Hi Adam,
I had a look at that and from my understanding it should be compatible with
the `conjugateGradientDescent` function which is fine for me.
I am not sure yet how this works out, but I'll see where I can get with
this.
Thanks,
Florian
2015-05-12 0:29 GMT+02:00 adam vogt
Hi Florian,
The restriction Edward described is gone: here's an unboxed vector of ForwardDouble: https://gist.github.com/aavogt/6bee72717148123a187c. But I don't think there's been any work to make hmatrix's Field class have a ForwardDouble instance. I don't know of a ReverseDouble either.
Regards, Adam
On Mon, May 11, 2015 at 6:24 AM, Florian Hofmann < fhofmann@techfak.uni-bielefeld.de> wrote:
In [1](this) email thread Edward Kmett and Dominic Steinitz discuss the state of interoperability of `ad` and `hmatric` (or `repa`) with Edward hinting that the 4.0 line of `ad` might improve on the situation.
Could someone elaborate on the state of affairs two years later?
Florian
[1]: https://mail.haskell.org/pipermail/haskell-cafe/2013-April/107561.html
_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe