[
https://issues.apache.org/jira/browse/MATH-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Luc Maisonobe resolved MATH-1036.
---------------------------------
Resolution: Fixed
Fix Version/s: 3.3
Fixed in subversion as of r1536073.
The name of the class has been changed to SparseGradient and the factory
methods for constant and variables changed from get to create.
The RealFieldElement<SparseGradient> interface is implemented, meaning
instances can be used in the same kind of computation as all other field
elements, including with functions like trigonometric, hyperbolic or power
functions. The high accuracy linear combinations are also available for these
instances.
Thanks for the report and the patch.
> GradStructure: A DerivativeStructure that works for large number of variables.
> ------------------------------------------------------------------------------
>
> Key: MATH-1036
> URL: https://issues.apache.org/jira/browse/MATH-1036
> Project: Commons Math
> Issue Type: Bug
> Reporter: Ajo Fod
> Fix For: 3.3
>
> Attachments: grad-patch.txt, GradStructure.java
>
>
> Ran into problems with DerivativeStructure today. It requires a lot of
> memory with 6000 or so independent variables. The problem starts with the
> number of DSCompiler objects instantiated.
> Here is a faster/leaner(less memory) GradStructure that only computes up to
> the first derivative for the case where there are a large number of
> independent variables.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
