[Idea]: Optimization Algorithms

[Planeshifter]

Idea

We currently do not have optimization algorithms in stdlib. Having support for Linear Programming, Convex Optimization, Quadratic Programming, and/or Non-Linear Optimization algorithms would be a great addition.

Expected Outcomes

stdlib will have a broad array of optimization algorithms for solving problems.

Involved Software

No other software should be necessary. However, we will need to do a needs analysis to determine which prerequisite packages/functionality is necessary in order to allow these algorithms to be implemented (e.g., BLAS).

Prerequisite Knowledge

JavaScript, Node.js. Familiarity with R, Python, C/C++ would be very useful, as will need to consult reference implementations.

Difficulty

Hard. Depends on the reference implementation requirements and algorithmic difficulty.

Project Length

350 hours.

Potential Mentors

@kgryte @Planeshifter @rreusser @Pranavchiku @czgdp1807

[kgryte]

@Planeshifter May be good to flesh out this idea a bit more, perhaps with a link or two to example APIs found in other libraries/programming languages.

[kgryte]

Some additional resources:

• Minpack: https://netlib.org/minpack/
• https://github.com/cbouilla/minpack-1.1 (contains additional links, as well)
• Fortran lang discussion: https://fortran-lang.discourse.group/t/lets-get-fortran-lang-minpack-into-scipy/2722/12

[kgryte]

Scalar root finding functions could be a decent place to start: https://docs.scipy.org/doc/scipy/reference/optimize.html#scalar-functions

[kgryte]

• SciPy's milp API wraps HiGHS: https://github.com/ERGO-Code/HiGHS/tree/master

[kgryte]

• HiGHS.ts: https://github.com/opvious/highs.ts

[kgryte]

• HiGHS ported to Wasm: https://github.com/lovasoa/highs-js