[3570545] | 1 | # This program is public domain |
---|
| 2 | """ |
---|
| 3 | Parameters and parameter sets. |
---|
| 4 | |
---|
| 5 | Parameter defines an individual parameter, and ParameterSet groups them |
---|
| 6 | into a hierarchy. |
---|
| 7 | |
---|
| 8 | Individual models need to provide a parameter set with the correct |
---|
| 9 | properties, either by using park.ParameterSet in their model definition, |
---|
| 10 | or by providing a wrapper which can translate assignment to parameter.value |
---|
| 11 | into the appropriate change in the wrapped model. See wrapper.py for |
---|
| 12 | an example. |
---|
| 13 | """ |
---|
| 14 | |
---|
| 15 | __all__ = ['Parameter', 'ParameterSet'] |
---|
| 16 | |
---|
[fb7180c] | 17 | import math |
---|
[3570545] | 18 | import numpy |
---|
| 19 | import expression |
---|
| 20 | |
---|
| 21 | class Pnormal(object): |
---|
| 22 | """ |
---|
| 23 | Negative log likelihood function for a parameter from a Gaussian |
---|
| 24 | distribution. |
---|
| 25 | |
---|
| 26 | Given P(v;mu,sigma) = exp(-1/2 (mu-v)^2/sigma^2)/sqrt(2 pi sigma^2) |
---|
| 27 | then -log(P) = 1/2 (mu-v)^2/sigma^2 + 1/2 log(2*pi*sigma^2) |
---|
| 28 | |
---|
| 29 | Assuming that parameter P is selected from a normal distribution, |
---|
| 30 | then P.likelihood = Pnormal(mu,sigma) |
---|
| 31 | """ |
---|
| 32 | def __init__(self, mean, std): |
---|
| 33 | self.mean = mean |
---|
| 34 | self.std = std |
---|
| 35 | self.const = math.log(2*math.pi*std**2)/2 |
---|
| 36 | def __call__(self, value): |
---|
| 37 | return ((value-self.mean)/self.std)**2/2 + self.const |
---|
| 38 | |
---|
| 39 | inf = numpy.inf |
---|
| 40 | class Parameter(object): |
---|
| 41 | """ |
---|
| 42 | A parameter is a box for communicating with the fitting service. |
---|
| 43 | Parameters can have a number of properties, |
---|
| 44 | |
---|
| 45 | Parameters have a number of properties: |
---|
| 46 | |
---|
| 47 | name "string" |
---|
| 48 | name of the parameter within the parameter set. |
---|
| 49 | |
---|
| 50 | The name is read only. You can rename a parameter but only |
---|
| 51 | in the context of the parameter set which contains it, using |
---|
| 52 | parameterset.rename(par,name). This will change all expressions |
---|
| 53 | containing the named parameter. |
---|
| 54 | |
---|
| 55 | path |
---|
| 56 | dotted name of the parameter within the set of models. The |
---|
| 57 | dotted name is automatically generated by the parameter set |
---|
| 58 | before expressions are parsed and evaluated. There are |
---|
| 59 | some operations on parameter sets (such as renaming the |
---|
| 60 | layer containing a parameter) which will force an adjustment |
---|
| 61 | of all the underlying parameter names, as well as any |
---|
| 62 | expressions in which they are referenced. |
---|
| 63 | |
---|
| 64 | limits (low, high) |
---|
| 65 | hard limits on the range of allowed parameter values, dictated |
---|
| 66 | by physics rather than by knowledge of the particular system. |
---|
| 67 | For example, thickness parameters would have limits (0,inf) |
---|
| 68 | because negative thicknesses are unphysical. These limits |
---|
| 69 | are enforced when setting range for the fit. |
---|
| 70 | |
---|
| 71 | units "string" |
---|
| 72 | units for the parameter. This should be a string, but |
---|
| 73 | be parsable by whatever units package your application |
---|
