apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-05-24 14:15:14 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

new getters and setters for self.params, added m['var'] getter and setter

Browse source
This commit is contained in:
Max Zwiessele 2013-04-24 11:16:33 +01:00
parent 2096d062cb
commit 000cd5dbb8
2 changed files with 160 additions and 123 deletions
Download patch file Download diff file Expand all files Collapse all files

31
GPy/core/model.py
View file

@ -84,37 +84,6 @@ class model(parameterised):
for w in which:
self.priors[w] = what
def __getitem__(self, name):
return self.get(name)
def __setitem(self, name, val):
return self.set(name, val)
def get(self, name, return_names=False):
"""
Get a model parameter by name. The name is applied as a regular expression and all parameters that match that regular expression are returned.
"""
matches = self.grep_param_names(name)
if len(matches):
if return_names:
return self._get_params()[matches], np.asarray(self._get_param_names())[matches].tolist()
else:
return self._get_params()[matches]
else:
raise AttributeError, "no parameter matches %s" % name
def set(self, name, val):
"""
Set model parameter(s) by name. The name is provided as a regular expression. All parameters matching that regular expression are set to ghe given value.
"""
matches = self.grep_param_names(name)
if len(matches):
x = self._get_params()
x[matches] = val
self._set_params(x)
else:
raise AttributeError, "no parameter matches %s" % name
def get_gradient(self, name, return_names=False):
"""
Get model gradient(s) by name. The name is applied as a regular expression and all parameters that match that regular expression are returned.

248
GPy/core/parameterised.py
View file

@ -8,24 +8,25 @@ import copy
import cPickle
import os
from ..util.squashers import sigmoid
import warnings
def truncate_pad(string,width,align='m'):
def truncate_pad(string, width, align='m'):
"""
A helper function to make aligned strings for parameterised.__str__
"""
width=max(width,4)
if len(string)>width:
return string[:width-3]+'...'
elif len(string)==width:
width = max(width, 4)
if len(string) > width:
return string[:width - 3] + '...'
elif len(string) == width:
return string
elif len(string)<width:
diff = width-len(string)
if align=='m':
return ' '*np.floor(diff/2.) + string + ' '*np.ceil(diff/2.)
elif align=='l':
return string + ' '*diff
elif align=='r':
return ' '*diff + string
elif len(string) < width:
diff = width - len(string)
if align == 'm':
return ' ' * np.floor(diff / 2.) + string + ' ' * np.ceil(diff / 2.)
elif align == 'l':
return string + ' ' * diff
elif align == 'r':
return ' ' * diff + string
else:
raise ValueError
@ -37,15 +38,15 @@ class parameterised(object):
self.tied_indices = []
self.constrained_fixed_indices = []
self.constrained_fixed_values = []
self.constrained_positive_indices = np.empty(shape=(0,),dtype=np.int64)
self.constrained_negative_indices = np.empty(shape=(0,),dtype=np.int64)
self.constrained_positive_indices = np.empty(shape=(0,), dtype=np.int64)
self.constrained_negative_indices = np.empty(shape=(0,), dtype=np.int64)
self.constrained_bounded_indices = []
self.constrained_bounded_uppers = []
self.constrained_bounded_lowers = []
def pickle(self,filename,protocol=-1):
f = file(filename,'w')
cPickle.dump(self,f,protocol)
def pickle(self, filename, protocol= -1):
f = file(filename, 'w')
cPickle.dump(self, f, protocol)
f.close()
def copy(self):
@ -55,18 +56,85 @@ class parameterised(object):
return copy.deepcopy(self)
@property
def params(self):
"""
Returns a **copy** of parameters in non transformed space
:see_also: :py:func:`GPy.core.parameterised.params_transformed`
"""
return self._get_params()
@params.setter
def params(self, params):
self._set_params(params)
@property
def params_transformed(self):
"""
Returns a **copy** of parameters in transformed space
:see_also: :py:func:`GPy.core.parameterised.params`
"""
return self._get_params_transformed()
@params_transformed.setter
def params_transformed(self, params):
self._set_params_transformed(params)
_get_set_deprecation = """get and set methods wont be available at next minor release
in the next releases you will get and set with following syntax:
Assume m is a model class:
print m['var'] # > prints all parameters matching 'var'
m['var'] = 2. # > sets all parameters matching 'var' to 2.
m['var'] = <array-like> # > sets parameters matching 'var' to <array-like>
"""
def get(self, name):
warnings.warn(self._get_set_deprecation, FutureWarning, stacklevel=2)
return self[name]
def set(self, name, val):
warnings.warn(self._get_set_deprecation, FutureWarning, stacklevel=2)
self[name] = val
def __getitem__(self, name, return_names=False):
"""
Get a model parameter by name. The name is applied as a regular expression and all parameters that match that regular expression are returned.
