Accelerating Simulation of Stiff Nonlinear Systems using Continuous-Time Echo State Networks — mentioned by ChrisRackauckas

Wed, 22 Sep 2021 10:58:56 GMT

Not just similar, the same. If you look through the documentation you'll see that https://github.com/SciML/ReservoirComputing.jl is a collection of reservoir architectures with high performance implementations, and some of our recent research has been pulling reservoir computing to the continuous domain for stiff ODEs (think of it almost like a neural ODE that you do not need to train via gradient descent): https://arxiv.org/abs/2010.04004 . We are definitely digging through this paper with some fascination and will incorporate a lot of its advancements (reservoir-free architectures) into the…

Paper and discussion history. A mention is not an endorsement.

Accelerating Simulation of Stiff Nonlinear Systems using Continuous-Time Echo State Networks — mentioned by ChrisRackauckas

Thu, 11 Mar 2021 16:50:35 GMT

There are ways to surrogatize portions of a simulation, replacing large portions with ML-trained surrogates. There are already demonstrations in different domains (on highly stiff differential-algebraic equations) showing that you can get these kinds of speedups at around 99% accuracy. https://arxiv.org/abs/2010.04004 https://arxiv.org/abs/2103.05244

Paper and discussion history. A mention is not an endorsement.

Accelerating Simulation of Stiff Nonlinear Systems using Continuous-Time Echo State Networks · HackerPapers discussions

Accelerating Simulation of Stiff Nonlinear Systems using Continuous-Time Echo State Networks — mentioned by ChrisRackauckas

Accelerating Simulation of Stiff Nonlinear Systems using Continuous-Time Echo State Networks — mentioned by ChrisRackauckas