N: the order of the HiPPO projection dt: discretization step size - should be roughly inverse to the length of the sequence

N: the order of the HiPPO projection dt: discretization step size - should be roughly inverse to the length of the sequence

Maps reals to (lo, hi), where 0 maps to (lo+hi)/2.

Maps real numbers to (lo, infinity), where 0 maps to ref.

r"""Accurately approximate the partition function Z(numer / denom). This uses the analytical formulation of the true partition function Z(alpha),

Collapses a wavelet decomposition made by construct() back into an image. Args: pyr: A numpy or TF tensor of single or double precision floats

Computes the Jacobian of function `f` with respect to input `x`.

num_levels a wavelet decomposition of an image. Args: im: A numpy or TF tensor of single or double precision floats of size (batch_size,

r"""Draw samples from the robust distribution. This function implements Algorithm 1 the paper. This code is written to allow for sampling

The same as torch.exp(x), but clamps the input to prevent NaNs.

The same as tf.math.expm1(x), but clamps the input to prevent NaNs.