def make_node(self, x, repeats): x = basic.as_tensor_variable(x) repeats = basic.as_tensor_variable(repeats) if repeats.dtype not in tensor.discrete_dtypes: raise TypeError("repeats.dtype must be an integer.") # Some dtypes are not supported by numpy's implementation of repeat. # Until another one is available, we should fail at graph construction # time, not wait for execution. int_bitwidth = theano.gof.cmodule.python_int_bitwidth() if int_bitwidth == 64: numpy_unsupported_dtypes = ('uint64', ) if int_bitwidth == 32: numpy_unsupported_dtypes = ('uint32', 'int64', 'uint64') if repeats.dtype in numpy_unsupported_dtypes: raise TypeError( ("dtypes %s are not supported by numpy.repeat " "for the 'repeats' parameter, " % numpy_unsupported_dtypes), repeats.dtype) if self.axis is None: out_type = theano.tensor.TensorType(dtype=x.dtype, broadcastable=[False]) else: out_type = x.type return theano.Apply(self, [x, repeats], [out_type()])
def make_node(self, x, repeats): x = basic.as_tensor_variable(x) repeats = basic.as_tensor_variable(repeats) if repeats.dtype not in tensor.discrete_dtypes: raise TypeError("repeats.dtype must be an integer.") # Some dtypes are not supported by numpy's implementation of repeat. # Until another one is available, we should fail at graph construction # time, not wait for execution. int_bitwidth = theano.gof.cmodule.python_int_bitwidth() if int_bitwidth == 64: numpy_unsupported_dtypes = ('uint64',) if int_bitwidth == 32: numpy_unsupported_dtypes = ('uint32', 'int64', 'uint64') if repeats.dtype in numpy_unsupported_dtypes: raise TypeError( ("dtypes %s are not supported by numpy.repeat " "for the 'repeats' parameter, " % numpy_unsupported_dtypes), repeats.dtype) if self.axis is None: out_type = theano.tensor.TensorType(dtype=x.dtype, broadcastable=[False]) else: out_type = x.type return theano.Apply(self, [x, repeats], [out_type()])
def binomial(random_state, size=None, n=1, p=0.5, ndim=None, dtype='int64', prob=None): """ Sample n times with probability of success prob for each trial, return the number of successes. If the size argument is ambiguous on the number of dimensions, ndim may be a plain integer to supplement the missing information. If size is None, the output shape will be determined by the shapes of n and prob. """ if prob is not None: p = prob print >> sys.stderr, "DEPRECATION WARNING: the parameter prob to the binomal fct have been renamed to p to have the same name as numpy." n = tensor.as_tensor_variable(n) p = tensor.as_tensor_variable(p) ndim, size, bcast = _infer_ndim_bcast(ndim, size, n, p) if n.dtype == 'int64': ### THIS WORKS AROUND A NUMPY BUG on 32bit machine ### Erase when the following works on a 32bit machine: ### numpy.random.binomial( # n=numpy.asarray([2,3,4], dtype='int64'), # p=numpy.asarray([.1, .2, .3], dtype='float64')) n = tensor.cast(n, 'int32') op = RandomFunction( 'binomial', tensor.TensorType(dtype=dtype, broadcastable=(False, ) * ndim)) return op(random_state, size, n, p)
def binomial(random_state, size=None, n=1, p=0.5, ndim=None, dtype='int64', prob=None): """ Sample n times with probability of success prob for each trial, return the number of successes. If the size argument is ambiguous on the number of dimensions, ndim may be a plain integer to supplement the missing information. If size is None, the output shape will be determined by the shapes of n and prob. """ if prob is not None: p = prob print >> sys.stderr, "DEPRECATION WARNING: the parameter prob to the binomal fct have been renamed to p to have the same name as numpy." n = tensor.as_tensor_variable(n) p = tensor.as_tensor_variable(p) ndim, size, bcast = _infer_ndim_bcast(ndim, size, n, p) if n.dtype=='int64': ### THIS WORKS AROUND A NUMPY BUG on 32bit machine ### Erase when the following works on a 32bit machine: ### numpy.random.binomial( # n=numpy.asarray([2,3,4], dtype='int64'), # p=numpy.asarray([.1, .2, .3], dtype='float64')) n = tensor.cast(n, 'int32') op = RandomFunction('binomial', tensor.TensorType(dtype = dtype, broadcastable = (False,)*ndim) ) return op(random_state, size, n, p)
def make_node(self, x, repeats): x = basic.as_tensor_variable(x) repeats = basic.as_tensor_variable(repeats) if self.axis == None: out_type = theano.tensor.TensorType(dtype=x.dtype, broadcastable=[False]) else: out_type = x.type return theano.Apply(self, [x, repeats], [out_type()])
def make_node(self, a, val): a = tensor.as_tensor_variable(a) val = tensor.as_tensor_variable(val) if a.ndim < 2: raise TypeError('%s: first parameter must have at least' ' two dimensions' % self.__class__.__name__) elif val.ndim != 0: raise TypeError('%s: second parameter must be a scalar' % self.__class__.__name__) val = tensor.cast(val, dtype=scalar.upcast(a.dtype, val.dtype)) if val.dtype != a.dtype: raise TypeError('%s: type of second parameter must be compatible' ' with first\'s' % self.__class__.__name__) return gof.Apply(self, [a, val], [a.type()])
def random_integers(random_state, size=None, low=0, high=1, ndim=None, dtype='int64'): """ Sample a random integer between low and high, both inclusive. If the size argument is ambiguous on the number of dimensions, ndim may be a plain integer to supplement the missing information. If size is None, the output shape will be determined by the shapes of low and high. """ low = tensor.as_tensor_variable(low) high = tensor.as_tensor_variable(high) ndim, size, bcast = _infer_ndim_bcast(ndim, size, low, high) op = RandomFunction(random_integers_helper, tensor.TensorType(dtype=dtype, broadcastable=bcast)) return op(random_state, size, low, high)
def make_node(self, x, weights): x = basic.as_tensor_variable(x) if x.dtype not in BinCountOp.compatible_type: raise TypeError("Inputs dtype must be an integer.") if x.ndim != 1: raise TypeError("Inputs must be of dimension 1.") if weights is None: weights = theano.gof.Constant(theano.gof.Generic(), None) out_type = x.type() else: weights = basic.as_tensor_variable(weights) out_type = weights.type() if weights.ndim != 1: raise TypeError("Weights cannot have a number of" "dimension different of 1.") return theano.Apply(self, [x, weights], [out_type])
def make_node(self, M): M = tensor.as_tensor_variable(M) if M.ndim != 0: raise TypeError('%s only works on scalar input' % self.__class__.__name__) elif (not M.dtype.startswith('int')) and \ (not M.dtype.startswith('uint')): # dtype is a theano attribute here raise TypeError('%s only works on integer input' % self.__class__.__name__) return gof.Apply(self, [M], [tensor.dvector()])
def uniform(random_state, size=None, low=0.0, high=1.0, ndim=None, dtype=None): """ Sample from a uniform distribution between low and high. If the size argument is ambiguous on the number of dimensions, ndim may be a plain integer to supplement the missing information. If size is None, the output shape will be determined by the shapes of low and high. If dtype is not specified, it will be inferred from the dtype of low and high, but will be at least as precise as floatX. """ low = tensor.as_tensor_variable(low) high = tensor.as_tensor_variable(high) if dtype is None: dtype = tensor.scal.upcast(theano.config.floatX, low.dtype, high.dtype) ndim, size, bcast = _infer_ndim_bcast(ndim, size, low, high) op = RandomFunction('uniform', tensor.TensorType(dtype=dtype, broadcastable=bcast)) return op(random_state, size, low, high)
def normal(random_state, size=None, avg=0.0, std=1.0, ndim=None, dtype=None): """ Sample from a normal distribution centered on avg with the specified standard deviation (std). If the size argument is ambiguous on the number of dimensions, ndim may be a plain integer to supplement the missing information. If size is None, the output shape will be determined by the shapes of avg and std. If dtype is not specified, it will be inferred from the dtype of avg and std, but will be at least as precise as floatX. """ avg = tensor.as_tensor_variable(avg) std = tensor.as_tensor_variable(std) if dtype is None: dtype = tensor.scal.upcast(theano.config.floatX, avg.dtype, std.dtype) ndim, size, bcast = _infer_ndim_bcast(ndim, size, avg, std) op = RandomFunction('normal', tensor.TensorType(dtype=dtype, broadcastable=bcast)) return op(random_state, size, avg, std)
