for testing and deploying your application
for finding and fixing issues
for empowering human code reviews
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import numpy as np
from theano import gof, tensor
import theano
import theano.tensor as T
class SampleMultivariateGaussian(gof.Op):
__props__ = ()
def make_node(self, x, cov):
if x.type.ndim != 1:
raise TypeError('x must be a 1-d vector')
if cov.type.ndim != 2:
raise TypeError('cov must be a 2-d matrix')
return gof.Apply(self, [x, cov], [T.vector(dtype="float32")])
def perform(self, node, inp, out):
x, cov = inp
z, = out
z[0] = np.random.multivariate_normal(x, cov, 1)[0].astype("float32")
def grad(self, inputs, outputs):
return [outputs[0], T.zeros_like(inputs[1])]
def R_op(self, inputs, eval_points):
if eval_points[0] is None:
return eval_points
return self.grad(inputs, eval_points)
zomux /
deepy
