commons-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Alexander Nozik (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (MATH-998) Minor design suggestions for RealVector and MultivariateDifferentiableFunction
Date Fri, 28 Jun 2013 11:10:20 GMT

    [ https://issues.apache.org/jira/browse/MATH-998?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13695394#comment-13695394
] 

Alexander Nozik commented on MATH-998:
--------------------------------------

OK, it is clear about RealVector.

For example something like this:

{code}
interface VetorToDoubleMapping{
    double value(RealVector vector)
    RealVector gradient(RealVector vector)
}
{code}

Very useful for multivariate optimization like simplex.

And something like this:

{code}
interface VetorToVectorMapping{
    RealVector value(RealVector vector)
    RealMatrix jacobian(RealVector vector)
}
{code}

I will try to post everything properly next time. Thanks for the comment. 
                
> Minor design suggestions for RealVector and MultivariateDifferentiableFunction
> ------------------------------------------------------------------------------
>
>                 Key: MATH-998
>                 URL: https://issues.apache.org/jira/browse/MATH-998
>             Project: Commons Math
>          Issue Type: Wish
>    Affects Versions: 3.1, 3.2
>            Reporter: Alexander Nozik
>            Priority: Minor
>
> I am working on some kind of data analysis framework and using Math as a platform. There
are some minor inconveniences in the architecture wich i want to note.
> The first problem concerning RealVector clas. It has a lot of features (too much from
my point of view) but does not implements any simple interface. Also it is not connected with
geometry package wich is not a problem in my case, but seems strange.
> The second problem is MultivariateDifferentiableFunction. It extends MultivariateFunction,
but since 3.1 it uses DerivativeStructure wich is good for symbolic calculations but is not
convinient for empirical and user defined functions. Perhaps it would be usefull to introduce
some additional interfaces for vector to double and vector to vector differentiable functions?

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message