CMM-22 / a6

0 stars 0 forks source link

Newton Method #1

Closed Memphis99 closed 2 years ago

Memphis99 commented 2 years ago

Hello, for point 4 I implemented the objective function with the regularizer on |u|^2. When I start the simulation with GD, all goes smoothly, but when I simulate with Newton Method, it does optimize the trajectory, but all the app and the simulation gets laggy and really slow. I tried using different log_c_reg but it lags everytime. Am I making some errors? Could be the app's problem?

joshuaoreilly commented 2 years ago

I also have a lot of lag, even in release mode. Not sure if it's normal.

MiguelZamoraM commented 2 years ago

This Is normal. The point is that we are computing both the gradient and the Hessian using finite differences, and that requires a lot of compute. Recall that if it gets laggy and you want to stop the optimization you can press the letter "o" in your keyboard.

Memphis99 commented 2 years ago

Ok thank you!