| 1 | |
|---|
| 2 | from sans.models.BaseComponent import BaseComponent |
|---|
| 3 | import numpy, math |
|---|
| 4 | import copy |
|---|
| 5 | from sans.models.pluginmodel import Model1DPlugin |
|---|
| 6 | class MultiplicationModel(BaseComponent): |
|---|
| 7 | """ |
|---|
| 8 | Use for P(Q)*S(Q); function call must be in the order of P(Q) and then S(Q): |
|---|
| 9 | The model parameters are combined from both models, P(Q) and S(Q), except 1) 'effect_radius' of S(Q) |
|---|
| 10 | which will be calculated from P(Q) via calculate_ER(), |
|---|
| 11 | and 2) 'scale' in P model which is synchronized w/ volfraction in S |
|---|
| 12 | then P*S is multiplied by a new param, 'scale_factor'. |
|---|
| 13 | The polydispersion is applicable only to P(Q), not to S(Q). |
|---|
| 14 | Note: P(Q) refers to 'form factor' model while S(Q) does to 'structure factor'. |
|---|
| 15 | """ |
|---|
| 16 | def __init__(self, p_model, s_model ): |
|---|
| 17 | BaseComponent.__init__(self) |
|---|
| 18 | """ |
|---|
| 19 | @param p_model: form factor, P(Q) |
|---|
| 20 | @param s_model: structure factor, S(Q) |
|---|
| 21 | """ |
|---|
| 22 | |
|---|
| 23 | ## Setting model name model description |
|---|
| 24 | self.description="" |
|---|
| 25 | self.name = p_model.name +" * "+ s_model.name |
|---|
| 26 | self.description= self.name+"\n" |
|---|
| 27 | self.fill_description(p_model, s_model) |
|---|
| 28 | |
|---|
| 29 | ## Define parameters |
|---|
| 30 | self.params = {} |
|---|
| 31 | |
|---|
| 32 | ## Parameter details [units, min, max] |
|---|
| 33 | self.details = {} |
|---|
| 34 | |
|---|
| 35 | # non-fittable parameters |
|---|
| 36 | self.non_fittable = p_model.non_fittable |
|---|
| 37 | |
|---|
| 38 | ##models |
|---|
| 39 | self.p_model= p_model |
|---|
| 40 | self.s_model= s_model |
|---|
| 41 | |
|---|
| 42 | |
|---|
| 43 | ## dispersion |
|---|
| 44 | self._set_dispersion() |
|---|
| 45 | ## Define parameters |
|---|
| 46 | self._set_params() |
|---|
| 47 | ## New parameter:Scaling factor |
|---|
| 48 | self.params['scale_factor'] = 1 |
|---|
| 49 | |
|---|
| 50 | ## Parameter details [units, min, max] |
|---|
| 51 | self._set_details() |
|---|
| 52 | self.details['scale_factor'] = ['', None, None] |
|---|
| 53 | |
|---|
| 54 | #list of parameter that can be fitted |
|---|
| 55 | self._set_fixed_params() |
|---|
| 56 | ## parameters with orientation |
|---|
| 57 | for item in self.p_model.orientation_params: |
|---|
| 58 | self.orientation_params.append(item) |
|---|
| 59 | |
|---|
| 60 | for item in self.s_model.orientation_params: |
|---|
| 61 | if not item in self.orientation_params: |
|---|
| 62 | self.orientation_params.append(item) |
|---|
| 63 | # get multiplicity if model provide it, else 1. |
|---|
| 64 | try: |
|---|
| 65 | multiplicity = p_model.multiplicity |
|---|
| 66 | except: |
|---|
| 67 | multiplicity = 1 |
|---|
| 68 | ## functional multiplicity of the model |
|---|
| 69 | self.multiplicity = multiplicity |
|---|
| 70 | |
|---|
| 71 | def _clone(self, obj): |
|---|
| 72 | """ |
|---|
| 73 | Internal utility function to copy the internal |
|---|
| 74 | data members to a fresh copy. |
|---|
| 75 | """ |
|---|
| 76 | obj.params = copy.deepcopy(self.params) |
|---|
| 77 | obj.description = copy.deepcopy(self.description) |
|---|
| 78 | obj.details = copy.deepcopy(self.details) |
|---|
| 79 | obj.dispersion = copy.deepcopy(self.dispersion) |
|---|
| 80 | obj.p_model = self.p_model.clone() |
|---|
| 81 | obj.s_model = self.s_model.clone() |
|---|
| 82 | #obj = copy.deepcopy(self) |
|---|
| 83 | return obj |
|---|
| 84 | |
|---|
| 85 | |
|---|
| 86 | def _set_dispersion(self): |
|---|
| 87 | """ |
|---|
| 88 | combined the two models dispersions |
|---|
| 89 | Polydispersion should not be applied to s_model |
|---|
| 90 | """ |
|---|
| 91 | ##set dispersion only from p_model |
|---|
| 92 | for name , value in self.p_model.dispersion.iteritems(): |
|---|
| 93 | self.dispersion[name]= value |
|---|
| 94 | |
|---|
| 95 | def getProfile(self): |
|---|
| 96 | """ |