"""
matches = self.grep_param_names(name)
if len(matches):
if return_names:
return self._get_params()[matches], np.asarray(self._get_param_names())[matches].tolist()
else:
return self._get_params()[matches]
else:
raise AttributeError, "no parameter matches %s" % name
def __setitem__(self, name, val):
"""
Set model parameter(s) by name. The name is provided as a regular expression. All parameters matching that regular expression are set to ghe given value.
"""
matches = self.grep_param_names(name)
if len(matches):
val = np.array(val)
assert (val.size == 1) or val.size == len(matches), "Shape mismatch: {}:({},)".format(val.size, len(matches))
x = self.params
x[matches] = val
self.params = x
# import ipdb;ipdb.set_trace()
# self.params[matches] = val
else:
raise AttributeError, "no parameter matches %s" % name
def tie_params(self, which):
matches = self.grep_param_names(which)
assert matches.size > 0, "need at least something to tie together"
if len(self.tied_indices):
assert not np.any(matches[:,None]==np.hstack(self.tied_indices)), "Some indices are already tied!"
assert not np.any(matches[:, None] == np.hstack(self.tied_indices)), "Some indices are already tied!"
self.tied_indices.append(matches)
#TODO only one of the priors will be evaluated. Give a warning message if the priors are not identical
if hasattr(self,'prior'):
# TODO only one of the priors will be evaluated. Give a warning message if the priors are not identical
if hasattr(self, 'prior'):
pass
self._set_params_transformed(self._get_params_transformed())# sets tied parameters to single value
self._set_params_transformed(self._get_params_transformed()) # sets tied parameters to single value
def untie_everything(self):
"""Unties all parameters by setting tied_indices to an empty list."""
@ -117,44 +185,44 @@ class parameterised(object):
which -- np.array(dtype=int), or regular expression object or string
"""
matches = self.grep_param_names(which)
assert not np.any(matches[:,None]==self.all_constrained_indices()), "Some indices are already constrained"
assert not np.any(matches[:, None] == self.all_constrained_indices()), "Some indices are already constrained"
self.constrained_positive_indices = np.hstack((self.constrained_positive_indices, matches))
#check to ensure constraint is in place
# check to ensure constraint is in place
x = self._get_params()
for i,xx in enumerate(x):
if (xx<0) & (i in matches):
for i, xx in enumerate(x):
if (xx < 0) & (i in matches):
x[i] = -xx
self._set_params(x)
def unconstrain(self,which):
def unconstrain(self, which):
"""Unconstrain matching parameters. does not untie parameters"""
matches = self.grep_param_names(which)
#positive/negative
self.constrained_positive_indices = np.delete(self.constrained_positive_indices,np.nonzero(np.sum(self.constrained_positive_indices[:,None]==matches[None,:],1))[0])
self.constrained_negative_indices = np.delete(self.constrained_negative_indices,np.nonzero(np.sum(self.constrained_negative_indices[:,None]==matches[None,:],1))[0])
#bounded
# positive/negative
self.constrained_positive_indices = np.delete(self.constrained_positive_indices, np.nonzero(np.sum(self.constrained_positive_indices[:, None] == matches[None, :], 1))[0])
self.constrained_negative_indices = np.delete(self.constrained_negative_indices, np.nonzero(np.sum(self.constrained_negative_indices[:, None] == matches[None, :], 1))[0])
# bounded
if len(self.constrained_bounded_indices):
