FAQ
Hello all,

After a short discussion with Dan Kortschak, we decided that it is
necessary to discuss Go on science at the list.

The thing started with an email about the 2 different libraries we have for
linear algebra and matrix operations
(go.matrix and biogo.matrix).

There are important things to talk about, for instance, compatibility
between our libraries. For instance, some
near-zero-cost biogo.matrix <--> go.matrix dense matrix converting
function, which should be possible since their internal representations are
already very similar and would be identical with some changes that Dan has
planned for biogo (if I understood him right). On the same line, it would
be important that we keep one or (in the matrix case)
two recommended standards for common things, don't have a proliferation of
similar, incompatible libraries. IMO it would be nice that the go
scientific libraries build on each other as with numpy and scipy.

Because of the previous, I think there is a place for a go-science
community of library developers, for instance, in G+. We might gather at
some point in a hangout or IRC to talk about compatibility and coordinate
efforts. We could also, maybe at some point, do other things to help the
cause, like some kind of wiki. Probably none of us has a lot of available
time, but a G+ community or similar shouldn't take much of it.

Thoughts?

Cheers,

Raúl

--

Search Discussions

  • Sebastien Binet at Jan 4, 2013 at 10:28 pm
    hi there,
    On Fri, Jan 4, 2013 at 11:22 PM, Raul Mera wrote:
    Hello all,

    After a short discussion with Dan Kortschak, we decided that it is necessary
    to discuss Go on science at the list.

    The thing started with an email about the 2 different libraries we have for
    linear algebra and matrix operations
    (go.matrix and biogo.matrix).

    There are important things to talk about, for instance, compatibility
    between our libraries. For instance, some
    near-zero-cost biogo.matrix <--> go.matrix dense matrix converting function,
    which should be possible since their internal representations are already
    very similar and would be identical with some changes that Dan has planned
    for biogo (if I understood him right). On the same line, it would be
    important that we keep one or (in the matrix case) two recommended standards
    for common things, don't have a proliferation of similar, incompatible
    libraries. IMO it would be nice that the go scientific libraries build on
    each other as with numpy and scipy.

    Because of the previous, I think there is a place for a go-science community
    of library developers, for instance, in G+. We might gather at some point in
    a hangout or IRC to talk about compatibility and coordinate efforts. We
    could also, maybe at some point, do other things to help the cause, like
    some kind of wiki. Probably none of us has a lot of available time, but a
    G+ community or similar shouldn't take much of it.

    Thoughts?
    sounds like a plan.
    having a set of BSD-licensed scientific libraries in Go would be great.
    count me in.

    -s

    --
  • Kosztka Imre Dávid at Jan 4, 2013 at 10:38 pm
    Please count me as well.


    2013/1/4 Sebastien Binet <seb.binet@gmail.com>
    hi there,
    On Fri, Jan 4, 2013 at 11:22 PM, Raul Mera wrote:
    Hello all,

    After a short discussion with Dan Kortschak, we decided that it is necessary
    to discuss Go on science at the list.

    The thing started with an email about the 2 different libraries we have for
    linear algebra and matrix operations
    (go.matrix and biogo.matrix).

    There are important things to talk about, for instance, compatibility
    between our libraries. For instance, some
    near-zero-cost biogo.matrix <--> go.matrix dense matrix converting function,
    which should be possible since their internal representations are already
    very similar and would be identical with some changes that Dan has planned
    for biogo (if I understood him right). On the same line, it would be
    important that we keep one or (in the matrix case) two recommended standards
    for common things, don't have a proliferation of similar, incompatible
    libraries. IMO it would be nice that the go scientific libraries build on
    each other as with numpy and scipy.

    Because of the previous, I think there is a place for a go-science community
    of library developers, for instance, in G+. We might gather at some point in
    a hangout or IRC to talk about compatibility and coordinate efforts. We
    could also, maybe at some point, do other things to help the cause, like
    some kind of wiki. Probably none of us has a lot of available time, but a
    G+ community or similar shouldn't take much of it.

    Thoughts?
    sounds like a plan.
    having a set of BSD-licensed scientific libraries in Go would be great.
    count me in.

    -s

    --


    --
    Name : Kosztka Imre Dávid
    E-mail: kosztkaid@gmail.com
    Phone number: +36309213462
    Mailing address: H-3700, Hungary Kazincbarcika Szeder utca 2.

