[Insight-users] vnl_svd suspicious return value error
Hans Johnson
hans-johnson at uiowa.edu
Thu Apr 26 17:31:36 EDT 2007
Sandy,
I've had this same problem, and unfortunatly I was forced to make a copy of
the vnl_svd files and delve into the netlib code making the recommended
changes to increase stability.
Hans
--
Hans J. Johnson, Ph.D.
Hans-johnson at uiowa.edu
278 GH
The University of Iowa
Iowa City, IA 52241
(319) 353 8587
On 4/26/07 3:06 PM, "frankmiller at jhmi.edu" <frankmiller at jhmi.edu> wrote:
> Sandy,
>
> I've had issues with vnl_svd on my Xeon system in the past that went
> away when I compiled itk with more conservative optimizer settings.
> With gcc the problem went away when I used -O2 instead of -O3.
>
> Give it a try.
>
> Frank
>
> On Thu, Apr 26, 2007 at 01:13:08PM -0400, Sandhitsu Das wrote:
>> While executing the following code segment
>>
>> this->m_RefinedLatticeCoefficients[i] = ( vnl_svd<RealType>
>> ( R ).solve( C ) ).extract( 2, C.cols() );
>>
>> I get the following runtime error:
>>
>> /InsightToolkit-2.8.1/Utilities/vxl/core/vnl/algo/vnl_svd.txx:
>> suspicious return value (4) from SVDC
>> /InsightToolkit-2.8.1/Utilities/vxl/core/vnl/algo/vnl_svd.txx: M is 4x4
>> M = [ ...
>> 0.1666666666667 0.6666666666667 0.1666666666667 0
>> -1.0000000000000 0 1.0000000000000 0
>> 2.0000000000000 -4.0000000000000 2.0000000000000 0
>> -1.3333333333333 4.0000000000000 -4.0000000000000 1.3333333333333 ]
>>
>> This happens on a variety of linux systems (P4, Xeon, Opteron), but
>> not on Mac OS X. RealType is double. Is there a fix for this other
>> than digging dip into netlib code and changing things as suggested in
>> vnl_svd.txx ?
>>
>> Thanks for any help,
>> Sandy
>>
>>
>> _______________________________________________
>> Insight-users mailing list
>> Insight-users at itk.org
>> http://www.itk.org/mailman/listinfo/insight-users
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
More information about the Insight-users
mailing list