My article [1] in Journal of Mathematical Sciences (Springer) will be released this October. It notes D standard library 3 times:

1. findRoot. D implementation is significantly better then 98% of others  for the problem because the problem behaves like pathological. Thanks to ieeeMean
2. logmdigamma
3. logmdigammaInverse

The article is already available online [1].
Of course the source code for the article was written in D [2].

If you want to use D for Science or Machine Learning go forward with Mir project [3].

Also you can help Mir in
 - 5 second: star the project https://github.com/libmir/mir
 - 1 hour+:
     - opt1: Write an article or about ndslice or mir.glas [6] (upcoming BLAS implementation in D)
     - opt2: Add small enhancement you want, see also [4]
     - opt3: Include new chapter about ndslice and Mir to the Dlang Tour [7]
 - 1 day+: Became an author for a new package, see also [5].

Companies can order numerical, statistical, and data mining algorithms and services. We work with web and big data.

[1] http://rdcu.be/kiKR -  "On Robust Algorithm for Finding Maximum Likelihood Estimation of the Generalized Inverse Gaussian Distribution"
[2] https://github.com/9il/atmosphere - library, which contains the source code for the article
[3] https://github.com/libmir/mir
[4] https://github.com/libmir/mir/issues?q=is%3Aissue+is%3Aopen+label%3Aenhancement
[5] https://github.com/libmir/mir/issues?q=is%3Aissue+is%3Aopen+label%3A%22New+Package%22
[6] http://docs.mir.dlang.io/latest/mir_glas_l3.html
[7] http://tour.dlang.org/

Best regards,
Ilya

On Tuesday, 13 September 2016 at 14:14:01 UTC, Ilya Yaroshenko wrote:

> Also you can help Mir in
>  - 5 second: star the project https://github.com/libmir/mir
>  - 1 hour+:
>      - opt1: Write an article or about ndslice or mir.glas [6] (upcoming BLAS implementation in D)
>      - opt2: Add small enhancement you want, see also [4]
>      - opt3: Include new chapter about ndslice and Mir to the Dlang Tour [7]
>  - 1 day+: Became an author for a new package, see also [5].
>
> Companies can order numerical, statistical, and data mining algorithms and services. We work with web and big data.
>
> [1] http://rdcu.be/kiKR -  "On Robust Algorithm for Finding Maximum Likelihood Estimation of the Generalized Inverse Gaussian Distribution"
> [2] https://github.com/9il/atmosphere - library, which contains the source code for the article
> [3] https://github.com/libmir/mir
> [4] https://github.com/libmir/mir/issues?q=is%3Aissue+is%3Aopen+label%3Aenhancement
> [5] https://github.com/libmir/mir/issues?q=is%3Aissue+is%3Aopen+label%3A%22New+Package%22
> [6] http://docs.mir.dlang.io/latest/mir_glas_l3.html
> [7] http://tour.dlang.org/
>
> Best regards,
> Ilya

How stable is Mir? I have recently stripped my library for embedding R inside D down to the minimum amount and created an R package to do the installation. Therefore it is trivial to install and get started on Linux.* I would like to test how it works to mix R and Mir code. However, I don't want to dig into that until Mir is in a stable state.

* Also Windows and Mac, but since I don't have either of those machines, I cannot do any work with them.

On Tuesday, 13 September 2016 at 14:43:11 UTC, bachmeier wrote:
> On Tuesday, 13 September 2016 at 14:14:01 UTC, Ilya Yaroshenko wrote:
>
>> [...]
>
> How stable is Mir? I have recently stripped my library for embedding R inside D down to the minimum amount and created an R package to do the installation. Therefore it is trivial to install and get started on Linux.* I would like to test how it works to mix R and Mir code. However, I don't want to dig into that until Mir is in a stable state.
>
> * Also Windows and Mac, but since I don't have either of those machines, I cannot do any work with them.

Recent release v0.15.3 is stable.
v0.17.0-alpha3 may have API changes in mir.glas and mir.random.

mir.ndslice will be removed in favor of std.experimental.ndslice, but redirection imports will work during long deprecation period (like in Phobos).

Thank you for the star)

On Tuesday, 13 September 2016 at 14:14:01 UTC, Ilya Yaroshenko wrote:
>
> 1. findRoot. D implementation is significantly better then 98% of others  for the problem because the problem behaves like pathological. Thanks to ieeeMean
> 2. logmdigamma
> 3. logmdigammaInverse

Damn, I didn't even realize that std.numeric had a root function!

On Tuesday, 13 September 2016 at 15:19:07 UTC, jmh530 wrote:
> On Tuesday, 13 September 2016 at 14:14:01 UTC, Ilya Yaroshenko wrote:
>>
>> 1. findRoot. D implementation is significantly better then 98% of others  for the problem because the problem behaves like pathological. Thanks to ieeeMean
>> 2. logmdigamma
>> 3. logmdigammaInverse
>
> Damn, I didn't even realize that std.numeric had a root function!

The next DMD release will also have findLocalMin

On Tuesday, 13 September 2016 at 16:22:31 UTC, Ilya Yaroshenko wrote:
>
> The next DMD release will also have findLocalMin

Cool.