self.constrained_bounded_indices = [np.delete(a,np.nonzero(np.sum(a[:,None]==matches[None,:],1))[0]) for a in self.constrained_bounded_indices]
self.constrained_bounded_indices = [np.delete(a, np.nonzero(np.sum(a[:, None] == matches[None, :], 1))[0]) for a in self.constrained_bounded_indices]
if np.hstack(self.constrained_bounded_indices).size:
self.constrained_bounded_uppers, self.constrained_bounded_lowers, self.constrained_bounded_indices = zip(*[(u,l,i) for u,l,i in zip(self.constrained_bounded_uppers, self.constrained_bounded_lowers, self.constrained_bounded_indices) if i.size])
self.constrained_bounded_uppers, self.constrained_bounded_lowers, self.constrained_bounded_indices = zip(*[(u, l, i) for u, l, i in zip(self.constrained_bounded_uppers, self.constrained_bounded_lowers, self.constrained_bounded_indices) if i.size])
self.constrained_bounded_uppers, self.constrained_bounded_lowers, self.constrained_bounded_indices = list(self.constrained_bounded_uppers), list(self.constrained_bounded_lowers), list(self.constrained_bounded_indices)
else:
self.constrained_bounded_uppers, self.constrained_bounded_lowers, self.constrained_bounded_indices = [],[],[]
#fixed:
for i,indices in enumerate(self.constrained_fixed_indices):
self.constrained_fixed_indices[i] = np.delete(indices,np.nonzero(np.sum(indices[:,None]==matches[None,:],1))[0])
#remove empty elements
tmp = [(i,v) for i,v in zip(self.constrained_fixed_indices, self.constrained_fixed_values) if len(i)]
self.constrained_bounded_uppers, self.constrained_bounded_lowers, self.constrained_bounded_indices = [], [], []
# fixed:
for i, indices in enumerate(self.constrained_fixed_indices):
self.constrained_fixed_indices[i] = np.delete(indices, np.nonzero(np.sum(indices[:, None] == matches[None, :], 1))[0])
# remove empty elements
tmp = [(i, v) for i, v in zip(self.constrained_fixed_indices, self.constrained_fixed_values) if len(i)]
if tmp:
self.constrained_fixed_indices, self.constrained_fixed_values = zip(*tmp)
self.constrained_fixed_indices, self.constrained_fixed_values = list(self.constrained_fixed_indices), list(self.constrained_fixed_values)
else:
self.constrained_fixed_indices, self.constrained_fixed_values = [],[]
self.constrained_fixed_indices, self.constrained_fixed_values = [], []
def constrain_negative(self,which):
def constrain_negative(self, which):
"""
Set negative constraints.
@ -163,12 +231,12 @@ class parameterised(object):
"""
matches = self.grep_param_names(which)
assert not np.any(matches[:,None]==self.all_constrained_indices()), "Some indices are already constrained"
assert not np.any(matches[:, None] == self.all_constrained_indices()), "Some indices are already constrained"
self.constrained_negative_indices = np.hstack((self.constrained_negative_indices, matches))
#check to ensure constraint is in place
# check to ensure constraint is in place
x = self._get_params()
for i,xx in enumerate(x):
if (xx>0.) and (i in matches):
for i, xx in enumerate(x):
if (xx > 0.) and (i in matches):
x[i] = -xx
self._set_params(x)
@ -184,20 +252,20 @@ class parameterised(object):
lower -- (float) the lower bound on the constraint
"""
matches = self.grep_param_names(which)
assert not np.any(matches[:,None]==self.all_constrained_indices()), "Some indices are already constrained"
assert not np.any(matches[:, None] == self.all_constrained_indices()), "Some indices are already constrained"
assert lower < upper, "lower bound must be smaller than upper bound!"