def normal(random_state, size=None, avg=0.0, std=1.0, ndim=None, dtype=None): """ Sample from a normal distribution centered on avg with the specified standard deviation (std). If the size argument is ambiguous on the number of dimensions, ndim may be a plain integer to supplement the missing information. If size is None, the output shape will be determined by the shapes of avg and std. If dtype is not specified, it will be inferred from the dtype of avg and std, but will be at least as precise as floatX. """ avg = tensor.as_tensor_variable(avg) std = tensor.as_tensor_variable(std) if dtype == None: dtype = tensor.scal.upcast(theano.config.floatX, avg.dtype, std.dtype) ndim, size, bcast = _infer_ndim_bcast(ndim, size, avg, std) op = RandomFunction('normal', tensor.TensorType(dtype=dtype, broadcastable=bcast)) return op(random_state, size, avg, std)
def make_node(self, x, weights): x = basic.as_tensor_variable(x) if x.dtype not in BinCountOp.compatible_type: raise TypeError("Inputs dtype must be an integer.") # Some dtypes are not supported by numpy's implementation of bincount. # Until another one is available, we should fail at graph construction # time, not wait for execution. int_bitwidth = theano.gof.cmodule.python_int_bitwidth() if int_bitwidth == 64: numpy_unsupported_dtypes = ('uint64',) if int_bitwidth == 32: numpy_unsupported_dtypes = ('uint32', 'int64', 'uint64') intp_bitwidth = theano.gof.cmodule.local_bitwidth() if intp_bitwidth == 32: out_type = basic.ivector() elif intp_bitwidth == 64: out_type = basic.lvector() if x.dtype in numpy_unsupported_dtypes: raise TypeError( ("Input dtypes %s are not supported by numpy.bincount, " % numpy_unsupported_dtypes), x.dtype) if x.ndim != 1: raise TypeError("Inputs must be of dimension 1.") if weights is None: weights = theano.gof.Constant(theano.gof.Generic(), None) else: weights = basic.as_tensor_variable(weights) out_type = basic.dvector() if weights.ndim != 1: raise TypeError("Weights cannot have a number of" "dimension different of 1.") return theano.Apply(self, [x, weights], [out_type])
def make_node(self, x, weights): x = basic.as_tensor_variable(x) if x.dtype not in BinCountOp.compatible_type: raise TypeError("Inputs dtype must be an integer.") # Some dtypes are not supported by numpy's implementation of bincount. # Until another one is available, we should fail at graph construction # time, not wait for execution. int_bitwidth = theano.gof.cmodule.python_int_bitwidth() if int_bitwidth == 64: numpy_unsupported_dtypes = ('uint64', ) if int_bitwidth == 32: numpy_unsupported_dtypes = ('uint32', 'int64', 'uint64') intp_bitwidth = theano.gof.cmodule.local_bitwidth() if intp_bitwidth == 32: out_type = basic.ivector() elif intp_bitwidth == 64: out_type = basic.lvector() if x.dtype in numpy_unsupported_dtypes: raise TypeError( ("Input dtypes %s are not supported by numpy.bincount, " % numpy_unsupported_dtypes), x.dtype) if x.ndim != 1: raise TypeError("Inputs must be of dimension 1.") if weights is None: weights = theano.gof.Constant(theano.gof.Generic(), None) else: weights = basic.as_tensor_variable(weights) out_type = basic.dvector() if weights.ndim != 1: raise TypeError("Weights cannot have a number of" "dimension different of 1.") return theano.Apply(self, [x, weights], [out_type])
def make_node(self, r, shape, *args): """ :param r: a numpy.RandomState instance, or a Variable of Type RandomStateType that will contain a RandomState instance. :param shape: an lvector with a shape defining how many samples to draw. In the case of scalar distributions, it is the shape of the tensor output by this Op. In that case, at runtime, the value associated with this lvector must have a length equal to the number of dimensions promised by `self.outtype`. In a more general case, the number of output dimensions, len(self.outtype), is equal to len(shape)+self.ndim_added. The special case where len(shape) == 0 means that the smallest shape compatible