Can't find test data

[stnava]

Hi - i am trying to compile test this very promising contribution ... just looking for a little help with this:

-- Found Eigen version 3.2.1: /usr/local/include/eigen3

CMake Error at CMakeLists.txt:28 (file): file COPY cannot find "/Users/stnava/code/Dense_HODLR/test/data".

CMake Error at CMakeLists.txt:29 (file): file COPY cannot find "/Users/stnava/code/Dense_HODLR/benchmarks/ACA".

CMake Error at CMakeLists.txt:30 (file): file COPY cannot find "/Users/stnava/code/Dense_HODLR/benchmarks/ACA_Blade".

[amiraa127]

Hi. Unfortunately, I have not included the test data in the repository as they are very large. If you just want to compile the package, kindly comment out the file(copy.......) lines in the CMakeLists.txt in the root directory. Because of the fact that I haven't included the test files, you will not be able to run all the HODLR tests available, although some can still be run.

Are you trying to just run the test or do you have any other applications in mind? I might be able to help more you more if I know more details about your specific application.

[amiraa127]

Since the package is relatively new, I have not modified the test files to be self sufficient. I will do that in the near feature.

[stnava]

i am thinking of porting HODLR or Dense_HODLR to R via an R package ....

brian

On Fri, Nov 21, 2014 at 5:36 PM, amiraa127 notifications@github.com wrote:

Since the package is relatively new, I have not modified the test files to be self sufficient. I will do that in the near feature.

— Reply to this email directly or view it on GitHub https://github.com/amiraa127/Dense_HODLR/issues/1#issuecomment-64049079.

[amiraa127]

Cool. I'll create a standalone test module and will let you know when it's ready. Meanwhile, let me know of any other issues.

Amir

[stnava]

brilliant! On Nov 21, 2014 5:55 PM, "AmirHossein Aminfar" notifications@github.com wrote:

Cool. I'll create a standalone test module and will let you know when it's ready. Meanwhile, let me know of any other issues.

Amir

— Reply to this email directly or view it on GitHub https://github.com/amiraa127/Dense_HODLR/issues/1#issuecomment-64051079.

[stnava]

quick q: how do application areas of HODLR package differ from Dense_HODLR? trying to figure out which to wrap ...

brian

On Fri, Nov 21, 2014 at 6:31 PM, brian avants stnava@gmail.com wrote:

brilliant! On Nov 21, 2014 5:55 PM, "AmirHossein Aminfar" notifications@github.com wrote:

Cool. I'll create a standalone test module and will let you know when it's ready. Meanwhile, let me know of any other issues.

Amir

— Reply to this email directly or view it on GitHub https://github.com/amiraa127/Dense_HODLR/issues/1#issuecomment-64051079 .

[amiraa127]

I'm not sure which two packages you are comparing. If you are comparing the two terms within my repository, I have to say that I might have used these two terms interchangeably but the essentially refer to the same thing. If you are referring to another package, I can give you a better comparison if you give me more details.

Amir

[stnava]

https://github.com/sivaramambikasaran/HODLR

brian

On Fri, Nov 21, 2014 at 11:26 PM, AmirHossein Aminfar < notifications@github.com> wrote:

I'm not sure which two packages you are comparing. If you are comparing the two terms within my repository, I have to say that I might have used these two terms interchangeably but the essentially refer to the same thing. If you are referring to another package, I can give you a better comparison if you give me more details.

Amir

— Reply to this email directly or view it on GitHub https://github.com/amiraa127/Dense_HODLR/issues/1#issuecomment-64068618.

[amiraa127]

This package is written by Sivaram Ambikasaran a former colleague of mine. There are several advantages to my package. First, my package supports a variety of low-rank approximation methods (SVD, partial pivoting LU, full pivoting LU, partial pivoting ACA and Pseudo skeleton). One other advantage of my package is the support for custom split at various levels. That is, you don't necessarily have to cut the matrix in half at each level and you can split your matrix at a user defined index. Finally and most importantly, you can create an HODLR matrix using a variety of methods. You can supply a dense matrix, a sparse matrix or a kernel function.

Unfortunately as the package is fairly new, I haven't had the time to write a comprehensive documentation and this is something that I'm working on. Having said that, I'm using this package as a core part of my fast sparse solver (which will become available shortly) and the package is also being used by a group in NYU Courant for their Stokes flow solver. Hence, I'm constantly monitoring the package for possible issues. I'm also happy to add any new futures that your specific application might require.

Amir

[stnava]

this sounds promising. i will try the adaptation soon and let you know if i encounter any issues. thanks, ahead of time, for your help and insight!

brian

On Fri, Nov 21, 2014 at 11:50 PM, AmirHossein Aminfar < notifications@github.com> wrote:

This package is written by Sivaram Ambikasaran a former colleague of mine. There are several advantages to my package. First, my package supports a variety of low-rank approximation methods (SVD, partial pivoting LU, full pivoting LU, partial pivoting ACA and Pseudo skeleton). One other advantage of my package is the support for custom split at various levels. That is, you don't necessarily have to cut the matrix in half at each level and you can split your matrix at a user defined index. Finally and most importantly, you can create an HODLR matrix using a variety of methods. You can supply a dense matrix, a sparse matrix or a kernel function.

Unfortunately as the package is fairly new, I haven't had the time to write a comprehensive documentation and this is something that I'm working on. Having said that, I'm using this package as a core part of my fast sparse solver (which will become available shortly) and the package is also being used by a group in NYU Courant for their Stokes flow solver. Hence, I'm constantly monitoring the package for possible issues. I'm also happy to add any new futures that your specific application might require.

Amir

— Reply to this email directly or view it on GitHub https://github.com/amiraa127/Dense_HODLR/issues/1#issuecomment-64069225.

[sivaramambikasaran]

@stnava As Amir pointed out this has few more features than mine. The only things missing are determinants and symmetric factorization. If Amir could add these routines as well, it should be great.

[stnava]

Closing this issue - my questions were addressed above. Opening new issue for ongoing R port issues.