mxfusion.util.testutils

Members

mxfusion.util.testutils.prepare_mxnet_array(array, array_has_samples, dtype)
mxfusion.util.testutils.numpy_array_reshape(var, isSamples, n_dim)

Reshape a numpy to a give dimensionality by adding dimensions with size one in front (broadcasting). If isSamples is true, keep the first dimension.

Parameters:
  • var (numpy.ndarray) – the variable to be reshaped.
  • isSamples (boolean) – whether the variable is a sampled variable with the first dimension being the number of samples.
  • n_dim (int) – the dimensionality that the variable is reshaped to.
Returns:

the reshaped numpy array
Return type:

numpy.ndarray
class mxfusion.util.testutils.MockMXNetRandomGenerator(samples)

Bases: mxfusion.components.distributions.random_gen.RandomGenerator

The MXNet pseudo-random number generator.

sample_normal(loc=0, scale=1, shape=None, dtype=None, out=None, ctx=None, F=None)
sample_multinomial(data, shape=None, get_prob=True, dtype=None, F=None)
sample_bernoulli(prob_true=0.5, shape=None, dtype=None, out=None, F=None)
sample_gamma(alpha=1, beta=1, shape=None, dtype=None, out=None, ctx=None, F=None)
sample_uniform(low=0.0, high=1.0, shape=None, dtype=None, out=None, ctx=None, F=None)
sample_laplace(location=0.0, scale=1.0, shape=None, dtype=None, out=None, ctx=None, F=None)
mxfusion.util.testutils.make_net()
mxfusion.util.testutils.make_basic_model(finalize=True, verbose=True)
mxfusion.util.testutils.make_bnn_model(finalize=True, verbose=True)
class mxfusion.util.testutils.DotProduct(prefix=None, params=None)

Bases: mxnet.gluon.block.HybridBlock

hybrid_forward(F, x, *args, **kwargs)

Overrides to construct symbolic graph for this Block.

Parameters:
  • x (Symbol or NDArray) – The first input tensor.
  • *args (list of Symbol or list of NDArray) – Additional input tensors.
class mxfusion.util.testutils.TestBlock

Bases: mxnet.gluon.block.HybridBlock

Block with standard functions for initializing and running an MXNet Gluon block for unit testing.

hybrid_forward(F, x, var1, var2)

Simple test function to enable MXFusion tests that use a Gluon block. :param F: MXNet computation type <mx.sym, mx.nd> :param x: MXNet dummy variable :param var1: MXNet dummy variable :param var2: MXNet dummy variable

mxfusion.util.testutils.plot_univariate(samples, dist, buffer=0, **kwargs)

Visual inspection by plotting the distribution: plots a histogram of the samples along with

Parameters:
  • samples ((mx.nd.NDArray, np.ndarray)) – Samples from the distribution
  • buffer – additional range to plot the distribution over
  • dist – Distribution that these are samples from (scipy.stats)
  • kwargs – Keyword arguments for the distribution (e.g. loc, scale)
mxfusion.util.testutils.plot_bivariate(samples, dist, buffer=0, **kwargs)

Visual inspection by plotting the distribution: plots a scatter plot of samples along with a contour plot

Parameters:
  • samples ((mx.nd.NDArray, np.ndarray)) – Samples from the distribution
  • buffer – additional range to plot the distribution over
  • dist – Distribution that these are samples from (scipy.stats)
  • kwargs – Keyword arguments for the distribution (e.g. loc, scale)
mxfusion.util.testutils.make_spd_matrix(dim)

Generate a random symmetric, positive-definite matrix.

Parameters:dim (int) – The matrix dimension (matrix is square).
Return X:The random symmetric, positive-definite matrix.
Return type:array of shape [n_dim, n_dim]