| 74 | supports. |
---|
| 75 | |
---|
| 76 | tip "string" |
---|
| 77 | parameter description, suitable for use in a tooltip |
---|
| 78 | |
---|
| 79 | value double |
---|
| 80 | current value of the parameter, either fixed, fitted or computed |
---|
| 81 | |
---|
| 82 | range (low, high) |
---|
| 83 | range of expected values for the parameter in the model |
---|
| 84 | |
---|
| 85 | expression "string" |
---|
| 86 | expression for the parameter in the model. This is a string |
---|
| 87 | containing a formula for updating the parameter value based |
---|
| 88 | on the values of other parameters in the system. The expression |
---|
| 89 | is ignored if 'calculated' is False. |
---|
| 90 | |
---|
| 91 | Note: the list of symbols available to the expression evaluator |
---|
| 92 | defaults to the contents of the math module. The caller will be |
---|
| 93 | able to override this within the fitting fitting class. |
---|
| 94 | |
---|
| 95 | status 'fixed'|'computed'|'fitted' |
---|
| 96 | the parameter type. Choose 'fixed' if the values is to |
---|
| 97 | remain fixed throughout the fit, even if a range and an |
---|
| 98 | expression have been specified. Choose 'computed' if the |
---|
| 99 | value of the parameter is to be computed each time the |
---|
| 100 | parameters are updated. Choose 'fitted' if an optimization |
---|
| 101 | algorithm is supposed to search the parameter space. |
---|
| 102 | |
---|
| 103 | likelihood |
---|
| 104 | function to return the negative log likelihood of seeing a |
---|
| 105 | particular parameter value. 2*likelihood(value) will be added |
---|
| 106 | to the total cost function for the particular parameter set |
---|
| 107 | during the fit. This will be on par with the probabilty |
---|
| 108 | of seeing the particular theory function given the observed |
---|
| 109 | datapoints when performing the fit (the residual term is |
---|
| 110 | closely related to the log likelihood of the normal distribution). |
---|
| 111 | |
---|
| 112 | Note: we are minimizing chi^2 = sum [ ((y-f(x;p))/dy)^2 ] rather |
---|
| 113 | than -log P = sum [ ((y-f(x;p))/dy)^2/2 + log(2 pi dy^2) ], |
---|
| 114 | where P is the probability of seeing f(x;p) given y,dy as the |
---|
| 115 | mean and standard deviation of a normal distribution. Because |
---|
| 116 | chi^2_p = - 2 * log P_p + constant, the minima of p are the same |
---|
| 117 | for chi^2 and negative log likelihood. However, to weight the |
---|
| 118 | likelihood properly when adding likelihood values to chisq, we |
---|
| 119 | need the extra factor of 2 mentioned above. The usual statistical |
---|
| 120 | implications of normalized chi^2 will of course be suspect, both |
---|
| 121 | because the assumption of independence between the points in |
---|
| 122 | chi^2 (which definitely do not hold for the new 'point' p_k), and |
---|
| 123 | because of the additional 2 log(2 pi dp_k^2) constant, but given |
---|
| 124 | the uncertainty in the estimate of the distribution parameters, |
---|
| 125 | this is likely a minor point. |
---|
| 126 | """ |
---|
| 127 | # Protect parameter name from user modification |
---|
| 128 | _name = "unknown" |
---|
| 129 | def _getname(self): return self._name |
---|
| 130 | name = property(_getname,doc="parameter name") |
---|
| 131 | |
---|
| 132 | # Do checking on the parameter range to make sure the model stays physical |
---|
