source: sasmodels/sasmodels/product.py @ 353e899

core_shell_microgelsmagnetic_modelticket-1257-vesicle-productticket_1156ticket_1265_superballticket_822_more_unit_tests
Last change on this file since 353e899 was 2d81cfe, checked in by Paul Kienzle <pkienzle@…>, 6 years ago

lint

  • Property mode set to 100644
File size: 13.2 KB
Line 
1"""
2Product model
3-------------
4
5The product model multiplies the structure factor by the form factor,
6modulated by the effective radius of the form.  The resulting model
7has a attributes of both the model description (with parameters, etc.)
8and the module evaluator (with call, release, etc.).
9
10To use it, first load form factor P and structure factor S, then create
11*make_product_info(P, S)*.
12"""
13from __future__ import print_function, division
14
15from copy import copy
16import numpy as np  # type: ignore
17
18from .modelinfo import ParameterTable, ModelInfo
19from .kernel import KernelModel, Kernel
20from .details import make_details, dispersion_mesh
21
22# pylint: disable=unused-import
23try:
24    from typing import Tuple
25except ImportError:
26    pass
27else:
28    from .modelinfo import ParameterSet
29# pylint: enable=unused-import
30
31# TODO: make estimates available to constraints
32#ESTIMATED_PARAMETERS = [
33#    ["est_radius_effective", "A", 0.0, [0, np.inf], "", "Estimated effective radius"],
34#    ["est_volume_ratio", "", 1.0, [0, np.inf], "", "Estimated volume ratio"],
35#]
36
37ER_ID = "radius_effective"
38VF_ID = "volfraction"
39
40# TODO: core_shell_sphere model has suppressed the volume ratio calculation
41# revert it after making VR and ER available at run time as constraints.
42def make_product_info(p_info, s_info):
43    # type: (ModelInfo, ModelInfo) -> ModelInfo
44    """
45    Create info block for product model.
46    """
47    # Make sure effective radius is the first parameter and
48    # make sure volume fraction is the second parameter of the
49    # structure factor calculator.  Structure factors should not
50    # have any magnetic parameters
51    if not len(s_info.parameters.kernel_parameters) >= 2:
52        raise TypeError("S needs {} and {} as its first parameters".format(ER_ID, VF_ID))
53    if not s_info.parameters.kernel_parameters[0].id == ER_ID:
54        raise TypeError("S needs {} as first parameter".format(ER_ID))
55    if not s_info.parameters.kernel_parameters[1].id == VF_ID:
56        raise TypeError("S needs {} as second parameter".format(VF_ID))
57    if not s_info.parameters.magnetism_index == []:
58        raise TypeError("S should not have SLD parameters")
59    p_id, p_name, p_pars = p_info.id, p_info.name, p_info.parameters
60    s_id, s_name, s_pars = s_info.id, s_info.name, s_info.parameters
61
62    # Create list of parameters for the combined model.  Skip the first
63    # parameter of S, which we verified above is effective radius.  If there
64    # are any names in P that overlap with those in S, modify the name in S
65    # to distinguish it.
66    p_set = set(p.id for p in p_pars.kernel_parameters)
67    s_list = [(_tag_parameter(par) if par.id in p_set else par)
68              for par in s_pars.kernel_parameters[1:]]
69    # Check if still a collision after renaming.  This could happen if for
70    # example S has volfrac and P has both volfrac and volfrac_S.
