CSparse.NEThttps://csparse.codeplex.com/project/feeds/rssA Concise Sparse Matrix Package for .NETNew Post: Multy Core supporthttps://csparse.codeplex.com/discussions/647166<div style="line-height: normal;">just asked it because i was interested in your CSParse.NET project.
<br />
All today computers have multicore processor, so i think multicore support makes your library scale-able and much better.<br />
</div>epsi1onWed, 18 Nov 2015 13:19:32 GMTNew Post: Multy Core support 20151118011932PNew Post: Multy Core supporthttps://csparse.codeplex.com/discussions/647166<div style="line-height: normal;">Sorry for the delay.
<br />
<br />
There are no plans to add multi core support to CSparse. How large are the matrices you are factorizing (size, nonzeros)? If it's from Brief Finite Element .NET, could you show some example code or upload an assembled matrix?<br />
</div>wo80Tue, 17 Nov 2015 17:09:16 GMTNew Post: Multy Core support 20151117050916PNew Post: Multy Core supporthttps://csparse.codeplex.com/discussions/647166<div style="line-height: normal;">Hello Christian,<br />
Do you still have no plan for making your sparse library to support multi core (parallel) calculations at same time? i believe it will have high impact on performance.<br />
<br />
Regards<br />
</div>epsi1onTue, 10 Nov 2015 08:37:12 GMTNew Post: Multy Core support 20151110083712AUpdated Wiki: Documentationhttps://csparse.codeplex.com/documentation?version=20<div class="wikidoc"><h1>See <a href="https://github.com/wo80/CSparse.NET/wiki">GitHub Wiki</a></h1></div><div class="ClearBoth"></div>wo80Sun, 20 Sep 2015 13:41:33 GMTUpdated Wiki: Documentation 20150920014133PUpdated Wiki: Documentationhttps://csparse.codeplex.com/documentation?version=19<div class="wikidoc"><h1>See <a href="https://github.com/wo80/CSparse.NET/wiki">GithHub Wiki</a></h1></div><div class="ClearBoth"></div>wo80Sun, 20 Sep 2015 13:41:18 GMTUpdated Wiki: Documentation 20150920014118PUpdated Wiki: Homehttps://csparse.codeplex.com/wikipage?version=8<div class="wikidoc"><h1>CSparse.NET has moved to <a href="https://github.com/wo80/CSparse.NET">GitHub</a></h1></div><div class="ClearBoth"></div>wo80Sun, 20 Sep 2015 13:39:01 GMTUpdated Wiki: Home 20150920013901PNew Post: Eigen value and eigen vectors for spd matrixhttps://csparse.codeplex.com/discussions/640021<div style="line-height: normal;">CSparse.NET won't help you compute eigenvalues and for large problems you'd usually resort to iterative methods.
<br />
<br />
Since 100x100 is rather small I suggest you do a dense Eigendecomposition. You can use the code from <a href="https://github.com/mathnet/mathnet-numerics/blob/master/src/Numerics/LinearAlgebra/Double/Factorization/DenseEvd.cs" rel="nofollow">MathNet</a>.
<br />
<br />
For the generalized Eigenvalue problem, you'll have to modify the system like this: <a href="http://www.cmth.ph.ic.ac.uk/people/a.mackinnon/Lectures/compphys/node72.html" rel="nofollow">http://www.cmth.ph.ic.ac.uk/people/a.mackinnon/Lectures/compphys/node72.html</a><br />
</div>wo80Wed, 17 Jun 2015 09:31:09 GMTNew Post: Eigen value and eigen vectors for spd matrix 20150617093109ANew Post: Eigen value and eigen vectors for spd matrixhttps://csparse.codeplex.com/discussions/640021<div style="line-height: normal;">Hello Christian,
<br />
Is it possible to find all eigenvalues and all eigenvectors for a general eigenvalue problem (engaging two matrices)? both matrixes A and B are SPD and they always have very small dimension like at most 100x100. if there be a solution for finding eigenvalues and eigenvectors for pair of dense matrices that also is useful for me and will be grate...
