www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.announce - libmir: recent updates

reply 9il <ilyayaroshenko gmail.com> writes:
Mir Algorithm http://mir-algorithm.dub.pm  - v2.0.0, new ndslice 
API, bitwise accelerated mir.algorithm.iteration, new mir.bitop 
module

Mir Optim http://mir-optim.dub.pm - v1.1.0, full rework of main 
LM loop, two times faster then Alglib for calibration of multiple 
curves.

Mir Random http://mir-random.dub.pm - v1.0.0, upgrade ndslice API

Mir http://mir.dub.pm - v2.0.0, upgrade ndslice API, `range` for 
n-dimensional random variables.

Asdf http://asdf.dub.pm - v0.2.5, null support for associative 
arrays (John Colvin)

Part of this work has been sponsored by Symmetry Investments and 
Kaleidic Associates.

http://symmetryinvestments.com
https://github.com/kaleidicassociates

Best,
Ilya Yaroshenko
Sep 09
next sibling parent reply Shigeki Karita <shigekikarita gmail.com> writes:
On Sunday, 9 September 2018 at 17:24:49 UTC, 9il wrote:
 Mir Algorithm http://mir-algorithm.dub.pm  - v2.0.0, new 
 ndslice API, bitwise accelerated mir.algorithm.iteration, new 
 mir.bitop module

 Mir Optim http://mir-optim.dub.pm - v1.1.0, full rework of main 
 LM loop, two times faster then Alglib for calibration of 
 multiple curves.

 Mir Random http://mir-random.dub.pm - v1.0.0, upgrade ndslice 
 API

 Mir http://mir.dub.pm - v2.0.0, upgrade ndslice API, `range` 
 for n-dimensional random variables.

 Asdf http://asdf.dub.pm - v0.2.5, null support for associative 
 arrays (John Colvin)

 Part of this work has been sponsored by Symmetry Investments 
 and Kaleidic Associates.

 http://symmetryinvestments.com
 https://github.com/kaleidicassociates

 Best,
 Ilya Yaroshenko
great. numir (numpy like helper functions library) also supports mir-algorithm v2 today https://github.com/libmir/numir/releases
Sep 09
parent jmh530 <john.michael.hall gmail.com> writes:
On Monday, 10 September 2018 at 04:50:37 UTC, Shigeki Karita 
wrote:
 [snip]

 great. numir (numpy like helper functions library) also 
 supports mir-algorithm v2 today

 https://github.com/libmir/numir/releases
Glad to hear it. I hope the v2 API helps with explaining mir.
Sep 10
prev sibling parent reply Arredondo <arm.plus gmail.com> writes:
On Sunday, 9 September 2018 at 17:24:49 UTC, 9il wrote:
 Mir Algorithm http://mir-algorithm.dub.pm  - v2.0.0, new 
 ndslice API,
Does new API mean that it's not backwards compatible? if so, where can we find the documentation for this new API? Arredondo.
Sep 10
parent reply 9il <ilyayaroshenko gmail.com> writes:
On Monday, 10 September 2018 at 12:55:01 UTC, Arredondo wrote:
 On Sunday, 9 September 2018 at 17:24:49 UTC, 9il wrote:
 Mir Algorithm http://mir-algorithm.dub.pm  - v2.0.0, new 
 ndslice API,
Does new API mean that it's not backwards compatible? if so, where can we find the documentation for this new API? Arredondo.
http://docs.algorithm.dlang.io - docs for recent Mir Algorithm release. The main difference is that Slice definition is changed to struct mir_slice(Iterator_, size_t N_ = 1, SliceKind kind_ = Contiguous) alias Slice = mir_slice; The parameters order was reversed, last two params are optional now. For example, see the PR for ndslice API update for old main Mir repo. https://github.com/libmir/mir/pull/397/files#diff-399a70373923e55f87a86928eb3cc731L43 Similar situation with Series definition.
Sep 10
parent reply jmh530 <john.michael.hall gmail.com> writes:
On Monday, 10 September 2018 at 14:33:11 UTC, 9il wrote:
 On Monday, 10 September 2018 at 12:55:01 UTC, Arredondo wrote:
 On Sunday, 9 September 2018 at 17:24:49 UTC, 9il wrote:
 Mir Algorithm http://mir-algorithm.dub.pm  - v2.0.0, new 
 ndslice API,
Does new API mean that it's not backwards compatible? if so, where can we find the documentation for this new API? Arredondo.
http://docs.algorithm.dlang.io - docs for recent Mir Algorithm release. The main difference is that Slice definition is changed to struct mir_slice(Iterator_, size_t N_ = 1, SliceKind kind_ = Contiguous) alias Slice = mir_slice; The parameters order was reversed, last two params are optional now. For example, see the PR for ndslice API update for old main Mir repo. https://github.com/libmir/mir/pull/397/files#diff-399a70373923e55f87a86928eb3cc731L43 Similar situation with Series definition.
mir-algorithm and lubeck do not interact well right now on run.dlang.org as lubeck depends on the earlier API. The maintainer of that might consider adding an older version of mir-algorithm until the situation is resolved.
Sep 10
parent reply jmh530 <john.michael.hall gmail.com> writes:
On Monday, 10 September 2018 at 17:42:16 UTC, jmh530 wrote:
 [snip]

 mir-algorithm and lubeck do not interact well right now on 
 run.dlang.org as lubeck depends on the earlier API. The 
 maintainer of that might consider adding an older version of 
 mir-algorithm until the situation is resolved.
Hmm, it looks like all you have to do just change the dependency to whatever works with lubeck and it works fine. Maybe it can be temporarily adjusted so that adding lubeck as a dependency adds the correct version of mir algorithm by default?
Sep 10
parent 9il <ilyayaroshenko gmail.com> writes:
On Monday, 10 September 2018 at 20:22:08 UTC, jmh530 wrote:
 On Monday, 10 September 2018 at 17:42:16 UTC, jmh530 wrote:
 [snip]

 mir-algorithm and lubeck do not interact well right now on 
 run.dlang.org as lubeck depends on the earlier API. The 
 maintainer of that might consider adding an older version of 
 mir-algorithm until the situation is resolved.
Hmm, it looks like all you have to do just change the dependency to whatever works with lubeck and it works fine. Maybe it can be temporarily adjusted so that adding lubeck as a dependency adds the correct version of mir algorithm by default?
Thanks for the report. Lubeck has been tagged to release. Plus tour update https://github.com/dlang-tour/english/pull/269 (requires mir v2.0.2)
Sep 11