commons-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Thomas Neidhart (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (MATH-1074) Refactor fixed size chromosomes
Date Mon, 09 Dec 2013 16:24:07 GMT

    [ https://issues.apache.org/jira/browse/MATH-1074?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13843286#comment-13843286
] 

Thomas Neidhart commented on MATH-1074:
---------------------------------------

It depends.

The Chromosome class could be similar to a CharSequence, which provides methods like length()
and substring() to get parts of the sequence. The actual underlying data would be completely
abstracted, thus avoiding to create a List<Double> which is quite inefficient, especially
for large populations.

I have not made up my mind, I was just working on some examples last weekend (will add another
one hopefully tonight), and found the current API very cumbersome to use. Will need to experiment
with more use-cases to see what is really useful.

> Refactor fixed size chromosomes
> -------------------------------
>
>                 Key: MATH-1074
>                 URL: https://issues.apache.org/jira/browse/MATH-1074
>             Project: Commons Math
>          Issue Type: Improvement
>    Affects Versions: 3.2
>            Reporter: Thomas Neidhart
>
> Currently, only chromosomes derived from AbstractListChromosome can be used for most
of the crossover policies. This is quite inconvenient to use, especially when working with
primitive types.
> Need to think about ways to refactor this for better support of other datatypes, e.g.
CharSequence or primitive arrays.



--
This message was sent by Atlassian JIRA
(v6.1.4#6159)

Mime
View raw message