# bayespy.utils.misc¶

General numerical functions and methods.

Functions

 `T`(X) Transpose the matrix. `add_axes`(X[, num, axis]) `add_leading_axes`(x, n) `add_trailing_axes`(x, n) `array_to_scalar`(x) `atleast_nd`(X, d) `axes_to_collapse`(shape_x, shape_to) `block_banded`(D, B) Construct a symmetric block-banded matrix. `block_diag`(*arrays) Form a block diagonal array from the given arrays. `broadcast`(*arrays[, ignore_axis]) Explicitly broadcast arrays to same shapes. `broadcasted_shape`(*shapes) Computes the resulting broadcasted shape for a given set of shapes. `broadcasted_shape_from_arrays`(*arrays) Computes the resulting broadcasted shape for a given set of arrays. `broadcasting_multiplier`(plates, *args) Compute the plate multiplier for given shapes. `ceildiv`(a, b) Compute a divided by b and rounded up. `composite_function`(function_list) Construct a function composition from a list of functions. `concatenate`(*arrays[, axis]) Concatenate arrays along a given axis. `diag`(X[, ndim]) Create a diagonal array given the diagonal elements. `diagonal`(A) `dist_haversine`(c1, c2[, radius]) `find_set_index`(index, set_lengths) Given set sizes and an index, returns the index of the set `first`(L) `flatten_axes`(X, *ndims) `gaussian_logpdf`(y_invcov_y, y_invcov_mu, …) `get_diag`(X[, ndim, ndim_to]) Get the diagonal of an array. `gradient`(f, x[, epsilon]) `grid`(x1, x2) Returns meshgrid as a (M*N,2)-shape array. `identity`(*shape) `invgamma`(x) Inverse gamma function. `invpsi`(x) Inverse digamma (psi) function. `is_callable`(f) `is_numeric`(a) `is_scalar_integer`(x) `is_shape_subset`(sub_shape, full_shape) `is_string`(s) `isinteger`(x) `logsumexp`(X[, axis, keepdims]) Compute log(sum(exp(X)) in a numerically stable way `m_digamma`(a, d) Returns the derivative of the log of multivariate gamma. `m_dot`(A, b) `make_diag`(X[, ndim, ndim_from]) Create a diagonal array given the diagonal elements. `make_equal_length`(*shapes) Make tuples equal length. `make_equal_ndim`(*arrays) Add trailing unit axes so that arrays have equal ndim `mean`(X[, axis, keepdims]) Compute the mean, ignoring NaNs. `moveaxis`(A, axis_from, axis_to) Move the axis axis_from to position axis_to. `multidigamma`(a, d) Returns the derivative of the log of multivariate gamma. `multiply`(*arrays) `multiply_shapes`(*shapes) Compute element-wise product of lists/tuples. `nans`([size]) `nested_iterator`(max_inds) `normalized_exp`(phi) Compute exp(phi) so that exp(phi) sums to one. `parse_command_line_arguments`(mandatory_args, …) Parse command line arguments of style “–parameter=value”. `put`(x, indices, y[, axis, ufunc]) A kind of inverse mapping of np.take `put_simple`(y, indices[, axis, length]) An inverse operation of np.take with accumulation and broadcasting. `remove_whitespace`(s) `repeat_to_shape`(A, s) `reshape_axes`(X, *shapes) `rmse`(y1, y2[, axis]) `safe_indices`(inds, shape) Makes sure that indices are valid for given shape. `squeeze`(X) Remove leading axes that have unit length. `squeeze_to_dim`(X, dim) `sum_multiply`(*args[, axis, sumaxis, keepdims]) `sum_multiply_to_plates`(*arrays[, to_plates, …]) Compute the product of the arguments and sum to the target shape. `sum_product`(*args[, axes_to_keep, …]) `sum_to_dim`(A, dim) Sum leading axes of A such that A has dim dimensions. `sum_to_shape`(X, s) Sum axes of the array such that the resulting shape is as given. `symm`(X) Make X symmetric. `tempfile`([prefix, suffix]) `trues`(shape) `unique`(l) Remove duplicate items from a list while preserving order. `write_to_hdf5`(group, data, name) Writes the given array into the HDF5 file. `zipper_merge`(*lists) Combines lists by alternating elements from them.

Classes

 `TestCase`([methodName]) Simple base class for unit testing.