OrdinaryDiffEq.jl icon indicating copy to clipboard operation
OrdinaryDiffEq.jl copied to clipboard

Published 20 hours ago • verified publisher icon SciML

New 5(4) RK coefficients

Open ChrisRackauckas opened this issue 3 years ago • 2 comments
trafficstars

https://arxiv.org/abs/2202.08443

ChrisRackauckas avatar Feb 18 '22 10:02 ChrisRackauckas

Found a precursor method https://arxiv.org/pdf/2108.12590.pdf

dynamic-queries avatar Mar 30 '22 12:03 dynamic-queries

  • [ ] Implement 6th order method from table 4 - MSRK6
  • [ ] Adaptive method MSRK54 and MSRK64

dynamic-queries avatar Apr 08 '22 13:04 dynamic-queries

Could give implementing the 6th order method a try.

pri1311 avatar Dec 31 '22 07:12 pri1311

It would be good to finish the 5h order method first. Adaptivity and interpolation. https://docs.sciml.ai/DiffEqDevDocs/stable/contributing/adding_algorithms/ is a good overview. Then see:

https://github.com/SciML/OrdinaryDiffEq.jl/blob/master/src/tableaus/low_order_rk_tableaus.jl#L501-L609 https://github.com/SciML/OrdinaryDiffEq.jl/blob/master/src/perform_step/low_order_rk_perform_step.jl#L903-L910

For how the adaptive estimates are computed using the btilde values. Let me know if you need any help there. One that and the interpolation are in, we can add it to our benchmarking suite: the author is curious about what the results will be.

ChrisRackauckas avatar Dec 31 '22 14:12 ChrisRackauckas