    --
  • Kun Li at Jan 5, 2013 at 2:54 am
    Would be ecstatic to see a comprehensive scientific library written in Go,
    like numpy and scipy
    On Friday, January 4, 2013 3:22:00 PM UTC-7, Raul Mera wrote:

    Hello all,

    After a short discussion with Dan Kortschak, we decided that it is
    necessary to discuss Go on science at the list.

    The thing started with an email about the 2 different libraries we have
    for linear algebra and matrix operations
    (go.matrix and biogo.matrix).

    There are important things to talk about, for instance, compatibility
    between our libraries. For instance, some
    near-zero-cost biogo.matrix <--> go.matrix dense matrix converting
    function, which should be possible since their internal representations are
    already very similar and would be identical with some changes that Dan has
    planned for biogo (if I understood him right). On the same line, it would
    be important that we keep one or (in the matrix case)
    two recommended standards for common things, don't have a proliferation of
    similar, incompatible libraries. IMO it would be nice that the go
    scientific libraries build on each other as with numpy and scipy.

    Because of the previous, I think there is a place for a go-science
    community of library developers, for instance, in G+. We might gather at
    some point in a hangout or IRC to talk about compatibility and coordinate
    efforts. We could also, maybe at some point, do other things to help the
    cause, like some kind of wiki. Probably none of us has a lot of available
    time, but a G+ community or similar shouldn't take much of it.

    Thoughts?

    Cheers,

    Raúl
    --
  • Job van der Zwan at Jan 5, 2013 at 1:09 pm

    On Friday, 4 January 2013 23:22:00 UTC+1, Raul Mera wrote:

    Thoughts?
    - I'm not a scientist (interaction design)
    - Most of my friends who also code, however, are (computer
    science, physics, math, chemistry, biology)
    - Whenever I want to try something... complicated, I turn to them for help
    - I like Go
    - They don't use Go yet
    - Good support for scientific stuff might change their mind

    So in short: even for non-scientists this project would be beneficial.
    Please go ahead with it!

    --
  • John Asmuth at Jan 5, 2013 at 3:09 pm
    Sounds good to me! I'd be happy to help.
    On Friday, January 4, 2013 5:22:00 PM UTC-5, Raul Mera wrote:

    Hello all,

    After a short discussion with Dan Kortschak, we decided that it is
    necessary to discuss Go on science at the list.

    The thing started with an email about the 2 different libraries we have
    for linear algebra and matrix operations
    (go.matrix and biogo.matrix).

    There are important things to talk about, for instance, compatibility
    between our libraries. For instance, some
    near-zero-cost biogo.matrix <--> go.matrix dense matrix converting
    function, which should be possible since their internal representations are
    already very similar and would be identical with some changes that Dan has
    planned for biogo (if I understood him right). On the same line, it would
    be important that we keep one or (in the matrix case)
    two recommended standards for common things, don't have a proliferation of
    similar, incompatible libraries. IMO it would be nice that the go
    scientific libraries build on each other as with numpy and scipy.

    Because of the previous, I think there is a place for a go-science
    community of library developers, for instance, in G+. We might gather at
    some point in a hangout or IRC to talk about compatibility and coordinate
    efforts. We could also, maybe at some point, do other things to help the
    cause, like some kind of wiki. Probably none of us has a lot of available
    time, but a G+ community or similar shouldn't take much of it.

    Thoughts?

    Cheers,

    Raúl
    --
  • Maxim Khitrov at Jan 5, 2013 at 4:00 pm

    On Fri, Jan 4, 2013 at 5:22 PM, Raul Mera wrote:
    Hello all,

    After a short discussion with Dan Kortschak, we decided that it is necessary
    to discuss Go on science at the list.

    The thing started with an email about the 2 different libraries we have for
    linear algebra and matrix operations
    (go.matrix and biogo.matrix).

    There are important things to talk about, for instance, compatibility
    between our libraries. For instance, some
    near-zero-cost biogo.matrix <--> go.matrix dense matrix converting function,
    which should be possible since their internal representations are already
    very similar and would be identical with some changes that Dan has planned
    for biogo (if I understood him right). On the same line, it would be
    important that we keep one or (in the matrix case) two recommended standards
    for common things, don't have a proliferation of similar, incompatible
    libraries. IMO it would be nice that the go scientific libraries build on
    each other as with numpy and scipy.

