MrToph / CFMMRouter.jl

Convex optimization for fun and profit. (Now in Julia!)

CFMMRouter

Overview

This package contains a fast solver for the CFMM Routing problem, as defined by Angeris et al. in Optimal Routing for Constant Function Market Makers. We partially decompose the problem to enable fast solutions when the number of CFMMs is large relative to the number of tokens.

For more information, check out the documentation.

Quick Start

First, add the package locally.

using Pkg; Pkg.add(url="https://github.com/bcc-research/CFMMRouter.jl")

Make some swap pools.

using LinearAlgebra
using CFMMRouter

equal_pool = ProductTwoCoin([1e6, 1e6], 1, [1, 2])
unequal_small_pool = ProductTwoCoin([1e3, 2e3], 1, [1, 2])
prices = ones(2)

Build a Router & route.

router = Router(
    LinearNonnegative(prices),
    [equal_pool, unequal_small_pool],
    2,
)
route!(router)

Check out the results.

Ψ = round.(Int, netflows(router))
println("Profit: $(dot(prices, Ψ))")

Performance

This routing algorithm scales approximately linearly in the number of swap pools for the arbitrage problem. These tests were run on a MacBook Pro with a 2.3GHz 8-Core Intel i9 processor. Several performance improvements are possible.

References

G Angeris, T Chitra, A Evans, and S Boyd. Optimal Routing for Constant Function Market Makers