[mlpack-svn] [MLPACK] #318: refactor LRSDP into separate LRSDPFunction class

MLPACK Trac trac at coffeetalk-1.cc.gatech.edu
Fri Mar 7 02:41:54 EST 2014


#318: refactor LRSDP into separate LRSDPFunction class
----------------------------------------+-----------------------------------
  Reporter:  rcurtin                    |        Owner:     
      Type:  defect                     |       Status:  new
  Priority:  trivial                    |    Milestone:     
 Component:  mlpack                     |   Resolution:     
  Keywords:  lrsdp, api, auglagrangian  |     Blocking:     
Blocked By:                             |  
----------------------------------------+-----------------------------------

Comment (by abhishek):

 Thanks rcurtin for suggesting this.
 We can put the implementation of LRSDPFunction lrsdp_function.cpp instead
 of lrsd_function_impl.hpp. It will be better because LRSDPFunction is not
 a template class. We can do it by making small changes
 1. rename the lrsdp_function_impl.cpp with lrsdp_function.cpp
 2. remove the line #include lrsdp_function_impl.hpp at the bottom of
 lrsdp_function.hpp
 3. Add following files in CMakeLists.txt
 lrsdp_function.hpp
 lrsdp_function.cpp
 4. Delete the following lines
 #ifndef __MLPACK_CORE_OPTIMIZERS_LRSDP_LRSDP_FUNCTION_IMPL_HPP
 #define __MLPACK_CORE_OPTIMIZERS_LRSDP_LRSDP_FUNCTION_IMPL_HPP

 #endif

 Similarly we can change lrsdp_impl.hpp to lrsdp.cpp which was already done
 in svn trunk.
 I have attached the lrsdp.zip in the attachment with the changes.

 Why lrsdp_impl.hpp is present if lrsdp.cpp already has implementation of
 Custom specializations of the AugmentedLagrangianFunction for the LRSDP
 case ?

 Replying to [comment:2 rcurtin]:
 > Why is the implementation of LRSDPFunction in lrsdp_impl.hpp despite the
 fact that it is not a templated class?

-- 
Ticket URL: <http://trac.research.cc.gatech.edu/fastlab/ticket/318#comment:3>
MLPACK <www.fast-lab.org>
MLPACK is an intuitive, fast, and scalable C++ machine learning library developed at Georgia Tech.


More information about the mlpack-svn mailing list