| 133 | _range = (-inf,inf) |
---|
| 134 | def _setrange(self, r): |
---|
| 135 | if self.limits[0]<=r[0]<=r[1] <=self.limits[1]: |
---|
| 136 | self._range = r |
---|
| 137 | else: |
---|
| 138 | raise ValueError("invalid range %s for %s"%(r,self.name)) |
---|
| 139 | def _getrange(self): return self._range |
---|
| 140 | range = property(_getrange,_setrange,doc="parameter range") |
---|
| 141 | |
---|
| 142 | path = "" |
---|
| 143 | value = 0. |
---|
| 144 | limits = (-inf,inf) |
---|
| 145 | expression = "" |
---|
| 146 | status = 'fixed' # fixed, computed or fitted |
---|
| 147 | likelihood = None |
---|
| 148 | units = "" |
---|
| 149 | tip = "Fitting parameter" |
---|
| 150 | deriv = False |
---|
| 151 | def __init__(self, name="unknown", **kw): |
---|
| 152 | self._name = name |
---|
| 153 | for k,v in kw.items(): |
---|
| 154 | if hasattr(self,k): setattr(self,k,v) |
---|
| 155 | else: raise AttributeError("Unknown attribute %s"%k) |
---|
| 156 | def __str__(self): return self.path if self.path != '' else self.name |
---|
| 157 | def __repr__(self): return "Parameter('%s')"%self.name |
---|
| 158 | def summarize(self): |
---|
| 159 | """ |
---|
| 160 | Return parameter range string. |
---|
| 161 | |
---|
| 162 | E.g., " Gold .....|.... 5.2043 in [2,7]" |
---|
| 163 | """ |
---|
| 164 | range = ['.']*10 |
---|
[fb7180c] | 165 | lo,hi = self.range |
---|
| 166 | portion = (self.value-lo)/(hi-lo) |
---|
[3570545] | 167 | if portion < 0: portion = 0. |
---|
| 168 | elif portion >= 1: portion = 0.99999999 |
---|
| 169 | bar = math.floor(portion*len(range)) |
---|
| 170 | range[bar] = '|' |
---|
| 171 | range = "".join(range) |
---|
[fb7180c] | 172 | return "%25s %s %g in [%g,%g]" % (self.name,range,self.value,lo,hi) |
---|
[3570545] | 173 | |
---|
| 174 | def isfitted(self): return self.status == 'fitted' |
---|
| 175 | def iscomputed(self): return self.status == 'computed' |
---|
| 176 | def isfixed(self): return self.status == 'fixed' |
---|
| 177 | def isrestrained(self): return self.likelihood is not None |
---|
| 178 | def isfeasible(self, value): |
---|
| 179 | """Return true if the value is in the range""" |
---|
| 180 | return self._range[0] <= value <= self._range[1] |
---|
| 181 | def setprefix(self, prefix): |
---|
| 182 | """ |
---|
| 183 | Set the full path to the parameter as used in expressions involving |
---|
| 184 | the parameter name. |
---|
| 185 | """ |
---|
| 186 | self.path = prefix+self.name |
---|
| 187 | def get(self): |
---|
| 188 | """ |
---|
| 189 | Return the current value for a parameter. |
---|
| 190 | """ |
---|
| 191 | return self.value |
---|
| 192 | def set(self, value): |
---|
| 193 | """ |
---|
| 194 | Set a parameter to a value, a range or an expression. If it is a value, |
---|
| 195 | the parameter will be fixed for the fit. If it is a range, the value |
---|
| 196 | will be varying for the fit. If it is an expression, the parameter will |
---|
| 197 | be calculated from the values of other parameters in the fit. |
---|
| 198 | |
---|
| 199 | Raises ValueError if the value could not be interpreted. |
---|
| 200 | """ |
---|
| 201 | |
---|
| 202 | # Expression |
---|
| 203 | if isinstance(value,basestring): |
---|
| 204 | self.expression = value |
---|
| 205 | self.status = 'computed' |
---|
| 206 | return |
---|
| 207 | |
---|
| 208 | # Fixed value |
---|
| 209 | if numpy.isscalar(value): |
---|
| 210 | self.value = value |