    Because of the previous, I think there is a place for a go-science community
    of library developers, for instance, in G+. We might gather at some point in
    a hangout or IRC to talk about compatibility and coordinate efforts. We
    could also, maybe at some point, do other things to help the cause, like
    some kind of wiki. Probably none of us has a lot of available time, but a
    G+ community or similar shouldn't take much of it.

    Thoughts?
    Sounds like a great plan, though I suggest keeping most of the
    discussion on golang-nuts to reach a wider audience.

    The main barrier to adoption of tools like scipy and Octave, at least
    in my organization, is the lack of interactive plotting. Scipy seems
    to do this a bit now (been a long time since I've looked at it), but
    MATLAB is, unfortunately, still the king here. We spend tens, if not
    hundreds, of thousands of dollars on the base licenses and various
    toolboxes each year. If this problem can be solved, perhaps by
    combining existing plot packages, the http server, and HTML5 canvas,
    then Go could become a major force in scientific data analysis.

    - Max

    --
  • Miki Tebeka at Jan 6, 2013 at 4:19 pm

    On Saturday, January 5, 2013 8:00:01 AM UTC-8, Maxim Khitrov wrote:

    The main barrier to adoption of tools like scipy and Octave, at least
    in my organization, is the lack of interactive plotting. Scipy seems
    to do this a bit now (been a long time since I've looked at it)
    A bit OT, but Scipy (with ipython + matplotlib) has great interactive
    plotting capabilities now.
    One thing we can think about is something like the IPython notebook<http://vimeo.com/53051817>
    .

    --
  • Ethan Burns at Jan 6, 2013 at 9:02 pm
    There have been a few mentions of interactive plotting in this thread.
    Someone contacted me a while ago about making an interactive front-end for
    my plotting package, Plotinum (code.google.com/p/plotinum), but I am not
    sure that anything has come from it. If anyone is interested in taking up
    this effort, I would be willing to help out a little bit on the Plotinum
    end of things.

    By the way, I made a small prototype a while ago just to see how difficult
    it may be to create an interactive front-end. You can find the code for
    that demo
    here: http://code.google.com/p/eaburns/source/browse/showplot/main.go. One
    thing that I noticed is that it is fairly slow. I seem to recall that
    profiling showed most of the time was going to the software rasterization
    either in draw2d or freetype-go. It should be possible to write an OpenGL
    implementation of the vg interface used by Plotinum. This would be a good
    first step for making an interactive plotting framework for Go.



    Best,
    Ethan
    On Sunday, January 6, 2013 11:19:06 AM UTC-5, Miki Tebeka wrote:
    On Saturday, January 5, 2013 8:00:01 AM UTC-8, Maxim Khitrov wrote:

    The main barrier to adoption of tools like scipy and Octave, at least
    in my organization, is the lack of interactive plotting. Scipy seems
    to do this a bit now (been a long time since I've looked at it)
    A bit OT, but Scipy (with ipython + matplotlib) has great interactive
    plotting capabilities now.
    One thing we can think about is something like the IPython notebook<http://vimeo.com/53051817>
    .
    --
  • Brendan Tracey at Feb 5, 2013 at 12:40 am
    Has anyone written any gradient based optimization algorithms (BFGS, etc.)?
    On Sunday, January 6, 2013 1:02:37 PM UTC-8, Ethan Burns wrote:

    There have been a few mentions of interactive plotting in this thread.
    Someone contacted me a while ago about making an interactive front-end for
    my plotting package, Plotinum (code.google.com/p/plotinum), but I am not
    sure that anything has come from it. If anyone is interested in taking up
    this effort, I would be willing to help out a little bit on the Plotinum
    end of things.

    By the way, I made a small prototype a while ago just to see how difficult
    it may be to create an interactive front-end. You can find the code for
    that demo here:
    http://code.google.com/p/eaburns/source/browse/showplot/main.go. One
    thing that I noticed is that it is fairly slow. I seem to recall that
    profiling showed most of the time was going to the software rasterization
    either in draw2d or freetype-go. It should be possible to write an OpenGL
    implementation of the vg interface used by Plotinum. This would be a good
    first step for making an interactive plotting framework for Go.



