Oyewale Abdulsobur
Oyewale Abdulsobur
Thank you for the detailed reply.
Hi, sorry i have been away for a while. I have tried refactoring to solve the issues requirement. Before submitting a new PR, i have provided the new update [here](https://gist.github.com/smilingprogrammer/4d3cbd765ee155eff6b77773aab2b3fe)....
> You do not have to open a new PR. You can add your corrections to this PR if you like. > > Is the algorithm behaviour in this graph...
What should be the expectation here? Will be better if there is a description of what the expectation should be. The Powell's method has a really low resource on it...
I have found the paper!!! https://sci-hub.se/10.1093/comjnl/7.2.155 Will take my time to go through it. Question; Should i still stick to the Hill Climb for the line search during implementation?
Hi, Greetings @ankurankan Is this issue still available to be worked on?
Hi @SimonBlanke. Looking at this issue it seems that the search is restricted to one coordinate dimension at a time. Therefore when the algorithm switches dimensions via new_dim, the new...
My proposed solution; Integrate a continuous line search at the end of each Powell cycle. Therefore at the end of each full cycle, we compute the net displacement.
What i'm actually describing is also how i will be implementing it as described in mathematical literature. As for the line search, i meant the “continuous‐angle” line search as described...
Update, Found the issue.