TaylorIntegration.jl
TaylorIntegration.jl copied to clipboard
ODE integration using Taylor's method, and more, in Julia
One way to reduce additional recompilations is to construct the ODE problem with `ODEProblem{true,false}(...)`. This is currently not supported by TaylorIntegration.jl, but it would be great to have this. Minimal...
See [this comment](https://github.com/JuliaReach/ReachabilityAnalysis.jl/pull/621#issuecomment-1107752118); the MWE there is: ```julia julia> @taylorize function f!(du, u, p, t) x, t = u return du end ERROR: syntax: "__tmpTaylor[2]" is not a valid function...
This is motivated by [this comment](https://github.com/gacarita).
Reference: https://github.com/JuliaReach/ReachabilityAnalysis.jl/pull/621#issuecomment-1107752118 The following is a minimal example: ```julia julia> @taylorize function f!(du, u, p, t) x, t = u return du end ERROR: syntax: "__tmpTaylor[2]" is not a valid...
In many applications, people is interested in the maximum Lyapunov exponent *only*. We should have a function which provides it.
After #114 is merged, we will need to review the keyword warning list. Related to https://github.com/PerezHz/TaylorIntegration.jl/pull/114#discussion_r524790167
```julia using DifferentialEquations using TaylorIntegration @taylorize function benchmark7!(du, u, p, t) du[1] = u[3]^3 - u[2] + u[4] du[2] = u[3] du[3] = 2 du[4] = u[4] return nothing end...
Currently, we have different ways to set the initial and final times for the integration, which defines which data is returned. DifferentialEquations distinguishes `tspan` (a tuple for the initial and...
This is necessary to spot easily performance regressions.
At present, `TaylorIntegration` (`TI`) supports jet transport (JT) "naively". This means that we assume that polynomials converge both with respect to time and JT variables. However, we don't enforce it....