Sacha Sokoloski

So I reached out to Edward Kmett about this, and he said he'd be willing to contribute to the AD side of things on such a project. He didn't say...

So I wrote to the Haskell commitee to see if they still recommended submitting a proposal, and apparently the "organization deadline" is much earlier then the actual project deadline. So...

Alright, done and done. The repo is [here](https://github.com/alex404/summer-of-haskell). I just wrote up something quite preliminary, so edit it however you wish, and I'll go over it again too. I think...

Hey, Edward Kmett added some info on AD to the project proposal. Would you mind filling it out with Accelerate side stuff? I'll edit it once you're done and run...

Adressing [ekmett/ad#2](https://github.com/ekmett/ad/issues/2) definitely sounds like a good way forward. My only concern is that if we only apply this solution to implementing AD in `hmatrix`, then we will have changed...

Yah, I think what you're saying is right, and we can of course always change our minds about how exactly to proceed if/when the project actually happens. My understanding of...

That's great news! How should we proceed now? Do you want to rework your contribution now that we're changing strategies? Based on our discussion I can also rework the draft...

Okay, I had another go at the draft. I've written it up saying there are 2 strategies (based on the suggestions of @ekmett ), and that one may be easier/less...

Okay, our idea is now online at the [summer of Haskell](https://summer.haskell.org/ideas.html). Any thoughts how we should find a student? Should we try to advertise or should we just wait for...

I've also been stuck using GHC 8.4 because of this issue. I personally use [ghc-typelits-natnormalise](https://hackage.haskell.org/package/ghc-typelits-natnormalise) and [ghc-typelits-knownnat](https://hackage.haskell.org/package/ghc-typelits-knownnat) for my work with type safe vectors. The aforementioned solution with deferred type...