[Insight-users] Why itk::LBFGSBOptimizer and itk::LBFGSOptimizer are classed as single-variant o

[Ali -]

Luis,

Thanks for pointing out where the 'implementation problem :)' is located. I'll try to extend the optimiser for a multi0variant version.

Since both ITK and VNL implementation are only wrappers, it should be starightforward to implement this. However, since the access to lbfgs is through VNL, it makes sense to implement both wrappers. The complexity starts where we need to extend some part of VNL: assume at some point VNL developers provide a featue like this, in this case (1) our implementation in VNL is redundant, and (2) the ITK wrapper should point to the official implementation. I am not sure what would be the best policy to choose.


PS It would be great if you could provide your opinion about my other post on filters implementing image to (ir)regular grid feature.

> 
> 
> 
> Hi Ali,
> 
> 
> 1) It is not a "problem", it is an "implementation feature".   :-)
> 
> 
> 
> 2) It is not in ITK, it is in VXL,
> 
>     Please look at the implementation in the file:
> 
>          Insight/Utilities/vxl/core/vnl/vnl_lbfgs.h
> 
>     Where the optimizer takes as input a vnl_cost_function
>     object, that derives from the vnl_unary_function class.
> 
> 
> 
> 3) It will be great to have multi-variate implementations
>     of these optimizers. If you are motivated to write one,
>     it will make a great contribution to the Insight Journal.
> 
> 
> 
>      Regards,
> 
> 
>         Luis
> 
> 
> 
> -------------
> Ali - wrote:
> > Hi,
> > 
> > The VNL wrappers itk::LBFGSBOptimizer and itk::LBFGSOptimizer are 
> > classified as single-variant optimiser, while being based on the 
> > quasi-Newton method, they are multi-variant optimisers by nature -- see 
> > http://www.alglib.net/optimization/ . Is this the ITK wrapper problem or 
> > a VNL issue?
> > 
> > _______________________________________________
> > Insight-users mailing list
> > Insight-users at itk.org
> > http://www.itk.org/mailman/listinfo/insight-users

_________________________________________________________________
Celeb spotting – Play CelebMashup and win cool prizes
https://www.celebmashup.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://public.kitware.com/pipermail/insight-users/attachments/20071107/32ac61a0/attachment-0001.html