[Fwd: Re: Error in LUDecomposition?]
- Subject: [Fwd: Re: Error in LUDecomposition?]
- From: Ronald Boisvert <boisvert@nist.gov>
- Date: Thu, 20 Aug 2009 10:05:05 -0400
- Content-Transfer-Encoding: 7bit
- Content-Type: text/plain; charset=ISO-8859-1; format=flowed
- User-Agent: Thunderbird 2.0.0.22 (Windows/20090605)
Sender: Anders Peterson <anders@optimatika.se>
Subject: Re: Error in LUDecomposition?
Believe you do need to change that. I did in ojAlgo's copy/version of
JAMA.
http://ojalgo.org/generated/org/ojalgo/matrix/jama/package-summary.html
/Anders
On 20 aug 2009, at 14.48, Rupert Parson wrote:
> OK - I've looked into the code a little more closely and see it is due
> to the fact that the LUDecomposition class only works for the case m
>> = n, and not for m < n, despite what is stated in its Javadoc.
>
> I am wondering when this might be rectified? I note that Ron Boisvert
> has commented on this in the past.
>
> Many thanks,
>
> Rupert
>
>
> 2009/8/20 Rupert Parson <rupert.parson@gmail.com>:
>> I am getting an java.lang.ArrayIndexOutOfBoundsException: at
>> Jama.LUDecomposition.<init>(LUDecomposition.java:107).
>>
>> Looking at the code, I see that this line is:
>>
>> if (j < m & LU[j][j] != 0.0) {
>>
>> and I am wondering if it should actually read:
>>
>> if (j < m && LU[j][j] != 0.0) {
>>
>> as using the non-short-circuiting logical operator "&" rather than
>> "&&" would seem to make the first condition of the conditional
>> statement rather redundant!?
>>
>> Many thanks,
>>
>> Rupert Parson
>>
>
>
>
>
>
>
Date Index |
Thread Index |
Problems or questions? Contact list-master@nist.gov