API Reference

This section provides a full API reference for this library. The functionality is split into several submodules, providing routines with varying functionality. Basic routines are provided in adrt and useful utilities are in adrt.utils.

Many of the functions in this module have requirements for the shapes and dtypes of their inputs. The core numerical routines support inputs with dtype either numpy.float32 or numpy.float64 which are referred to as “numpy.ndarray of float”.

Our main routines often have requirements for the shapes of input arrays. Often, these functions also support an optional batch dimension which makes it possible to process multiple independent inputs at once without looping in Python.

Frequently, we will refer to an “ADRT output” or an “ADRT output of size N” which is an array with a shape matching:

(batch?, 4, 2*N-1, N)

which specifies

  1. an optional batch dimension

  2. a dimension for the ADRT quadrants, of size exactly four

  3. a dimension for ADRT offsets with size given by the formula, referencing the last dimension of size N

  4. a dimension for the angles of size N, which must be a power of two

Contents