---|
| 211 | self.status = 'fixed' |
---|
| 212 | return |
---|
| 213 | |
---|
| 214 | # Likelihood |
---|
| 215 | if hasattr(value,'__call__'): |
---|
| 216 | self.range = value.range |
---|
| 217 | self.likelihood = value |
---|
| 218 | self.status = 'fitted' |
---|
| 219 | return |
---|
| 220 | |
---|
| 221 | # Range |
---|
| 222 | try: |
---|
| 223 | lo,hi = numpy.asarray(value) |
---|
| 224 | if not numpy.isscalar(lo) or not numpy.isscalar(hi): |
---|
| 225 | raise Exception |
---|
| 226 | self.range = (lo,hi) |
---|
| 227 | self.status = 'fitted' |
---|
| 228 | return |
---|
| 229 | except: |
---|
| 230 | pass |
---|
| 231 | |
---|
| 232 | raise ValueError,\ |
---|
| 233 | "parameter %s expects value, expression or range: %s"%(self.name,value) |
---|
| 234 | |
---|
| 235 | class ParameterSet(list): |
---|
| 236 | """ |
---|
| 237 | The set of parameters used to fit theory to data. |
---|
| 238 | |
---|
| 239 | ParameterSet forms a hierarchy of parameters. The parameters themselves |
---|
| 240 | are referred to by the path through the hierarchy, usually:: |
---|
| 241 | |
---|
| 242 | fitname.component.parameter |
---|
| 243 | |
---|
| 244 | Though more or fewer levels are permitted. Parameters are assumed to |
---|
| 245 | have a unique label throughout the fit. This is required so that |
---|
| 246 | expressions tying the results of one fit to another can uniquely |
---|
| 247 | reference a parameter. |
---|
| 248 | |
---|
| 249 | Attributes: |
---|
| 250 | |
---|
| 251 | name |
---|
| 252 | the name of the parameter set |
---|
| 253 | path |
---|
| 254 | the full dotted name of the parameter set |
---|
| 255 | context |
---|
| 256 | a dictionary providing additional context for evaluating parameters; |
---|
| 257 | Note that this namespace is shared with other theory functions, so |
---|
| 258 | populate it carefully. |
---|
| 259 | """ |
---|
| 260 | # Protect parameter set name from user modification |
---|
| 261 | def _getname(self): return self._name |
---|
| 262 | def _setname(self, name): |
---|
| 263 | raise NotImplementedError("parameter.name is protected; use fit.rename_parameter() to change it") |
---|
| 264 | name = property(_getname,doc="parameter name") |
---|
| 265 | path = "" |
---|
| 266 | def __init__(self, name="unknown", pars=[]): |
---|
| 267 | super(ParameterSet,self).__init__(pars) |
---|
| 268 | self._name = name |
---|
| 269 | self.context = {} |
---|
| 270 | |
---|
| 271 | def _byname(self, parts): |
---|
| 272 | """byname recursion function""" |
---|
| 273 | if len(parts) == 1: return self |
---|
| 274 | for p in self: |
---|
| 275 | if parts[1] == p.name: |
---|
[fb7180c] | 276 | if len(parts) == 2: |
---|
[3570545] | 277 | return p |
---|
| 278 | elif isinstance(p, ParameterSet): |
---|
| 279 | return p._byname(parts[1:]) |
---|
[fb7180c] | 280 | else: |
---|
| 281 | raise |
---|
[3570545] | 282 | return None |
---|
| 283 | |
---|
| 284 | def byname(self, name): |
---|
| 285 | """Lookup parameter from dotted path""" |
---|
| 286 | parts = name.split('.') |
---|
| 287 | if parts[0] == self.name: |
---|
[fb7180c] | 288 | p = self._byname(name.split('.')) |
---|
[3570545] | 289 | if p: return p |
---|
| 290 | raise KeyError("parameter %s not in parameter set"%name) |
---|
| 291 | |
---|
| 292 | def __getitem__(self, idx): |
---|
| 293 | """Allow indexing by name or by number""" |
---|
| 294 | if isinstance(idx,basestring): |
