In the MKL provider output directory, there should also be libiomp5md.dll (the Intel openmp rumtime). That also needs to be copied to the same directory as your executable. I find it useful to add both the provider and openmp runtime to the project solution and then set copy on build to always. This way the both DLLs are copied to your to output directory.
↧
New Post: Problem using Native Linear Algebra Provider
↧
New Post: Problem using Native Linear Algebra Provider
Thanks very much for the quick replies! Sorry I should have mentioned that I'm also copying the libiomp5md.dll file along with the MathNET.Numerics.MKL.dll file, and I'm running 2.3 right now, developing in VS2010. Also, apparently our system/network config makes NuGet tricky, so unfortunately that's not an option.
Cuda, you mentioned adding the provider and the openmp runtime to my project solution - do you mean my solution (i.e., the software I'm developing, not any of the Math.NET Numerics solutions)? I tried adding the MKL dll as a Reference just to see what would happen, but as expected it didn't work. Is there another way to point Visual Studio to these files (aside from just dropping them in the folder with MathNet.Numerics.dll manually)? Maybe that's what I'm missing?
Thanks,
Brian
Cuda, you mentioned adding the provider and the openmp runtime to my project solution - do you mean my solution (i.e., the software I'm developing, not any of the Math.NET Numerics solutions)? I tried adding the MKL dll as a Reference just to see what would happen, but as expected it didn't work. Is there another way to point Visual Studio to these files (aside from just dropping them in the folder with MathNet.Numerics.dll manually)? Maybe that's what I'm missing?
Thanks,
Brian
↧
↧
New Post: Problem using Native Linear Algebra Provider
In your project, click add existing item and then select the two DLLs. Then select those two items in
the solution explorer and click properties. From there set copy to always.
But I wonder if the problem is the visual c++ runtime. Could you try installing the VC++ 2012 runtime?
http://www.microsoft.com/en-us/download/details.aspx?id=30679
the solution explorer and click properties. From there set copy to always.
But I wonder if the problem is the visual c++ runtime. Could you try installing the VC++ 2012 runtime?
http://www.microsoft.com/en-us/download/details.aspx?id=30679
↧
New Post: Problem using Native Linear Algebra Provider
But you are building it yourself so the runtime probably isn't the problem.
↧
New Post: Problem using Native Linear Algebra Provider
I just checked the MKL provider DLL with dependency walker (http://www.dependencywalker.com/) and it only shows libiomp5md.dll and kernel32.dll as dependencies (as expected). If both your executable and the two DLLs are all in the same directory, I'm stumped.
The only thing I can think of is adding the directory that contains the two DLLs to your PATH and then launch your exe via the command line. If that works, then we know the problem is loading the DLLs locally.
Another option, is try it with mono. Mono has a great log facility.
From the mono command line prompt:
set MONO_LOG_LEVEL=debug
set MONO_LOG_MASK=dll
mono your.exe
Then mono will log all DLL errors and calls.
The only thing I can think of is adding the directory that contains the two DLLs to your PATH and then launch your exe via the command line. If that works, then we know the problem is loading the DLLs locally.
Another option, is try it with mono. Mono has a great log facility.
From the mono command line prompt:
set MONO_LOG_LEVEL=debug
set MONO_LOG_MASK=dll
mono your.exe
Then mono will log all DLL errors and calls.
↧
↧
New Post: Problem using Native Linear Algebra Provider
That did the trick - for some reason VS didn't like the DLLs unless it put them there itself. Adding them manually to the project and selecting copy always worked on my test project and looks like it's working on the main project as well. Thanks very much for the help and the very quick responses!
↧
Created Issue: Control.ParallelizeOperation BUG [5723]
There is a potential bug in the method:
`MathNet.Numerics.Control.ParallelizeOperation(int elements)`
This method should return true if the operation should be parallelized; and false otherwise.
Instead, this method returns the exact opposite of this (see the source code).
I would prefer, if the property `DisableParallelization` was set to false (by default) in the future revisions.
`MathNet.Numerics.Control.ParallelizeOperation(int elements)`
This method should return true if the operation should be parallelized; and false otherwise.
