Lazylinop: fast and versatile linear operators in Python#
Lazylinop is a Python toolbox to ease and accelerate computations with (“matrix-free”) linear operators. Operators can be easily manipulated, transformed or aggregated with other linear operators to form more complex LazyLinOp objects as easily as with NumPy arrays and SciPy matrices.
A LazyLinOp is a high-level linear operator based on an arbitrary underlying implementation including transparent use of efficient third-party libraries (e.g for wavelets or non-uniform FFT). The evaluation is delayed (lazy paradigm) until the LazyLinOp is actually applied to a collection of vectors resulting in potential performance gains and memory sparing.
It also offers torch tensors and CuPy arrays compatibility, standard signal/image processing linear operators, polynomials of operators, SVDs of linear operators as well as advanced tools to implement efficient Butterfly operators, to approximate and quantize large linear operators.