|---|
| 97 | Get SLD profile of p_model if exists |
|---|
| 98 | |
|---|
| 99 | : return: (r, beta) where r is a list of radius of the transition points |
|---|
| 100 | beta is a list of the corresponding SLD values |
|---|
| 101 | : Note: This works only for func_shell# = 2 (exp function). |
|---|
| 102 | """ |
|---|
| 103 | try: |
|---|
| 104 | x,y = self.p_model.getProfile() |
|---|
| 105 | except: |
|---|
| 106 | x = None |
|---|
| 107 | y = None |
|---|
| 108 | |
|---|
| 109 | return x, y |
|---|
| 110 | |
|---|
| 111 | def _set_params(self): |
|---|
| 112 | """ |
|---|
| 113 | Concatenate the parameters of the two models to create |
|---|
| 114 | this model parameters |
|---|
| 115 | """ |
|---|
| 116 | |
|---|
| 117 | for name , value in self.p_model.params.iteritems(): |
|---|
| 118 | if not name in self.params.keys() and name != 'scale': |
|---|
| 119 | self.params[name]= value |
|---|
| 120 | |
|---|
| 121 | for name , value in self.s_model.params.iteritems(): |
|---|
| 122 | #Remove the effect_radius from the (P*S) model parameters. |
|---|
| 123 | if not name in self.params.keys() and name != 'effect_radius': |
|---|
| 124 | self.params[name]= value |
|---|
| 125 | |
|---|
| 126 | # Set "scale and effec_radius to P and S model as initializing |
|---|
| 127 | # since run P*S comes from P and S separately. |
|---|
| 128 | self._set_scale_factor() |
|---|
| 129 | self._set_effect_radius() |
|---|
| 130 | |
|---|
| 131 | def _set_details(self): |
|---|
| 132 | """ |
|---|
| 133 | Concatenate details of the two models to create |
|---|
| 134 | this model details |
|---|
| 135 | """ |
|---|
| 136 | for name ,detail in self.p_model.details.iteritems(): |
|---|
| 137 | if name != 'scale': |
|---|
| 138 | self.details[name]= detail |
|---|
| 139 | |
|---|
| 140 | for name , detail in self.s_model.details.iteritems(): |
|---|
| 141 | if not name in self.details.keys() or name != 'effect_radius': |
|---|
| 142 | self.details[name]= detail |
|---|
| 143 | |
|---|
| 144 | def _set_scale_factor(self): |
|---|
| 145 | """ |
|---|
| 146 | Set scale=volfraction to P model |
|---|
| 147 | """ |
|---|
| 148 | value = self.params['volfraction'] |
|---|
| 149 | if value != None: |
|---|
| 150 | self.p_model.setParam( 'scale', value) |
|---|
| 151 | |
|---|
| 152 | |
|---|
| 153 | def _set_effect_radius(self): |
|---|
| 154 | """ |
|---|
| 155 | Set effective radius to S(Q) model |
|---|
| 156 | """ |
|---|
| 157 | effective_radius = self.p_model.calculate_ER() |
|---|
| 158 | #Reset the effective_radius of s_model just before the run |
|---|
| 159 | if effective_radius != None and effective_radius != NotImplemented: |
|---|
| 160 | self.s_model.setParam('effect_radius',effective_radius) |
|---|
| 161 | |
|---|
| 162 | def setParam(self, name, value): |
|---|
| 163 | """ |
|---|
| 164 | Set the value of a model parameter |
|---|
| 165 | |
|---|
| 166 | @param name: name of the parameter |
|---|
| 167 | @param value: value of the parameter |
|---|
| 168 | """ |
|---|
| 169 | # set param to P*S model |
|---|
| 170 | self._setParamHelper( name, value) |
|---|
| 171 | |
|---|
| 172 | ## setParam to p model |
|---|
| 173 | # set 'scale' in P(Q) equal to volfraction |
|---|
| 174 | if name == 'volfraction': |
|---|
| 175 | self._set_scale_factor() |
|---|
| 176 | elif name in self.p_model.getParamList(): |
|---|
| 177 | self.p_model.setParam( name, value) |
|---|
| 178 | |
|---|
| 179 | ## setParam to s model |
|---|
| 180 | # This is a little bit abundant: Todo: find better way |
|---|
| 181 | self._set_effect_radius() |
|---|
| 182 | if name in self.s_model.getParamList(): |
|---|
| 183 | self.s_model.setParam( name, value) |
|---|
| 184 | |
|---|
| 185 | |
|---|
| 186 | #self._setParamHelper( name, value) |
|---|
| 187 | |
|---|
| 188 | def _setParamHelper(self, name, value): |
|---|
| 189 | """ |
|---|
| 190 | Helper function to setparam |
|---|
| 191 | """ |
|---|
| 192 | # Look for dispersion parameters |
|---|
| 193 | toks = name.split('.') |
|---|
| 194 | if len(toks)==2: |
