search

CMM-22 / a6

0 stars 0 forks source link

Ex.4/5 Objective function and Newton's method #5

Closed kiremicev closed 2 years ago

kiremicev commented 2 years ago
  1. Can we calculate the objective function in any way we want or is there a recommended/required way to calculate it.
  2. In Ex.4 it is said: Your solution should work really, really well. What is good enough? My solution for both gradient descent and Newton's method looks as follows: https://drive.google.com/file/d/1tN_d7TKd7OO3I-3G2xHnAgkKp9E2zW7a/view?usp=sharing (number of steps implemented in the GUI) Is this be good enough?
  3. How exactly are we supposed to answer question of Ex.5: Try optimizing for a few different targets with Newton’s method. How many steps does Newton’s method usually take to converge for this problem? Why? Why it takes that amount of steps or why it takes less steps than gradient descent method?

Thank you in advance for your help!

kiremicev commented 2 years ago

I think I got it. I made a mistake which made the Newton's method very unpredictable.