with the argument's shape will be used. :param args: the values associated with these variables will be passed to the RandomState function during perform as extra "*args"-style arguments. These should be castable to variables of Type TensorType. :rtype: Apply :return: Apply with two outputs. The first output is a gof.generic Variable from which to draw further random numbers. The second output is the outtype() instance holding the random draw. """ shape = tensor.as_tensor_variable(shape, ndim=1) assert shape.type.ndim == 1 assert (shape.type.dtype == 'int64') or (shape.type.dtype == 'int32') if not isinstance(r.type, RandomStateType): print >> sys.stderr, 'WARNING: RandomState instances should be in RandomStateType' if 0: raise TypeError('r must be RandomStateType instance', r) # the following doesn't work because we want to ignore the broadcastable flags in # shape.type # assert shape.type == tensor.lvector # convert args to TensorType instances # and append enough None's to match the length of self.args args = map(tensor.as_tensor_variable, args) return gof.Apply(self, [r, shape] + args, [r.type(), self.outtype()])
def make_node(self, x): x = basic.as_tensor_variable(x) return theano.Apply(self, [x], [x.type()])
def multinomial(random_state, size=None, n=1, pvals=[0.5, 0.5], ndim=None, dtype='int64'): """Sample from one or more multinomial distributions defined by one-dimensional slices in pvals. :param pvals: a tensor of shape "nmulti+(L,)" describing each multinomial distribution. This tensor must have the property that numpy.allclose(pvals.sum(axis=-1), 1) is true. :param size: a vector of shape information for the output; this can also specify the "nmulti" part of pvals' shape. A -1 in the k'th position from the right means to borrow the k'th position from the right in nmulti. (See examples below.) Default ``None`` means size=nmulti. :param n: the number of experiments to simulate for each multinomial. This can be a scalar, or tensor, it will be broadcasted to have shape "nmulti". :param dtype: the dtype of the return value (which will represent counts) :returns: tensor of len(size)+1 dimensions, and shape[-1]==L, with the specified ``dtype``, with the experiment counts. See examples to understand the shape of the return value, which is derived from both size and pvals.shape. In return value rval, "numpy.allclose(rval.sum(axis=-1), n)" will be true. For example, to simulate n experiments from each multinomial in a batch of size B: size=None, pvals.shape=(B,L) --> rval.shape=[B,L] rval[i,j] is the count of possibility j in the i'th distribution (row) in pvals. Using size: size=(1,-1), pvals.shape=(A,B,L) --> rval.shape=[1,B,L], and requires that A==1. rval[k,i,j] is the count of possibility j in the distribution specified by pvals[k,i]. Using size for broadcasting of pvals: size=(10, 1, -1), pvals.shape=(A, B, L) --> rval.shape=[10,1,B,L], and requires that A==1. rval[l,k,i,j] is the count of possibility j in the distribution specified by pvals[k,i], in the l'th of 10 draws. """ n = tensor.as_tensor_variable(n) pvals = tensor.as_tensor_variable(pvals) # until ellipsis is implemented (argh) tmp = pvals.T[0].T ndim, size, bcast = _infer_ndim_bcast(ndim, size, n, tmp) bcast = bcast + (pvals.type.broadcastable[-1], ) op = RandomFunction(multinomial_helper, tensor.TensorType(dtype=dtype, broadcastable=bcast), ndim_added=1) return op(random_state, size, n, pvals)
def make_node(self, x): x = basic.as_tensor_variable(x) out_type = theano.tensor.TensorType(dtype=x.dtype, broadcastable=[False] * self.out_nd) return theano.Apply(self, [x], [out_type()])
