Skip to content

Experiments with using neural networks to solve differential equations.

Notifications You must be signed in to change notification settings

thecomputationguy/neural-network-differential-equations

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

neural-network-differential-equations

Experiments with using neural networks to solve differential equations. In this particular example, the non-linear ODE of a damped simple harmonic oscillator is considered. The ideas are based on the work of Raissi et al. from https://arxiv.org/abs/1801.01236. Using an MLP with just 1 fully-connected layer with 32 neurons is usually sufficient to replicate the dynamics of the system considered here, as shown.

Case with initial condition (5,5)

To check for general model behaviour, 2 additional cases are considered, each with a different initial condition:

Case with initial condition (2,2)

Case with initial condition (1,1)

About

Experiments with using neural networks to solve differential equations.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published