
454 | def evalDistribution(self, qdist): |
---|
455 | # type: (Union[np.ndarray, Tuple[np.ndarray, np.ndarray], List[np.ndarray]]) -> np.ndarray |
---|
456 | r""" |
---|
457 | Evaluate a distribution of q-values. |
---|
458 | |
---|
459 | :param qdist: array of q or a list of arrays [qx,qy] |
---|
460 | |
---|
461 | * For 1D, a numpy array is expected as input |
---|
462 | |
---|
463 | :: |
---|
464 | |
---|
465 | evalDistribution(q) |
---|
466 | |
---|
467 | where *q* is a numpy array. |
---|
468 | |
---|
469 | * For 2D, a list of *[qx,qy]* is expected with 1D arrays as input |
---|
470 | |
---|
471 | :: |
---|
472 | |
---|
473 | qx = [ qx[0], qx[1], qx[2], ....] |
---|
474 | qy = [ qy[0], qy[1], qy[2], ....] |
---|
475 | |
---|
476 | If the model is 1D only, then |
---|
477 | |
---|
478 | .. math:: |
---|
479 | |
---|
480 | q = \sqrt{q_x^2+q_y^2} |
---|
481 | |
---|
482 | """ |
---|
483 | if isinstance(qdist, (list, tuple)): |
---|
484 | # Check whether we have a list of ndarrays [qx,qy] |
---|
485 | qx, qy = qdist |
---|
486 | if not self._model_info.parameters.has_2d: |
---|
487 | return self.calculate_Iq(np.sqrt(qx ** 2 + qy ** 2)) |
---|
488 | else: |
---|
489 | return self.calculate_Iq(qx, qy) |
---|
490 | |
---|
491 | elif isinstance(qdist, np.ndarray): |
---|
492 | # We have a simple 1D distribution of q-values |
---|
493 | return self.calculate_Iq(qdist) |
---|
494 | |
---|
495 | else: |
---|
496 | raise TypeError("evalDistribution expects q or [qx, qy], not %r" |
---|
497 | % type(qdist)) |
---|
498 | |
---|
499 | def calculate_Iq(self, qx, qy=None): |
---|
500 | # type: (Sequence[float], Optional[Sequence[float]]) -> np.ndarray |
---|
501 | """ |
---|
502 | Calculate Iq for one set of q with the current parameters. |
---|
503 | |
---|
504 | If the model is 1D, use *q*. If 2D, use *qx*, *qy*. |
---|
505 | |
---|
506 | This should NOT be used for fitting since it copies the *q* vectors |
---|
507 | to the card for each evaluation. |
---|
508 | """ |
---|
509 | if self._model is None: |
---|
510 | self._model = core.build_model(self._model_info) |
---|
511 | if qy is not None: |
---|
512 | q_vectors = [np.asarray(qx), np.asarray(qy)] |
---|
513 | else: |
---|
514 | q_vectors = [np.asarray(qx)] |
---|
515 | calculator = self._model.make_kernel(q_vectors) |
---|
516 | pairs = [self._get_weights(p) |
---|
517 | for p in self._model_info.parameters.call_parameters] |
---|
518 | call_details, value = kernel.build_details(calculator, pairs) |
---|
519 | result = calculator(call_details, value, cutoff=self.cutoff) |
---|
520 | calculator.release() |
---|
521 | return result |
---|
522 | |
---|
523 | def calculate_ER(self): |
---|
524 | # type: () -> float |
---|
525 | """ |
---|
526 | Calculate the effective radius for P(q)*S(q) |
---|
527 | |
---|
528 | :return: the value of the effective radius |
---|
529 | """ |
---|
530 | if self._model_info.ER is None: |
---|
531 | return 1.0 |
---|
532 | else: |
---|
533 | value, weight = self._dispersion_mesh() |
---|
534 | fv = self._model_info.ER(*value) |
---|
535 | #print(values[0].shape, weights.shape, fv.shape) |
---|
536 | return np.sum(weight * fv) / np.sum(weight) |
---|
537 | |
---|
538 | def calculate_VR(self): |
---|
539 | # type: () -> float |
---|
540 | """ |
---|
541 | Calculate the volf ratio for P(q)*S(q) |
---|
542 | |
---|
543 | :return: the value of the volf ratio |
---|
544 | """ |
---|
545 | if self._model_info.VR is None: |
---|
546 | return 1.0 |
---|
547 | else: |
---|
548 | value, weight = self._dispersion_mesh() |
---|
549 | whole, part = self._model_info.VR(*value) |
---|
550 | return np.sum(weight * part) / np.sum(weight * whole) |
---|
551 | |
---|
552 | def set_dispersion(self, parameter, dispersion): |
---|
553 | # type: (str, weights.Dispersion) -> Dict[str, Any] |
---|
554 | """ |
---|
555 | Set the dispersion object for a model parameter |
---|
556 | |
---|
557 | :param parameter: name of the parameter [string] |
