[Insight-users] Iteration Monitoring for LBFGSB optimization method

[Luis Ibanez]

Hi Youbing,

Thanks for pointing this out.

This problem was fixed on March 15 2008:
http://www.itk.org/cgi-bin/viewcvs.cgi/Examples/Registration/DeformableRegistration8.cxx?root=Insight&r1=1.17&r2=1.18

Please replace the current line 110 of DeformableRegistration8.cxx

    std::cout << optimizer->GetValue() << "   ";

with the following:

    std::cout << optimizer->GetCachedValue() << "   ";


The LBFGSBOptimizer does iteration reporting by using the function
adaptor, and that forces it to report a "cached" function value.


BTW:
You may want to simply use ITK 3.6 that will be released later today.


     Regards,


        Luis


----------------
Youbing wrote:
> Hi all,
> 
> I am learning how to use LBFGSB (itk::LBFGSOptimizer.h). I ran the 
> example "DeformableRegistration8.cxx" and found all current value of the 
> cost function are zero (shown below). The three values correspond to 
> iteration number, current cost function value, and 
> InfinityNormOfProjectedGradient.
> 
> Starting Registration
> 
> 0   0   0
> 
> 0   0   0
> 
> 0   0   0.000960382
> 
> 1   0   0.000960382
> 
> 1   0   0.00139707
> 
> 2   0   0.00139707
> 
> 2   0   0.00139707
> 
> 2   0   0.00139707
> 
> 2   0   0.00100493
> 
> 3   0   0.00100493
> 
> 3   0   0.00100493
> 
> 3   0   0.00100493
> 
> 3   0   0.00100493
> 
> 3   0   0.00100493
> 
> 3   0   0.00100493
> 
> 3   0   0.00100493
> 
> The itk version I used is “InsightToolkit_2_25_2008”, not the released 
> version. Is it caused by the itk version? Could anyone give me some 
> hints? Thanks!
> 
> Youbing
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users