71    if any(p.id in p_set for p in s_list):
72        raise TypeError("name collision: P has P.name and P.name_S while S has S.name")
73
74    translate_name = dict((old.id, new.id) for old, new
75                          in zip(s_pars.kernel_parameters[1:], s_list))
76    combined_pars = p_pars.kernel_parameters + s_list
77    parameters = ParameterTable(combined_pars)
78    parameters.max_pd = p_pars.max_pd + s_pars.max_pd
79    def random():
80        combined_pars = p_info.random()
81        s_names = set(par.id for par in s_pars.kernel_parameters[1:])
82        combined_pars.update((translate_name[k], v)
83                             for k, v in s_info.random().items()
84                             if k in s_names)
85        return combined_pars
86
87    model_info = ModelInfo()
88    model_info.id = '@'.join((p_id, s_id))
89    model_info.name = '@'.join((p_name, s_name))
90    model_info.filename = None
91    model_info.title = 'Product of %s and %s'%(p_name, s_name)
92    model_info.description = model_info.title
93    model_info.docs = model_info.title
94    model_info.category = "custom"
95    model_info.parameters = parameters
96    model_info.random = random
97    #model_info.single = p_info.single and s_info.single
98    model_info.structure_factor = False
99    model_info.variant_info = None
100    #model_info.tests = []
101    #model_info.source = []
102    # Iq, Iqxy, form_volume, ER, VR and sesans
103    # Remember the component info blocks so we can build the model
104    model_info.composition = ('product', [p_info, s_info])
105    model_info.control = p_info.control
106    model_info.hidden = p_info.hidden
107    if getattr(p_info, 'profile', None) is not None:
108        profile_pars = set(p.id for p in p_info.parameters.kernel_parameters)
109        def profile(**kwargs):
110            # extract the profile args
111            kwargs = dict((k, v) for k, v in kwargs.items() if k in profile_pars)
112            return p_info.profile(**kwargs)
113    else:
114        profile = None
115    model_info.profile = profile
116    model_info.profile_axes = p_info.profile_axes
117
118    # TODO: delegate random to p_info, s_info
119    #model_info.random = lambda: {}
120
121    ## Show the parameter table
122    #from .compare import get_pars, parlist
123    #print("==== %s ====="%model_info.name)
124    #values = get_pars(model_info)
125    #print(parlist(model_info, values, is2d=True))
126    return model_info
127
128def _tag_parameter(par):
129    """
130    Tag the parameter name with _S to indicate that the parameter comes from
131    the structure factor parameter set.  This is only necessary if the
132    form factor model includes a parameter of the same name as a parameter
133    in the structure factor.
134    """
135    par = copy(par)
136    # Protect against a vector parameter in S by appending the vector length
137    # to the renamed parameter.  Note: haven't tested this since no existing
138    # structure factor models contain vector parameters.
139    vector_length = par.name[len(par.id):]
140    par.id = par.id + "_S"
141    par.name = par.id + vector_length
142    return par
143
144class ProductModel(KernelModel):
145    def __init__(self, model_info, P, S):
146        # type: (ModelInfo, KernelModel, KernelModel) -> None
147        #: Combined info plock for the product model
148        self.info = model_info
149        #: Form factor modelling individual particles.
150        self.P = P
151        #: Structure factor modelling interaction between particles.
152        self.S = S
153        #: Model precision. This is not really relevant, since it is the
154        #: individual P and S models that control the effective dtype,
155        #: converting the q-vectors to the correct type when the kernels
156        #: for each are created. Ideally this should be set to the more
157        #: precise type to avoid loss of precision, but precision in q is
158        #: not critical (single is good enough for our purposes), so it just
159        #: uses the precision of the form factor.
160        self.dtype = P.dtype  # type: np.dtype
161
162    def make_kernel(self, q_vectors):
163        # type: (List[np.ndarray]) -> Kernel
164        # Note: may be sending the q_vectors to the GPU twice even though they
165        # are only needed once.  It would mess up modularity quite a bit to
166        # handle this optimally, especially since there are many cases where
167        # separate q vectors are needed (e.g., form in python and structure
168        # in opencl; or both in opencl, but one in single precision and the
169        # other in double precision).
170        p_kernel = self.P.make_kernel(q_vectors)
171        s_kernel = self.S.make_kernel(q_vectors)
172        return ProductKernel(self.info, p_kernel, s_kernel)
173
174    def release(self):
175        # type: (None) -> None
176        """
177        Free resources associated with the model.
178        """
179        self.P.release()
180        self.S.release()
181
182
183class ProductKernel(Kernel):
184    def __init__(self, model_info, p_kernel, s_kernel):
185        # type: (ModelInfo, Kernel, Kernel) -> None
186        self.info = model_info
187        self.p_kernel = p_kernel
188        self.s_kernel = s_kernel
189        self.dtype = p_kernel.dtype
190        self.results = []  # type: List[np.ndarray]
191
192    def __call__(self, call_details, values, cutoff, magnetic):
193        # type: (CallDetails, np.ndarray, float, bool) -> np.ndarray
194        p_info, s_info = self.info.composition[1]
195
196        # if there are magnetic parameters, they will only be on the
197        # form factor P, not the structure factor S.
198        nmagnetic = len(self.info.parameters.magnetism_index)
199        if nmagnetic:
200            spin_index = self.info.parameters.npars + 2
201            magnetism = values[spin_index: spin_index+3+3*nmagnetic]
202        else:
203            magnetism = []
204        nvalues = self.info.parameters.nvalues
205        nweights = call_details.num_weights
206        weights = values[nvalues:nvalues + 2*nweights]
207
208        # Construct the calling parameters for P.
209        p_npars = p_info.parameters.npars
210        p_length = call_details.length[:p_npars]
211        p_offset = call_details.offset[:p_npars]
212        p_details = make_details(p_info, p_length, p_offset, nweights)
213        # Set p scale to the volume fraction in s, which is the first of the
214        # 'S' parameters in the parameter list, or 2+np in 0-origin.