    Best,
    Ethan
    On Sunday, January 6, 2013 11:19:06 AM UTC-5, Miki Tebeka wrote:
    On Saturday, January 5, 2013 8:00:01 AM UTC-8, Maxim Khitrov wrote:

    The main barrier to adoption of tools like scipy and Octave, at least
    in my organization, is the lack of interactive plotting. Scipy seems
    to do this a bit now (been a long time since I've looked at it)
    A bit OT, but Scipy (with ipython + matplotlib) has great interactive
    plotting capabilities now.
    One thing we can think about is something like the IPython notebook<http://vimeo.com/53051817>
    .
    --
    You received this message because you are subscribed to the Google Groups "golang-nuts" group.
    To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscribe@googlegroups.com.
    For more options, visit https://groups.google.com/groups/opt_out.
  • Peter A. Cejchan at Feb 5, 2013 at 7:51 am
    I have a working implementation of an Ant Colony Optimization, and
    Simulated Annealig,
    but these are not gradient-based methods.

    ++pac
    On Tuesday, 5 February 2013 01:40:19 UTC+1, Brendan Tracey wrote:

    Has anyone written any gradient based optimization algorithms (BFGS, etc.)?
    On Sunday, January 6, 2013 1:02:37 PM UTC-8, Ethan Burns wrote:

    There have been a few mentions of interactive plotting in this thread.
    Someone contacted me a while ago about making an interactive front-end for
    my plotting package, Plotinum (code.google.com/p/plotinum), but I am not
    sure that anything has come from it. If anyone is interested in taking up
    this effort, I would be willing to help out a little bit on the Plotinum
    end of things.

    By the way, I made a small prototype a while ago just to see how
    difficult it may be to create an interactive front-end. You can find the
    code for that demo here:
    http://code.google.com/p/eaburns/source/browse/showplot/main.go. One
    thing that I noticed is that it is fairly slow. I seem to recall that
    profiling showed most of the time was going to the software rasterization
    either in draw2d or freetype-go. It should be possible to write an OpenGL
    implementation of the vg interface used by Plotinum. This would be a good
    first step for making an interactive plotting framework for Go.



    Best,
    Ethan
    On Sunday, January 6, 2013 11:19:06 AM UTC-5, Miki Tebeka wrote:
    On Saturday, January 5, 2013 8:00:01 AM UTC-8, Maxim Khitrov wrote:

    The main barrier to adoption of tools like scipy and Octave, at least
    in my organization, is the lack of interactive plotting. Scipy seems
    to do this a bit now (been a long time since I've looked at it)
    A bit OT, but Scipy (with ipython + matplotlib) has great interactive
    plotting capabilities now.
    One thing we can think about is something like the IPython notebook<http://vimeo.com/53051817>
    .
    --
    You received this message because you are subscribed to the Google Groups "golang-nuts" group.
    To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscribe@googlegroups.com.
    For more options, visit https://groups.google.com/groups/opt_out.
  • John Asmuth at Jan 5, 2013 at 7:30 pm
    I made a google+ community called "go.science":
    https://plus.google.com/communities/115354877257015196825.
    On Friday, January 4, 2013 5:22:00 PM UTC-5, Raul Mera wrote:

    Hello all,

    After a short discussion with Dan Kortschak, we decided that it is
    necessary to discuss Go on science at the list.

    The thing started with an email about the 2 different libraries we have
    for linear algebra and matrix operations
    (go.matrix and biogo.matrix).

    There are important things to talk about, for instance, compatibility
    between our libraries. For instance, some
    near-zero-cost biogo.matrix <--> go.matrix dense matrix converting
    function, which should be possible since their internal representations are
    already very similar and would be identical with some changes that Dan has
    planned for biogo (if I understood him right). On the same line, it would
    be important that we keep one or (in the matrix case)
    two recommended standards for common things, don't have a proliferation of
    similar, incompatible libraries. IMO it would be nice that the go
    scientific libraries build on each other as with numpy and scipy.

    Because of the previous, I think there is a place for a go-science
    community of library developers, for instance, in G+. We might gather at
    some point in a hangout or IRC to talk about compatibility and coordinate
    efforts. We could also, maybe at some point, do other things to help the
    cause, like some kind of wiki. Probably none of us has a lot of available
    time, but a G+ community or similar shouldn't take much of it.

    Thoughts?

    Cheers,

    Raúl
    --
  • Raul Mera at Jan 5, 2013 at 9:22 pm
    Great, John, I'm joining.

