SymbolicAnalysis.jl

Symbolics-based function property propagation for optimization
Author Vaibhavdixit02
Popularity
13 Stars
Updated Last
1 Month Ago
Started In
August 2023

SymbolicAnalysis.jl

Symbolics.jl based function property propagation for optimization

SymbolicAnalysis is a package for implementing the Disciplined Programming approach to optimization, As demonstrated by the DCP framework, and further followups to it for further classes of functions https://www.cvxpy.org/tutorial/index.html such as DGP, DQP etc, symbolic representation of problems can be leveraged to identify and facilitate building Convex (or similar function properties) expressions.

This package aims to utilize expression graph rewriting and metadata propagation supported by Symbolics.jl, to support propagation of several of these properties - limited right now to Euclidean Convexity and Geodesic Convexity on the Symmetric Positive Definite manifold. This package provides an easier to expand implementation of functional properties than the previous implementations CVXPY and Convex.jl as well as a more performant implementation of the function property propagation.