---|
558 | :param dispersion: dispersion object of type Dispersion |
---|
559 | """ |
---|
560 | if parameter in self.params: |
---|
561 | # TODO: Store the disperser object directly in the model. |
---|
562 | # The current method of relying on the sasview GUI to |
---|
563 | # remember them is kind of funky. |
---|
564 | # Note: can't seem to get disperser parameters from sasview |
---|
565 | # (1) Could create a sasview model that has not yet # been |
---|
566 | # converted, assign the disperser to one of its polydisperse |
---|
567 | # parameters, then retrieve the disperser parameters from the |
---|
568 | # sasview model. (2) Could write a disperser parameter retriever |
---|
569 | # in sasview. (3) Could modify sasview to use sasmodels.weights |
---|
570 | # dispersers. |
---|
571 | # For now, rely on the fact that the sasview only ever uses |
---|
572 | # new dispersers in the set_dispersion call and create a new |
---|
573 | # one instead of trying to assign parameters. |
---|
574 | from . import weights |
---|
575 | disperser = weights.dispersers[dispersion.__class__.__name__] |
---|
576 | dispersion = weights.MODELS[disperser]() |
---|
577 | self.dispersion[parameter] = dispersion.get_pars() |
---|
578 | else: |
---|
579 | raise ValueError("%r is not a dispersity or orientation parameter") |
---|
580 | |
---|
581 | def _dispersion_mesh(self): |
---|
582 | # type: () -> List[Tuple[np.ndarray, np.ndarray]] |
---|
583 | """ |
---|
584 | Create a mesh grid of dispersion parameters and weights. |
---|
585 | |
---|
586 | Returns [p1,p2,...],w where pj is a vector of values for parameter j |
---|
587 | and w is a vector containing the products for weights for each |
---|
588 | parameter set in the vector. |
---|
589 | """ |
---|
590 | pars = [self._get_weights(p) |
---|
591 | for p in self._model_info.parameters.call_parameters |
---|
592 | if p.type == 'volume'] |
---|
593 | return kernel.dispersion_mesh(self._model_info, pars) |
---|
594 | |
---|
595 | def _get_weights(self, par): |
---|
596 | # type: (Parameter) -> Tuple[np.ndarray, np.ndarray] |
---|
597 | """ |
---|
598 | Return dispersion weights for parameter |
---|
599 | """ |
---|
600 | if par.name not in self.params: |
---|
601 | if par.name == self.multiplicity_info.control: |
---|
602 | return [self.multiplicity], [] |
---|
603 | else: |
---|
604 | return [np.NaN], [] |
---|
605 | elif par.polydisperse: |
---|
606 | dis = self.dispersion[par.name] |
---|
607 | value, weight = weights.get_weights( |
---|
608 | dis['type'], dis['npts'], dis['width'], dis['nsigmas'], |
---|
609 | self.params[par.name], par.limits, par.relative_pd) |
---|
610 | return value, weight / np.sum(weight) |
---|
611 | else: |
---|
612 | return [self.params[par.name]], [] |
---|
613 | |
---|
614 | def test_model(): |
---|
615 | # type: () -> float |
---|
616 | """ |
---|
617 | Test that a sasview model (cylinder) can be run. |
---|
618 | """ |
---|
619 | Cylinder = _make_standard_model('cylinder') |
---|
620 | cylinder = Cylinder() |
---|
621 | return cylinder.evalDistribution([0.1,0.1]) |
---|
622 | |
---|
623 | def test_rpa(): |
---|
624 | # type: () -> float |
---|
625 | """ |
---|
626 | Test that a sasview model (cylinder) can be run. |
---|
627 | """ |
---|
628 | RPA = _make_standard_model('rpa') |
---|
629 | rpa = RPA(3) |
---|
630 | return rpa.evalDistribution([0.1,0.1]) |
---|
631 | |
---|
632 | |
---|
633 | def test_model_list(): |
---|
634 | # type: () -> None |
---|
635 | """ |
---|
636 | Make sure that all models build as sasview models. |
---|
637 | """ |
---|
638 | from .exception import annotate_exception |
---|
639 | for name in core.list_models(): |
---|
640 | try: |
---|
641 | _make_standard_model(name) |
---|
642 | except: |
---|
643 | annotate_exception("when loading "+name) |
---|
644 | raise |
---|
645 | |
---|
646 | if __name__ == "__main__": |
---|
647 | print("cylinder(0.1,0.1)=%g"%test_model()) |
---|