self.constrained_bounded_indices.append(matches)
self.constrained_bounded_uppers.append(upper)
self.constrained_bounded_lowers.append(lower)
#check to ensure constraint is in place
# check to ensure constraint is in place
x = self._get_params()
for i,xx in enumerate(x):
if ((xx<=lower)|(xx>=upper)) & (i in matches):
x[i] = sigmoid(xx)*(upper-lower) + lower
for i, xx in enumerate(x):
if ((xx <= lower) | (xx >= upper)) & (i in matches):
x[i] = sigmoid(xx) * (upper - lower) + lower
self._set_params(x)
def constrain_fixed(self, which, value = None):
def constrain_fixed(self, which, value=None):
"""
Arguments
---------
@ -211,14 +279,14 @@ class parameterised(object):
To fix multiple parameters to the same value, simply pass a regular expression which matches both parameter names, or pass both of the indexes
"""
matches = self.grep_param_names(which)
assert not np.any(matches[:,None]==self.all_constrained_indices()), "Some indices are already constrained"
assert not np.any(matches[:, None] == self.all_constrained_indices()), "Some indices are already constrained"
self.constrained_fixed_indices.append(matches)
if value != None:
self.constrained_fixed_values.append(value)
else:
self.constrained_fixed_values.append(self._get_params()[self.constrained_fixed_indices[-1]])
#self.constrained_fixed_values.append(value)
# self.constrained_fixed_values.append(value)
self._set_params_transformed(self._get_params_transformed())
def _get_params_transformed(self):
@ -226,40 +294,40 @@ class parameterised(object):
x = self._get_params()
x[self.constrained_positive_indices] = np.log(x[self.constrained_positive_indices])
x[self.constrained_negative_indices] = np.log(-x[self.constrained_negative_indices])
[np.put(x,i,np.log(np.clip(x[i]-l,1e-10,np.inf)/np.clip(h-x[i],1e-10,np.inf))) for i,l,h in zip(self.constrained_bounded_indices, self.constrained_bounded_lowers, self.constrained_bounded_uppers)]
[np.put(x, i, np.log(np.clip(x[i] - l, 1e-10, np.inf) / np.clip(h - x[i], 1e-10, np.inf))) for i, l, h in zip(self.constrained_bounded_indices, self.constrained_bounded_lowers, self.constrained_bounded_uppers)]
to_remove = self.constrained_fixed_indices+[t[1:] for t in self.tied_indices]
to_remove = self.constrained_fixed_indices + [t[1:] for t in self.tied_indices]
if len(to_remove):
return np.delete(x,np.hstack(to_remove))
return np.delete(x, np.hstack(to_remove))
else:
return x
def _set_params_transformed(self,x):
def _set_params_transformed(self, x):
""" takes the vector x, which is then modified (by untying, reparameterising or inserting fixed values), and then call self._set_params"""
#work out how many places are fixed, and where they are. tricky logic!
# work out how many places are fixed, and where they are. tricky logic!
Nfix_places = 0.
if len(self.tied_indices):
Nfix_places += np.hstack(self.tied_indices).size-len(self.tied_indices)
Nfix_places += np.hstack(self.tied_indices).size - len(self.tied_indices)
if len(self.constrained_fixed_indices):
Nfix_places += np.hstack(self.constrained_fixed_indices).size
if Nfix_places:
fix_places = np.hstack(self.constrained_fixed_indices+[t[1:] for t in self.tied_indices])
fix_places = np.hstack(self.constrained_fixed_indices + [t[1:] for t in self.tied_indices])
else:
fix_places = []
free_places = np.setdiff1d(np.arange(Nfix_places+x.size,dtype=np.int),fix_places)
free_places = np.setdiff1d(np.arange(Nfix_places + x.size, dtype=np.int), fix_places)
#put the models values in the vector xx
xx = np.zeros(Nfix_places+free_places.size,dtype=np.float64)
# put the models values in the vector xx
xx = np.zeros(Nfix_places + free_places.size, dtype=np.float64)
xx[free_places] = x
[np.put(xx,i,v) for i,v in zip(self.constrained_fixed_indices, self.constrained_fixed_values)]
[np.put(xx,i,v) for i,v in [(t[1:],xx[t[0]]) for t in self.tied_indices] ]
[np.put(xx, i, v) for i, v in zip(self.constrained_fixed_indices, self.constrained_fixed_values)]
[np.put(xx, i, v) for i, v in [(t[1:], xx[t[0]]) for t in self.tied_indices] ]
xx[self.constrained_positive_indices] = np.exp(xx[self.constrained_positive_indices])
xx[self.constrained_negative_indices] = -np.exp(xx[self.constrained_negative_indices])