---|
| 295 | for p in self: |
---|
| 296 | if p.name == idx: |
---|
| 297 | return p |
---|
| 298 | raise KeyError("%s is not in the parameter set"%idx) |
---|
| 299 | else: |
---|
| 300 | return super(ParameterSet,self).__getitem__(idx) |
---|
| 301 | |
---|
| 302 | def flatten(self): |
---|
| 303 | """ |
---|
| 304 | Iterate over the elements in depth first order. |
---|
| 305 | """ |
---|
| 306 | import park |
---|
| 307 | L = [] |
---|
| 308 | for p in self: |
---|
| 309 | # Yuck! I really only want to try flattening if p is a |
---|
| 310 | # ParameterSet but it seems we have parameter.ParameterSet |
---|
| 311 | # and park.parameter.ParameterSet as separate entities, |
---|
| 312 | # depending on how things were included. The solution is |
---|
| 313 | # probably to force absolute include paths always. |
---|
| 314 | try: |
---|
| 315 | L += p.flatten() |
---|
| 316 | except: |
---|
| 317 | L.append(p) |
---|
| 318 | return L |
---|
| 319 | """ |
---|
| 320 | # Iterators are cute but too hard to use since you can |
---|
| 321 | # only use them in a [p for p in generator()] once. |
---|
| 322 | for p in self: |
---|
| 323 | if isinstance(p, ParameterSet): |
---|
| 324 | for subp in p.flatten(): |
---|
| 325 | yield subp |
---|
| 326 | else: |
---|
| 327 | yield p |
---|
| 328 | """ |
---|
| 329 | |
---|
| 330 | def _fixed(self): |
---|
| 331 | """ |
---|
| 332 | Return the subset of the parameters which are fixed |
---|
| 333 | """ |
---|
| 334 | return [p for p in self.flatten() if p.isfixed()] |
---|
| 335 | fixed = property(_fixed,doc=_fixed.__doc__) |
---|
| 336 | |
---|
| 337 | def _fitted(self): |
---|
| 338 | """ |
---|
| 339 | Return the subset of the paramters which are varying |
---|
| 340 | """ |
---|
| 341 | return [p for p in self.flatten() if p.isfitted()] |
---|
| 342 | fitted = property(_fitted,doc=_fitted.__doc__) |
---|
| 343 | |
---|
| 344 | def _computed(self): |
---|
| 345 | """ |
---|
| 346 | Return the subset of the parameters which are calculated |
---|
| 347 | """ |
---|
| 348 | return [p for p in self.flatten() if p.iscomputed()] |
---|
| 349 | computed = property(_computed,doc=_computed.__doc__) |
---|
| 350 | |
---|
| 351 | def _restrained(self): |
---|
| 352 | """ |
---|
| 353 | Return the subset of the parameters which have a likelihood |
---|
| 354 | function associated with them. |
---|
| 355 | """ |
---|
| 356 | return [p for p in self.flatten() if p.isrestrained()] |
---|
| 357 | restrained = property(_restrained,doc=_restrained.__doc__) |
---|
| 358 | |
---|
| 359 | def setprefix(self,prefix=None): |
---|
| 360 | """ |
---|
| 361 | Fill in the full path name for all the parameters in the tree. |
---|
| 362 | |
---|
| 363 | Note: this function must be called from the root parameter set |
---|
| 364 | to build proper path names. |
---|
| 365 | |
---|
| 366 | This is required before converting parameter expressions into |
---|
| 367 | function calls. |
---|
| 368 | """ |
---|
| 369 | if prefix == None: |
---|
| 370 | # We are called from root, so we don't have a path |
---|
| 371 | self.path = "" |
---|
| 372 | prefix = "" |
---|
| 373 | else: |
---|
| 374 | self.path = prefix+self.name |
---|
| 375 | prefix = self.path+'.' |
---|
| 376 | for p in self: |
---|
| 377 | #print "setting prefix for",p,prefix |
---|
| 378 | p.setprefix(prefix) |
---|
| 379 | |
---|
| 380 | def rename(self, par, name): |