|---|
| 195 | for item in self.dispersion.keys(): |
|---|
| 196 | if item.lower()==toks[0].lower(): |
|---|
| 197 | for par in self.dispersion[item]: |
|---|
| 198 | if par.lower() == toks[1].lower(): |
|---|
| 199 | self.dispersion[item][par] = value |
|---|
| 200 | return |
|---|
| 201 | else: |
|---|
| 202 | # Look for standard parameter |
|---|
| 203 | for item in self.params.keys(): |
|---|
| 204 | if item.lower()==name.lower(): |
|---|
| 205 | self.params[item] = value |
|---|
| 206 | return |
|---|
| 207 | |
|---|
| 208 | raise ValueError, "Model does not contain parameter %s" % name |
|---|
| 209 | |
|---|
| 210 | |
|---|
| 211 | def _set_fixed_params(self): |
|---|
| 212 | """ |
|---|
| 213 | fill the self.fixed list with the p_model fixed list |
|---|
| 214 | """ |
|---|
| 215 | for item in self.p_model.fixed: |
|---|
| 216 | self.fixed.append(item) |
|---|
| 217 | |
|---|
| 218 | self.fixed.sort() |
|---|
| 219 | |
|---|
| 220 | |
|---|
| 221 | def run(self, x = 0.0): |
|---|
| 222 | """ Evaluate the model |
|---|
| 223 | @param x: input q-value (float or [float, float] as [r, theta]) |
|---|
| 224 | @return: (scattering function value) |
|---|
| 225 | """ |
|---|
| 226 | # set effective radius and scaling factor before run |
|---|
| 227 | self._set_effect_radius() |
|---|
| 228 | self._set_scale_factor() |
|---|
| 229 | return self.params['scale_factor']*self.p_model.run(x)*self.s_model.run(x) |
|---|
| 230 | |
|---|
| 231 | def runXY(self, x = 0.0): |
|---|
| 232 | """ Evaluate the model |
|---|
| 233 | @param x: input q-value (float or [float, float] as [qx, qy]) |
|---|
| 234 | @return: scattering function value |
|---|
| 235 | """ |
|---|
| 236 | # set effective radius and scaling factor before run |
|---|
| 237 | self._set_effect_radius() |
|---|
| 238 | self._set_scale_factor() |
|---|
| 239 | return self.params['scale_factor']*self.p_model.runXY(x)* self.s_model.runXY(x) |
|---|
| 240 | |
|---|
| 241 | ## Now (May27,10) directly uses the model eval function |
|---|
| 242 | ## instead of the for-loop in Base Component. |
|---|
| 243 | def evalDistribution(self, x = []): |
|---|
| 244 | """ Evaluate the model in cartesian coordinates |
|---|
| 245 | @param x: input q[], or [qx[], qy[]] |
|---|
| 246 | @return: scattering function P(q[]) |
|---|
| 247 | """ |
|---|
| 248 | # set effective radius and scaling factor before run |
|---|
| 249 | self._set_effect_radius() |
|---|
| 250 | self._set_scale_factor() |
|---|
| 251 | return self.params['scale_factor']*self.p_model.evalDistribution(x)* self.s_model.evalDistribution(x) |
|---|
| 252 | |
|---|
| 253 | def set_dispersion(self, parameter, dispersion): |
|---|
| 254 | """ |
|---|
| 255 | Set the dispersion object for a model parameter |
|---|
| 256 | @param parameter: name of the parameter [string] |
|---|
| 257 | @dispersion: dispersion object of type DispersionModel |
|---|
| 258 | """ |
|---|
| 259 | value= None |
|---|
| 260 | try: |
|---|
| 261 | if parameter in self.p_model.dispersion.keys(): |
|---|
| 262 | value= self.p_model.set_dispersion(parameter, dispersion) |
|---|
| 263 | self._set_dispersion() |
|---|
| 264 | return value |
|---|
| 265 | except: |
|---|
| 266 | raise |
|---|
| 267 | |
|---|
| 268 | def fill_description(self, p_model, s_model): |
|---|
| 269 | """ |
|---|
| 270 | Fill the description for P(Q)*S(Q) |
|---|
| 271 | """ |
|---|
| 272 | description = "" |
|---|
| 273 | description += "Note:1) The effect_radius (effective radius) of %s \n"% (s_model.name) |
|---|
| 274 | description +=" is automatically calculated from size parameters (radius...).\n" |
|---|
| 275 | description += " 2) For non-spherical shape, this approximation is valid \n" |
|---|
| 276 | description += " only for limited systems. Thus, use it at your own risk.\n" |
|---|
| 277 | description +="See %s description and %s description \n"%( p_model.name, s_model.name ) |
|---|
| 278 | description += " for details of individual models." |
|---|
| 279 | self.description += description |
|---|
| 280 | |
|---|
