kamilkhanlab Goto Github PK

abs-normal

Tools in Julia for solving equations and optimization problems involving piecewise linear/affine functions expressed in "absolute normal form".

computational-graph-tools

Tools in Julia for automatically constructing the computational graph/tape of a composite function, and performing the reverse AD mode.

continuous-convex-adjoints

A Julia/C++ implementation and numerical examples of adjoint subgradient evaluation for convex ODE relaxations, as described by Zhang and Khan (2024).

convex-ode-subgradients

Contains code for the numerical examples in an article by Song and Khan (2024). This code evaluates subgradients for convex relaxations of parametric ordinary differential equations (ODEs)

convexsampling.jl

Constructs linear underestimators of convex functions by tractable black-box sampling. Implemented in Julia.

implicit-func-relaxations

Proof-of-concept implementation of a new method for computing convex relaxations for implicit functions and inverse functions.

kamilkhanlab.github.io

nonogram-ilp

ILP formulation in GAMS for solving nonograms

nonogramsolver.jl

Formulates and solves nonogram puzzles (a.k.a. Picross and paint-by-number) in Julia. Uses a new integer linear programming formulation, and solves it with JuMP.

nonsmooth-forward-ad

Implementation of the vector forward mode of automatic differentiation (AD) for generalized differentiation of nonsmooth functions. Uses operator overloading in Julia.

ob-ode-relaxations

Proof-of-concept implementation of a method by Song and Khan (2022) for computing convex relaxations for parametric ordinary differential equations.