---|
| 381 | """ |
---|
| 382 | Rename the parameter to something new. |
---|
| 383 | Called from root of the parameter hierarchy, rename the particular |
---|
| 384 | parameter object to something else. |
---|
| 385 | |
---|
| 386 | This changes the internal name of the parameter, as well as all |
---|
| 387 | expressions in which it occurs. If the parameter is actually |
---|
| 388 | a parameter set, then it renames all parameters in the set. |
---|
| 389 | """ |
---|
| 390 | |
---|
| 391 | # Must run from root for self.setprefix and self.computed to work |
---|
| 392 | if self.path != "": |
---|
| 393 | raise RuntimeError,"rename must be called from root parameter set" |
---|
| 394 | |
---|
| 395 | # Change the name of the node |
---|
| 396 | par._name = name |
---|
| 397 | |
---|
| 398 | # Determine which nodes (self and children) are affected |
---|
| 399 | if isinstance(par,ParameterSet): |
---|
| 400 | changeset = par.flatten() |
---|
| 401 | else: |
---|
| 402 | changeset = [par] |
---|
| 403 | |
---|
| 404 | # Map the old names into the new names |
---|
| 405 | old = [p.path for p in changeset] |
---|
| 406 | self.setprefix() # Reset the path names of all parameters |
---|
| 407 | new = [p.path for p in changeset] |
---|
| 408 | mapping = dict(zip(old,new)) |
---|
| 409 | |
---|
| 410 | # Perform the substitution into all of the expressions |
---|
| 411 | exprs = self.computed |
---|
| 412 | for p in exprs: |
---|
| 413 | p.expression = expression.substitute(p.expression, mapping) |
---|
| 414 | |
---|
| 415 | def gather_context(self): |
---|
| 416 | """ |
---|
| 417 | Gather all additional symbols that can be used in expressions. |
---|
| 418 | |
---|
| 419 | For example, if reflectometry provides a volume fraction |
---|
| 420 | function volfrac(rho1,rho2,frac) to compute densities, then |
---|
| 421 | this function can be added as a context dictionary to the |
---|
| 422 | reflectometry parameter set. Note that there is no guarantee |
---|
| 423 | which function will be used if the same function exists in |
---|
| 424 | two separate contexts. |
---|
| 425 | """ |
---|
| 426 | context = {} # Create a new dictionary |
---|
| 427 | context.update(self.context) |
---|
| 428 | for p in self: |
---|
| 429 | if hasattr(p,'gather_context'): |
---|
| 430 | context.update(p.gather_context()) |
---|
| 431 | return context |
---|
| 432 | |
---|
| 433 | |
---|
| 434 | def test(): |
---|
| 435 | # Check single parameter |
---|
| 436 | a = Parameter('a') |
---|
| 437 | assert a.name == 'a' |
---|
| 438 | a.value = 5 |
---|
| 439 | assert a.value == 5 |
---|
| 440 | # Check the setters |
---|
| 441 | a.set(7) |
---|
| 442 | assert a.value == 7 and a.status == 'fixed' and a.isfixed() |
---|
| 443 | a.set([3,5]) |
---|
| 444 | assert a.value == 7 and a.range[0] == 3 and a.range[1]==5 and a.isfitted() |
---|
| 445 | a.set('3*M.b') |
---|
| 446 | assert a.iscomputed() and a.expression == '3*M.b' |
---|
| 447 | |
---|
| 448 | # Check limits |
---|
| 449 | a.limits = (0,inf) |
---|
| 450 | try: |
---|
| 451 | a.range = (-1,1) |
---|
| 452 | raise Exception,"Failed to check range in limits" |
---|
| 453 | except ValueError: pass # Correct failure |
---|
| 454 | |
---|
| 455 | # Check that we can't change name directly |
---|
| 456 | try: |
---|
| 457 | a.name = 'Q' |
---|
| 458 | raise Exception,"Failed to protect name" |
---|
| 459 | except AttributeError: pass # Correct failure |