    Also, I don't quite thing we need a wide audience for this, since its
    mostly to keep developers coordinated.

    Of course anyone interested can join the community :-)


    On Saturday, January 5, 2013 9:30:48 PM UTC+2, John Asmuth wrote:

    I made a google+ community called "go.science":
    https://plus.google.com/communities/115354877257015196825.
    On Friday, January 4, 2013 5:22:00 PM UTC-5, Raul Mera wrote:

    Hello all,

    After a short discussion with Dan Kortschak, we decided that it is
    necessary to discuss Go on science at the list.

    The thing started with an email about the 2 different libraries we have
    for linear algebra and matrix operations
    (go.matrix and biogo.matrix).

    There are important things to talk about, for instance, compatibility
    between our libraries. For instance, some
    near-zero-cost biogo.matrix <--> go.matrix dense matrix converting
    function, which should be possible since their internal representations are
    already very similar and would be identical with some changes that Dan has
    planned for biogo (if I understood him right). On the same line, it would
    be important that we keep one or (in the matrix case)
    two recommended standards for common things, don't have a proliferation of
    similar, incompatible libraries. IMO it would be nice that the go
    scientific libraries build on each other as with numpy and scipy.

    Because of the previous, I think there is a place for a go-science
    community of library developers, for instance, in G+. We might gather at
    some point in a hangout or IRC to talk about compatibility and coordinate
    efforts. We could also, maybe at some point, do other things to help the
    cause, like some kind of wiki. Probably none of us has a lot of available
    time, but a G+ community or similar shouldn't take much of it.

    Thoughts?

    Cheers,

    Raúl
    --
  • Ryan.bressler at Jan 5, 2013 at 8:51 pm
    I joined the G+ community and excited to see how this pans out.

    The lab I work for at the Institute for Systems Biology are big fans of go.
    We use it for our job distribution system (http://code.google.com/p/golem/)
    and have used it to serve web sockets streaming data into interactive html5
    (d3 and protovis) visualizations but rarely use it for analysis.

    One place I have started using it analytically is for tree based machine
    learning (ie random forest). I haven't implemented a full algorithm yet
    (just messy code to parse, apply, and gather statistics
    about existing forests: https://github.com/ryanbressler/CloudForest) but I
    think that go is the best language I have used for this sort of
    thing...easy/fast to prototype in like python but you get lower level
    things like pointers for building data structures and the performance of a
    compiled code for loops and recursion. I expect an equivalent to
    scikits.learn containing common data analysis algorithms could get a pretty
    active community of contributors quickly once the matrix stuff has matured
    a bit. It would be a great code base for people looking to apply stuff and
    also a great starting point for researchers tweaking algorithms.

    Ryan

    On Friday, January 4, 2013 3:22:00 PM UTC-7, Raul Mera wrote:

    Hello all,

    After a short discussion with Dan Kortschak, we decided that it is
    necessary to discuss Go on science at the list.

    The thing started with an email about the 2 different libraries we have
    for linear algebra and matrix operations
    (go.matrix and biogo.matrix).

    There are important things to talk about, for instance, compatibility
    between our libraries. For instance, some
    near-zero-cost biogo.matrix <--> go.matrix dense matrix converting
    function, which should be possible since their internal representations are
    already very similar and would be identical with some changes that Dan has
    planned for biogo (if I understood him right). On the same line, it would
    be important that we keep one or (in the matrix case)
    two recommended standards for common things, don't have a proliferation of
    similar, incompatible libraries. IMO it would be nice that the go
    scientific libraries build on each other as with numpy and scipy.

    Because of the previous, I think there is a place for a go-science
    community of library developers, for instance, in G+. We might gather at
    some point in a hangout or IRC to talk about compatibility and coordinate
    efforts. We could also, maybe at some point, do other things to help the
    cause, like some kind of wiki. Probably none of us has a lot of available
    time, but a G+ community or similar shouldn't take much of it.

    Thoughts?

    Cheers,

    Raúl
    --
  • Visan Ovidiu at Jan 5, 2013 at 9:31 pm
    Count me in. I'm glad to help, more libraries make Go become the language
    we all need.

    --

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupgolang-nuts @
categoriesgo
postedJan 4, '13 at 10:22p
activeFeb 5, '13 at 7:51a
posts15
users13
websitegolang.org

People

Translate

site design / logo © 2022 Grokbase