@bwbush Sounds nice! Perhaps I just need to take the dive and install NixOS on my personal computer. Right now I only use it to manage GHC/Stack and Pandoc on Arch
@bwbush Is NixOS your daily use OS?
@rufousboi It means "manager who is hiring", no?
I think I understand your question. I don't think Haskell needs to concretely resolve every forall in order to produce a value. For example:
empt :: [a] -- same as forall a. [a]
empt = 
lenEmpt :: Int
lenEmpt = length empt
I made a stupid example for u:
Explained it better than I could. Though I think AD is a slightly better example in your case since you can actually have two terms with different s in the same scope
What do #Coq people think about the ability to generate Haskell? Is the code actually reasonable? Wouldn't it be better to target something like GHC Core?
I hope P = NP but the optimal algorithm for subset sum is like O(n^(Graham's number)) with massive constant factors so it doesn't actually help with solving NP problems
@abs U can use mine but u can't change the pass because my brother uses it sometimes. I can also just buy it for u
@iitalics I deeply respect ur opinion because I love u like a brother but I find Scala to be super unergonomic. Haven't used it in a bit but I felt like it had all of functional programming and all of object oriented programming jammed into it. Also the type inference is wack. Also some of the more complicated features are wack (thinking of implicit parameters.) I wrote a somewhat long compiler for a toy OOP language with affine types a short time ago in scala. It was great because I could just add a mutable field whenever I wanted one, unlike haskell, but it was really annoying to deviate from (java + case classes) and there were some surprisingly nasty runtime bugs that I'm not used to in a strongly typed language. Also I felt like it was hard to write code outside of IntelliJ
Miaourt hand reveal
Yeah, it is pretty interesting they're the exact same at the Core level!
Perhaps this is a stupid way of looking at it, but everything gets compiled to assembly code, where it seems there is only one way of reversing an immutable linked list in linear time, so maybe it's inevitable reverse' and reverse'' end up equivalent 🤔
I think traditional factor investing has not been too good since 2007, for many reasons. Overcrowding and strong cross correlation between funds could be one, especially given the insane leverage some of these firms have. Good summary: https://boston.qwafafew.org/wp-content/uploads/sites/3/2017/08/KCMAugust2007.pdf.
That being said, there are some quant giants that continue to have incredible success, perhaps due to the execution factor 22 mentioned. Market makers also seem to be doing exceptionally well recently.
Based on experience I do think crashes is where HF shines; though perhaps too many funds claim to deliver uncorrelated returns, a lot of places I'm familiar with really did exceptionally well in 2008. But given that it can be hard to determine which funds really deliver (other than some select impossible-to-enter funds like 2S and Renaissance,) I tend to be more interested in a Taleb-type approach of always being positively exposed to tail risk. I feel like I'm fine with slower growth in bull periods such as this one. That, and I don't have a high enough net worth to go in Hedge Funds anyways