DiffEqParamEstim.jl
DiffEqParamEstim.jl copied to clipboard
SciML
Easy scientific machine learning (SciML) parameter estimation with pre-built loss functions
Fixes #248 ## Checklist - [ ] Appropriate tests were added - [ ] Any code changes were done in a way that does not break public API - [...
@Vaibhavdixit02 could you have a look? I want to add some CI enhancements here like I did in other repos, and this is blocking me.
**Describe the bug 🐞** The cost function returned by `multiple_shooting_objective` is not differentiable by Zygote, which prevents using the AutoZygote AD setting in optimisation. **Expected behavior** Zygote succesfully computes the...
@Vaibhavdixit02 Tests are not passing on master in various ways. Can you take a look at that such that I can finish this PR?
Could it be documented if/how the width and weights of the kernels are adapted based on the frequency of the data? In particular if the time between measurements is not...
Hi, IMHO removing `build_lsoptim_objective()` breaks existing code unnecessarily. Moreover, this used to be in usage examples, and I did not find a clear upgrade recipe to the current version.
Since the package is defined as "_a simple interface for users who want to quickly run standard parameter estimation routines for model calibration_", it would be interesting to add a...
DiffEqFlux has copy pastes of some of the functions here. They should be removed in one of the two. https://github.com/SciML/DiffEqFlux.jl/issues/773 I would also suggest deprecating passing the kernel as a...
Consider the following example: https://diffeq.sciml.ai/stable/analysis/parameter_estimation/#Optimization-Based-ODE-Inference-Examples If one makes an `L2Loss` there is nothing that stops one from doing: ``` using DiffEqParamEstim, DifferentialEquations, Plots, RecursiveArrayTools function f(du, u, p, t) du[1]...