<br />
<br />
Thanks<br />
</div>epsi1onWed, 17 Jun 2015 05:56:54 GMTNew Post: Eigen value and eigen vectors for spd matrix 20150617055654ANew Post: Main considerations of this library compared to other(s)https://csparse.codeplex.com/discussions/638199<div style="line-height: normal;">Okay, it is interesting that the largest contributor of mathnet recognizes what you just described. Maybe it could be useful for that project to integrate the sparse solvers you published here. Anyhow thanks for sharing . <br />
</div>a5rTue, 19 May 2015 15:38:16 GMTNew Post: Main considerations of this library compared to other(s) 20150519033816PNew Post: Main considerations of this library compared to other(s)https://csparse.codeplex.com/discussions/638199<div style="line-height: normal;">Support for solving large systems is very limited in MathNet. You have a couple of iterative solvers and that's it (<a href="https://github.com/mathnet/mathnet-numerics/issues/227" rel="nofollow">https://github.com/mathnet/mathnet-numerics/issues/227</a>). You might know that iterative solvers need good preconditioners to be efficient (or even to converge at all). The only perconditioner in MathNet that is optimized for sparse matrices is MILU(0), which I contributed. The library also lacks any optimization for sparse, symmetric matrices.<br />
<br />
After all, I did this project mainly to learn about sparse direct solvers. Tim Davis' book (Direct Methods for Sparse Linear Systems) is a really good starting point and since the C# version performs pretty well, I thought: why not publish it.<br />
<br />
I don't know about other implementations for the .NET platform, so people might find it useful.<br />
</div>wo80Tue, 19 May 2015 15:01:45 GMTNew Post: Main considerations of this library compared to other(s) 20150519030145PNew Post: Main considerations of this library compared to other(s)https://csparse.codeplex.com/discussions/638199<div style="line-height: normal;">Hi,
<br />
I've been wondering how this library relates to other library code (e.g. MathNet). There is also code in the MathNET library for sparse matrices. Given that MathNet is a larger library, what is the particular purpose of this library? Is the main idea to make it faster or simpler than the larger libs? Because i know your library Triangle.Net (which is really nice) i was wondering about this one as well. So my main question is what are your considerations for this library? Thanks in advance.<br />
</div>a5rTue, 19 May 2015 11:04:30 GMTNew Post: Main considerations of this library compared to other(s) 20150519110430ANew Post: increase performance in double SparseQR https://csparse.codeplex.com/discussions/561748<div style="line-height: normal;">Do you have any plan for making the library parallel? (specificly the cholesky decomposition)<br />
</div>epsi1onMon, 29 Sep 2014 16:43:27 GMTNew Post: increase performance in double SparseQR 20140929044327PNew Post: increase performance in double SparseQR https://csparse.codeplex.com/discussions/561748<div style="line-height: normal;">I played around with metis' reordering routines but haven't seen any significant improvement over AMD.
<br />
<br />
There's a new constructor in the Cholesky class which takes a given permutation as input, so you can now use custom permutations (metis nd, colamd, whatever...)<br />
</div>wo80Mon, 29 Sep 2014 16:41:40 GMTNew Post: increase performance in double SparseQR 20140929044140PSource code checked in, #75514http://csparse.codeplex.com/SourceControl/changeset/75514Add Cholesky constructor (custom permutation)wo80Mon, 29 Sep 2014 16:32:34 GMTSource code checked in, #75514 20140929043234PSource code checked in, #75513http://csparse.codeplex.com/SourceControl/changeset/75513Sync changes with complex QRwo80Mon, 29 Sep 2014 16:25:00 GMTSource code checked in, #75513 20140929042500PSource code checked in, #75294http://csparse.codeplex.com/SourceControl/changeset/75294Minor optimizationsepsi1onTue, 26 Aug 2014 09:33:09 GMTSource code checked in, #75294 20140826093309ANew Post: increase performance in double SparseQR https://csparse.codeplex.com/discussions/561748<div style="line-height: normal;">Wow, great result.
<br />
It was good if other ordering methods (like metis nested dissection) included in package. was noted is some references (like <a href="http://milamin.sourceforge.net/technical-notes/reordering" rel="nofollow">this</a>) that it can have significant effects.
<br />
<br />
Thanks for information anyways...<br />
</div>epsi1onSat, 23 Aug 2014 06:02:51 GMTNew Post: increase performance in double SparseQR 20140823060251ANew Post: increase performance in double SparseQR https://csparse.codeplex.com/discussions/561748<div style="line-height: normal;">I used the defaults, which is <em>Maximize Speed (/O2)</em> for release mode.<br />
</div>wo80Fri, 22 Aug 2014 18:56:44 GMTNew Post: increase performance in double SparseQR 20140822065644PNew Post: increase performance in double SparseQR https://csparse.codeplex.com/discussions/561748<div style="line-height: normal;">Thanks.
<br />
In your benchmark was native files compiler optimized? (I mean is it best performance of the native files)<br />
</div>epsi1onFri, 22 Aug 2014 18:10:18 GMTNew Post: increase performance in double SparseQR 20140822061018PNew Post: increase performance in double SparseQR https://csparse.codeplex.com/discussions/561748<div style="line-height: normal;">Yes, I did some benchmarks before. The managed code performs pretty well:<br />
<pre><code>Matrix crystm03: 24696 x 24696, 583770 non-zeros
Native Cholesky: 3464ms, residual = 3,976e-26, result = 1
Managed Cholesky: 4584ms, residual = 3,976e-26
Native LU: 8464ms, residual = 4,377e-26, result = 1
Managed LU: 10891ms, residual = 4,377e-26
Native QR: 13310ms, residual = 1,542e-25, result = 1
Managed QR: 15907ms, residual = 1,542e-25</code></pre>
In case you are interested, here's the Visual Studio project: <a href="http://wo80.bplaced.net/files/CXSparse.Benchmark.rar" rel="nofollow">CXSparse.Benchmark.rar</a><br />
</div>wo80Fri, 22 Aug 2014 17:41:20 GMTNew Post: increase performance in double SparseQR 20140822054120P