Instead, this method returns the exact opposite of this (see the source code).
I would prefer, if the property `DisableParallelization` was set to false (by default) in the future revisions.
↧
New Post: Inverse of StudentT distribution
Hi,
I'm just wondering if there are any plans to add that function? I'm currently using the code that I ported to C# from the C port of the cephes library, so using Math.Net would be more convenient. I can post that code if there is interest.
Thanks,
Andrew
I'm just wondering if there are any plans to add that function? I'm currently using the code that I ported to C# from the C port of the cephes library, so using Math.Net would be more convenient. I can post that code if there is interest.
Thanks,
Andrew
↧
Commented Issue: Control.ParallelizeOperation BUG [5723]
There is a potential bug in the method:
`MathNet.Numerics.Control.ParallelizeOperation(int elements)`
This method should return true if the operation should be parallelized; and false otherwise.
Instead, this method returns the exact opposite of this (see the source code).
I would prefer, if the property `DisableParallelization` was set to false (by default) in the future revisions.
Comments: Added to github [#89](https://github.com/mathnet/mathnet-numerics/issues/89)
`MathNet.Numerics.Control.ParallelizeOperation(int elements)`
This method should return true if the operation should be parallelized; and false otherwise.
Instead, this method returns the exact opposite of this (see the source code).
I would prefer, if the property `DisableParallelization` was set to false (by default) in the future revisions.
Comments: Added to github [#89](https://github.com/mathnet/mathnet-numerics/issues/89)
↧
↧
Closed Issue: Control.ParallelizeOperation BUG [5723]
There is a potential bug in the method:
`MathNet.Numerics.Control.ParallelizeOperation(int elements)`
This method should return true if the operation should be parallelized; and false otherwise.
Instead, this method returns the exact opposite of this (see the source code).
I would prefer, if the property `DisableParallelization` was set to false (by default) in the future revisions.
Comments: Fixed, thanks for reporting.
`MathNet.Numerics.Control.ParallelizeOperation(int elements)`
This method should return true if the operation should be parallelized; and false otherwise.
Instead, this method returns the exact opposite of this (see the source code).
I would prefer, if the property `DisableParallelization` was set to false (by default) in the future revisions.
Comments: Fixed, thanks for reporting.
↧
New Post: Maxima and Saddle Points
How would I use Math.NET numerics to find local and global maxima and/or saddle points?
↧
Created Issue: (minor) Documentation fix [5724]
Hi,
On http://numerics.mathdotnet.com/api/MathNet.Numerics.Integration/Integrate.htm , the method descriptions (both) say :
Approximation of the definite interval of an analytic smooth function on a closed interval.
I suggest the description be changed to :
Approximation of the definite integral of an analytic smooth function on a closed interval.
... just a suggestion,
Chuck
On http://numerics.mathdotnet.com/api/MathNet.Numerics.Integration/Integrate.htm , the method descriptions (both) say :
Approximation of the definite interval of an analytic smooth function on a closed interval.
I suggest the description be changed to :
Approximation of the definite integral of an analytic smooth function on a closed interval.
... just a suggestion,
Chuck
↧
New Post: Is there a DataVisualiser for matrices in the project?
Here's my debugger visualizer for sparse matrices: MatrixDebuggerVisualizer.zip
↧
↧
New Post: Is there a DataVisualiser for matrices in the project?
↧
Source code checked in, #12c9cd486d91
LA: Generic Matrix partial Matrix.BCL class
↧
Source code checked in, #6a085491ef53
LA: Vector.Negate should implement common pattern
↧
Source code checked in, #2df2f0cb839e
LA: Sparse complex DoConjugate should not call base.Conjugate
↧
↧
Source code checked in, #2cb8a2a0a78f
LA: Matrix group operators together, some eol whitespaces fixes
↧
Source code checked in, #5d4c296a74c2
LA: Result should be called 'result' not 'target'
↧
Source code checked in, #b215abc89002
LA: Obsolete Vector.Plus, drop some redundant operator overloads
↧