[np.put(xx,i,low+sigmoid(xx[i])*(high-low)) for i,low,high in zip(self.constrained_bounded_indices, self.constrained_bounded_lowers, self.constrained_bounded_uppers)]
[np.put(xx, i, low + sigmoid(xx[i]) * (high - low)) for i, low, high in zip(self.constrained_bounded_indices, self.constrained_bounded_lowers, self.constrained_bounded_uppers)]
self._set_params(xx)
def _get_param_names_transformed(self):
@ -269,31 +337,31 @@ class parameterised(object):
"""
n = self._get_param_names()
#remove/concatenate the tied parameter names
# remove/concatenate the tied parameter names
if len(self.tied_indices):
for t in self.tied_indices:
n[t[0]] = "<tie>".join([n[tt] for tt in t])
remove = np.hstack([t[1:] for t in self.tied_indices])
else:
remove=np.empty(shape=(0,),dtype=np.int)
remove = np.empty(shape=(0,), dtype=np.int)
#also remove the fixed params
# also remove the fixed params
if len(self.constrained_fixed_indices):
remove = np.hstack((remove, np.hstack(self.constrained_fixed_indices)))
#add markers to show that some variables are constrained
# add markers to show that some variables are constrained
for i in self.constrained_positive_indices:
n[i] = n[i]+'(+ve)'
n[i] = n[i] + '(+ve)'
for i in self.constrained_negative_indices:
n[i] = n[i]+'(-ve)'
for i,l,h in zip(self.constrained_bounded_indices, self.constrained_bounded_lowers, self.constrained_bounded_uppers):
n[i] = n[i] + '(-ve)'
for i, l, h in zip(self.constrained_bounded_indices, self.constrained_bounded_lowers, self.constrained_bounded_uppers):
for ii in i:
n[ii] = n[ii]+'(bounded)'
n[ii] = n[ii] + '(bounded)'
n = [nn for i,nn in enumerate(n) if not i in remove]
n = [nn for i, nn in enumerate(n) if not i in remove]
return n
def __str__(self,nw=30):
def __str__(self, nw=30):
"""
Return a string describing the parameter names and their ties and constraints
"""
@ -302,10 +370,10 @@ class parameterised(object):
if not N:
return "This object has no free parameters."
header = ['Name','Value','Constraints','Ties']
values = self._get_params() #map(str,self._get_params())
#sort out the constraints
constraints = ['']*len(names)
header = ['Name', 'Value', 'Constraints', 'Ties']
values = self._get_params() # map(str,self._get_params())
# sort out the constraints
constraints = [''] * len(names)
for i in self.constrained_positive_indices:
constraints[i] = '(+ve)'
for i in self.constrained_negative_indices:
@ -313,14 +381,14 @@ class parameterised(object):
for i in self.constrained_fixed_indices:
for ii in i:
constraints[ii] = 'Fixed'
for i,u,l in zip(self.constrained_bounded_indices, self.constrained_bounded_uppers, self.constrained_bounded_lowers):
for i, u, l in zip(self.constrained_bounded_indices, self.constrained_bounded_uppers, self.constrained_bounded_lowers):
for ii in i:
constraints[ii] = '('+str(l)+', '+str(u)+')'
#sort out the ties
ties = ['']*len(names)
for i,tie in enumerate(self.tied_indices):
constraints[ii] = '(' + str(l) + ', ' + str(u) + ')'
# sort out the ties
ties = [''] * len(names)
for i, tie in enumerate(self.tied_indices):
for j in tie:
ties[j] = '('+str(i)+')'
ties[j] = '(' + str(i) + ')'
values = ['%.4f' % float(v) for v in values]
max_names = max([len(names[i]) for i in range(len(names))] + [len(header[0])])
@ -330,10 +398,10 @@ class parameterised(object):
cols = np.array([max_names, max_values, max_constraint, max_ties]) + 4
columns = cols.sum()
header_string = ["{h:^{col}}".format(h = header[i], col = cols[i]) for i in range(len(cols))]
header_string = ["{h:^{col}}".format(h=header[i], col=cols[i]) for i in range(len(cols))]
header_string = map(lambda x: '|'.join(x), [header_string])
separator = '-'*len(header_string[0])
param_string = ["{n:^{c0}}|{v:^{c1}}|{c:^{c2}}|{t:^{c3}}".format(n = names[i], v = values[i], c = constraints[i], t = ties[i], c0 = cols[0], c1 = cols[1], c2 = cols[2], c3 = cols[3]) for i in range(len(values))]
separator = '-' * len(header_string[0])
param_string = ["{n:^{c0}}|{v:^{c1}}|{c:^{c2}}|{t:^{c3}}".format(n=names[i], v=values[i], c=constraints[i], t=ties[i], c0=cols[0], c1=cols[1], c2=cols[2], c3=cols[3]) for i in range(len(values))]
return ('\n'.join([header_string[0], separator]+param_string)) + '\n'
return ('\n'.join([header_string[0], separator] + param_string)) + '\n'