Skip to content

Interpolating Adjoint Method for DDEs with constant lags

License

Notifications You must be signed in to change notification settings

andrschl/dde_adjoint_method

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

20 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

dde_adjoint_method

Julia code for interpolating adjoint sensitivity method for Neural DDEs with constant lags [1].

dde_adjoint.jl contains the interpolating adjoint code followed by a short neural ODE example. It uses Differentialequations.jl [2] for the DDE solvers. The own implementation is compared to built-in AD sensitivity methods ForwardDiffSensitivity() and ReverseDiffAdjoint().

The method assumes autonomous dynamics, constant lags, and currently also a smooth enough transition between initial history and DDE solution.

References

[1] Calver, Jonathan, and Wayne Enright. “Numerical Methods for Computing Sensitivities for ODEs and DDEs.” Numerical Algorithms 74, no. 4 (April 2017): 1101–17. https://doi.org/10.1007/s11075-016-0188-6.

[2] Rackauckas, Christopher and Nie, Qing. “Differentialequations.jl--a performant and feature-rich ecosystem for solving differential equations in julia.“ Journal of Open Research Software, 2017

About

Interpolating Adjoint Method for DDEs with constant lags

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages