search

ZichaoLong / PDE-Net

PDE-Net: Learning PDEs from Data
311 stars 104 forks source link

Optimization Method #9

Open ankitesh97 opened 4 years ago

ankitesh97 commented 4 years ago

I was wondering why did you use BFGS optimization instead of inbuilt ADAM/Gradient descent optimization method in pytorch?

ZichaoLong commented 4 years ago

I was wondering why did you use BFGS optimization instead of inbuilt ADAM/Gradient descent optimization method in pytorch?

BFGS with line search converges quicker. ADAM and SGD with a fixed learning rate are not stable.