def multinomial(random_state, size=None, n=1, pvals=[0.5, 0.5], ndim=None, dtype='int64'): """ Sample from one or more multinomial distributions defined by one-dimensional slices in pvals. :param pvals: a tensor of shape "nmulti+(L,)" describing each multinomial distribution. This tensor must have the property that numpy.allclose(pvals.sum(axis=-1), 1) is true. :param size: a vector of shape information for the output; this can also specify the "nmulti" part of pvals' shape. A -1 in the k'th position from the right means to borrow the k'th position from the right in nmulti. (See examples below.) Default ``None`` means size=nmulti. :param n: the number of experiments to simulate for each multinomial. This can be a scalar, or tensor, it will be broadcasted to have shape "nmulti". :param dtype: the dtype of the return value (which will represent counts) :returns: tensor of len(size)+1 dimensions, and shape[-1]==L, with the specified ``dtype``, with the experiment counts. See examples to understand the shape of the return value, which is derived from both size and pvals.shape. In return value rval, "numpy.allclose(rval.sum(axis=-1), n)" will be true. For example, to simulate n experiments from each multinomial in a batch of size B: size=None, pvals.shape=(B,L) --> rval.shape=[B,L] rval[i,j] is the count of possibility j in the i'th distribution (row) in pvals. Using size: size=(1,-1), pvals.shape=(A,B,L) --> rval.shape=[1,B,L], and requires that A==1. rval[k,i,j] is the count of possibility j in the distribution specified by pvals[k,i]. Using size for broadcasting of pvals: size=(10,1,-1), pvals.shape=(A,B,L) --> rval.shape=[10,1,B,L], and requires that A==1. rval[l,k,i,j] is the count of possibility j in the distribution specified by pvals[k,i], in the l'th of 10 draws. """ n = tensor.as_tensor_variable(n) pvals = tensor.as_tensor_variable(pvals) # until ellipsis is implemented (argh) tmp = pvals.T[0].T ndim, size, bcast = _infer_ndim_bcast(ndim, size, n, tmp) bcast = bcast+(pvals.type.broadcastable[-1],) op = RandomFunction(multinomial_helper, tensor.TensorType(dtype = dtype, broadcastable = bcast), ndim_added=1) return op(random_state, size, n, pvals)
def _infer_ndim_bcast(ndim, shape, *args): """ Infer the number of dimensions from the shape or the other arguments. :rtype: (int, variable, tuple) triple, where the variable is an integer vector, and the tuple contains Booleans. :returns: the first element returned is the inferred number of dimensions. The second element is the shape inferred (combining symbolic and constant informations from shape and args). The third element is a broadcasting pattern corresponding to that shape. """ # Find the minimum value of ndim required by the *args if args: args_ndim = max(arg.ndim for arg in args) else: args_ndim = 0 # there is a convention that -1 means the corresponding shape of a # potentially-broadcasted symbolic arg if (isinstance(shape, (tuple, list)) and numpy.all(numpy.asarray(shape)>=0)): bcast = [(s==1) for s in shape] v_shape = tensor.TensorConstant(type=tensor.lvector, data=theano._asarray(shape, dtype='int64')) shape_ndim = len(shape) if ndim is None: ndim = shape_ndim else: if shape_ndim != ndim: raise ValueError('ndim should be equal to len(shape), but\n', 'ndim = %s, len(shape) = %s, shape = %s' % (ndim, shape_ndim, shape)) elif isinstance(shape, (tuple, list)): # there is a convention that -1 means the corresponding shape of a # potentially-broadcasted symbolic arg # # This case combines together symbolic and non-symbolic shape # information if ndim is None: ndim=args_ndim else: ndim = max(args_ndim, ndim) ndim = max(args_ndim, len(shape)) shape = [-1]*(ndim - len(shape))+list(shape) bcast = [] pre_v_shape = [] for i,s in enumerate(shape): if hasattr(s, 'type'): # s is symbolic bcast.append(False) # todo - introspect further pre_v_shape.append(s) else: if s >= 0: pre_v_shape.append(tensor.as_tensor_variable(s)) bcast.append((s==1)) elif s == -1: n_a_i = 0 for a in args: # ndim: _ _ _ _ _ _ # ashp: s0 s1 s2 s3 # i if i >= ndim - a.ndim: n_a_i += 1 a_i = i + a.ndim -ndim if not a.broadcastable[a_i]: pre_v_shape.append(a.shape[a_i]) bcast.append(False) break else: if n_a_i == 0: raise ValueError(('Auto-shape of -1 must overlap' 'with the shape of one of the broadcastable' 'inputs')) else: pre_v_shape.append(tensor.as_tensor_variable(1)) bcast.append(True) else: ValueError('negative shape', s) # post-condition: shape may still contain both symbolic and non-symbolic things v_shape = tensor.stack(*pre_v_shape) elif shape is None: # The number of drawn samples will be determined automatically, # but we need to know ndim if not args: raise TypeError(('_infer_ndim_bcast cannot infer shape without' ' either shape or args')) template = reduce(lambda a,b:a+b, args) v_shape = template.shape bcast = template.broadcastable ndim = template.ndim else: v_shape = tensor.as_tensor_variable(shape) if ndim is None: ndim = tensor.get_vector_length(v_shape) bcast = [False]*ndim if not (v_shape.dtype.startswith('int') or v_shape.dtype.startswith('uint')): raise TypeError('shape must be an integer vector or list', v_shape.dtype) if args_ndim > ndim: raise ValueError('ndim should be at least as big as required by args value', (ndim, args_ndim), args) assert ndim == len(bcast) return ndim, tensor.cast(v_shape, 'int32'), tuple(bcast)
def _infer_ndim_bcast(ndim, shape, *args): """ Infer the number of dimensions from the shape or the other arguments. :rtype: (int, variable, tuple) triple, where the variable is an integer vector, and the tuple contains Booleans. :returns: the first element returned is the inferred number of dimensions. The second element is the shape inferred (combining symbolic and constant informations from shape and args). The third element is a broadcasting pattern corresponding to that shape. """ # Find the minimum value of ndim required by the *args if args: args_ndim = max(arg.ndim for arg in args) else: args_ndim = 0 # there is a convention that -1 means the corresponding shape of a # potentially-broadcasted symbolic arg if (isinstance(shape, (tuple, list)) and numpy.all(numpy.asarray(shape) >= 0)): bcast = [(s == 1) for s in shape] v_shape = tensor.TensorConstant(type=tensor.lvector, data=theano._asarray(shape, dtype='int64')) shape_ndim = len(shape) if ndim is None: ndim = shape_ndim else: if shape_ndim != ndim: raise ValueError( 'ndim should be equal to len(shape), but\n', 'ndim = %s, len(shape) = %s, shape = %s' % (ndim, shape_ndim, shape)) elif isinstance(shape, (tuple, list)): # there is a convention that -1 means the corresponding shape of a # potentially-broadcasted symbolic arg # # This case combines together symbolic and non-symbolic shape # information if ndim is None: ndim = args_ndim else: ndim = max(args_ndim, ndim) ndim = max(args_ndim, len(shape)) shape = [-1] * (ndim - len(shape)) + list(shape) bcast = [] pre_v_shape = [] for i, s in enumerate(shape): if hasattr(s, 'type'): # s is symbolic bcast.append(False) # todo - introspect further pre_v_shape.append(s) else: if s >= 0: pre_v_shape.append(tensor.as_tensor_variable(s)) bcast.append((s == 1)) elif s == -1: n_a_i = 0 for a in args: # ndim: _ _ _ _ _ _ # ashp: s0 s1 s2 s3 # i if i >= ndim - a.ndim: n_a_i += 1 a_i = i + a.ndim - ndim if not a.broadcastable[a_i]: pre_v_shape.append(a.shape[a_i]) bcast.append(False) break else: if n_a_i == 0: raise ValueError( ('Auto-shape of -1 must overlap' 'with the shape of one of the broadcastable' 'inputs')) else: pre_v_shape.append(tensor.as_tensor_variable(1)) bcast.append(True) else: ValueError('negative shape', s) # post-condition: shape may still contain both symbolic and # non-symbolic things v_shape = tensor.stack(*pre_v_shape) elif shape is None: # The number of drawn samples will be determined automatically, # but we need to know ndim if not args: raise TypeError(('_infer_ndim_bcast cannot infer shape without' ' either shape or args')) template = reduce(lambda a, b: a + b, args) v_shape = template.shape bcast = template.broadcastable ndim = template.ndim else: v_shape = tensor.as_tensor_variable(shape) if ndim is None: ndim = tensor.get_vector_length(v_shape) bcast = [False] * ndim if (not (v_shape.dtype.startswith('int') or v_shape.dtype.startswith('uint'))): raise TypeError('shape must be an integer vector or list', v_shape.dtype) if args_ndim > ndim: raise ValueError( 'ndim should be at least as big as required by args value', (ndim, args_ndim), args) assert ndim == len(bcast) return ndim, tensor.cast(v_shape, 'int32'), tuple(bcast)