---|
| 460 | |
---|
| 461 | assert str(a) == 'a' # Before setpath, just print name |
---|
| 462 | a.setprefix('M.') |
---|
| 463 | assert a.path == 'M.a' |
---|
| 464 | assert str(a) == 'M.a' # After setpath, print path |
---|
| 465 | assert a.units == '' |
---|
| 466 | a.units = 'kg' |
---|
| 467 | assert a.units == 'kg' |
---|
| 468 | assert repr(a) == "Parameter('a')" |
---|
| 469 | |
---|
| 470 | # Check parameter set |
---|
| 471 | b,c = Parameter('b'),Parameter('c') |
---|
| 472 | M1 = ParameterSet('M1',[a,b,c]) |
---|
| 473 | assert M1[0] is a |
---|
| 474 | assert M1['a'] is a |
---|
| 475 | a.set(5) # one value |
---|
| 476 | b.set([3,5]) |
---|
| 477 | c.set('3*M1.a') |
---|
| 478 | assert M1.computed == [c] |
---|
| 479 | assert M1.fitted == [b] |
---|
| 480 | assert M1.fixed == [a] |
---|
| 481 | d = Parameter('d') |
---|
| 482 | M1.insert(0,d) |
---|
| 483 | assert M1.fixed == [d,a] |
---|
| 484 | e = Parameter('e') |
---|
| 485 | M1.append(e) |
---|
| 486 | assert M1.fixed == [d,a,e] |
---|
| 487 | |
---|
| 488 | a2,b2,c2 = [Parameter(s) for s in ('a','b','c')] |
---|
| 489 | a2.set(15) |
---|
| 490 | M2 = ParameterSet('M2',[a2,b2,c2]) |
---|
| 491 | # Adjust parameter in set |
---|
| 492 | b2.set([3,5]) |
---|
| 493 | assert M2.fitted == [b2] |
---|
| 494 | |
---|
| 495 | # Hierarchical parameter sets |
---|
| 496 | r = Parameter('r') |
---|
| 497 | root = ParameterSet('root',[M1,r,M2]) |
---|
| 498 | assert root.fixed == [d,a,e,r,a2,c2] |
---|
| 499 | assert root.fitted == [b,b2] |
---|
| 500 | assert root.computed == [c] |
---|
| 501 | root.setprefix() |
---|
| 502 | assert a2.path == "M2.a" |
---|
| 503 | # Rename individual parameter |
---|
| 504 | root.rename(a,'a1') |
---|
| 505 | assert a.path == "M1.a1" |
---|
| 506 | assert c.expression == "3*M1.a1" |
---|
| 507 | # Rename parameter set |
---|
| 508 | root.rename(M1,'m1') |
---|
| 509 | assert c.path == "m1.c" |
---|
| 510 | assert c.expression == "3*m1.a1" |
---|
| 511 | |
---|
| 512 | # Integration test: parameter and expression working together. |
---|
| 513 | fn = expression.build_eval(root.flatten(), root.gather_context()) |
---|
| 514 | #import dis; dis.dis(fn) |
---|
| 515 | fn() |
---|
| 516 | assert c.value == 3*a.value |
---|
| 517 | |
---|
| 518 | # Test context |
---|
| 519 | M2.context['plus2'] = lambda x: x+2 |
---|
| 520 | c2.set('plus2(M2.a)') |
---|
| 521 | assert set(root.computed) == set([c,c2]) |
---|
| 522 | fn = expression.build_eval(root.flatten(), root.gather_context()) |
---|
| 523 | #print dis.dis(fn) |
---|
| 524 | fn() |
---|
| 525 | assert c2.value == a2.value+2 |
---|
| 526 | |
---|
| 527 | # Multilevel hierarchy |
---|
| 528 | # Forming M3.a.x, M3.a.y, M3.b with M3.a.y = 2*M3.b+M3.a.x |
---|
| 529 | x = Parameter('x'); x.set(15) |
---|
| 530 | y = Parameter('y'); y.set('2*M3.b+M3.a.x') |
---|
| 531 | b = Parameter('b'); b.set(10) |
---|
| 532 | a = ParameterSet('a',[x,y]) |
---|
| 533 | M3 = ParameterSet('M3',[a,b]) |
---|
| 534 | root = ParameterSet('root',[M3]) |
---|
| 535 | root.setprefix() |
---|
| 536 | fn = expression.build_eval(root.flatten(), root.gather_context()) |
---|
| 537 | #import dis; dis.dis(fn) |
---|
| 538 | fn() |
---|
| 539 | #print "y-value:",y.value |
---|
| 540 | assert y.value == 35 |
---|
| 541 | |
---|
| 542 | |
---|
| 543 | |
---|
| 544 | |
---|
| 545 | if __name__ == "__main__": test() |
---|