215        volfrac = values[2+p_npars]
216        p_values = [[volfrac, 0.0], values[2:2+p_npars], magnetism, weights]
217        spacer = (32 - sum(len(v) for v in p_values)%32)%32
218        p_values.append([0.]*spacer)
219        p_values = np.hstack(p_values).astype(self.p_kernel.dtype)
220
221        # Call ER and VR for P since these are needed for S.
222        p_er, p_vr = calc_er_vr(p_info, p_details, p_values)
223        s_vr = (volfrac/p_vr if p_vr != 0. else volfrac)
224        #print("volfrac:%g p_er:%g p_vr:%g s_vr:%g"%(volfrac,p_er,p_vr,s_vr))
225
226        # Construct the calling parameters for S.
227        # The  effective radius is not in the combined parameter list, so
228        # the number of 'S' parameters is one less than expected.  The
229        # computed effective radius needs to be added into the weights
230        # vector, especially since it is a polydisperse parameter in the
231        # stand-alone structure factor models.  We will added it at the
232        # end so the remaining offsets don't need to change.
233        s_npars = s_info.parameters.npars-1
234        s_length = call_details.length[p_npars:p_npars+s_npars]
235        s_offset = call_details.offset[p_npars:p_npars+s_npars]
236        s_length = np.hstack((1, s_length))
237        s_offset = np.hstack((nweights, s_offset))
238        s_details = make_details(s_info, s_length, s_offset, nweights+1)
239        v, w = weights[:nweights], weights[nweights:]
240        s_values = [
241            # scale=1, background=0, radius_effective=p_er, volfraction=s_vr
242            [1., 0., p_er, s_vr],
243            # structure factor parameters start after scale, background and
244            # all the form factor parameters.  Skip the volfraction parameter
245            # as well, since it is computed elsewhere, and go to the end of the
246            # parameter list.
247            values[2+p_npars+1:2+p_npars+s_npars],
248            # no magnetism parameters to include for S
249            # add er into the (value, weights) pairs
250            v, [p_er], w, [1.0]
251        ]
252        spacer = (32 - sum(len(v) for v in s_values)%32)%32
253        s_values.append([0.]*spacer)
254        s_values = np.hstack(s_values).astype(self.s_kernel.dtype)
255
256        # 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
260        #print("p_npars",p_npars,s_npars,p_er,s_vr,values[2+p_npars+1:2+p_npars+s_npars])
261        #call_details.show(values)
262        #print("values", values)
263        #p_details.show(p_values)
264        #print("=>", p_result)
265        #s_details.show(s_values)
266        #print("=>", s_result)
267
268        # remember the parts for plotting later
269        self.results = [p_result, s_result]
270
271        #import pylab as plt
272        #plt.subplot(211); plt.loglog(self.p_kernel.q_input.q, p_result, '-')
273        #plt.subplot(212); plt.loglog(self.s_kernel.q_input.q, s_result, '-')
274        #plt.figure()
275
276        return values[0]*(p_result*s_result) + values[1]
277
278    def release(self):
279        # type: () -> None
280        self.p_kernel.release()
281        self.s_kernel.release()
282
283
284def calc_er_vr(model_info, call_details, values):
285    # type: (ModelInfo, ParameterSet) -> Tuple[float, float]
286
287    if model_info.ER is None and model_info.VR is None:
288        return 1.0, 1.0
289
290    nvalues = model_info.parameters.nvalues
291    value = values[nvalues:nvalues + call_details.num_weights]
292    weight = values[nvalues + call_details.num_weights: nvalues + 2*call_details.num_weights]
293    npars = model_info.parameters.npars
294    # Note: changed from pairs ([v], [w]) to triples (p, [v], [w]), but the
295    # dispersion mesh code doesn't actually care about the nominal parameter
296    # value, p, so set it to None.
297    pairs = [(None, value[offset:offset+length], weight[offset:offset+length])
298             for p, offset, length
299             in zip(model_info.parameters.call_parameters[2:2+npars],
300                    call_details.offset,
301                    call_details.length)
302             if p.type == 'volume']
303    value, weight = dispersion_mesh(model_info, pairs)
304
305    if model_info.ER is not None:
306        individual_radii = model_info.ER(*value)
307        radius_effective = np.sum(weight*individual_radii) / np.sum(weight)
308    else:
309        radius_effective = 1.0
310
311    if model_info.VR is not None:
312        whole, part = model_info.VR(*value)
313        volume_ratio = np.sum(weight*part)/np.sum(weight*whole)
314    else:
315        volume_ratio = 1.0
316
317    return radius_effective, volume_ratio
Note: See TracBrowser for help on using the repository browser.