|Number of watchers on Github||32|
|Number of open issues||1|
|Open pull requests||0+|
|Closed pull requests||0+|
|Last commit||almost 7 years ago|
|Repo Created||almost 7 years ago|
|Repo Last Updated||over 1 year ago|
|Organization / Author||karpathy|
|Do you use MatlabWrapper? Leave a review!|
|View MatlabWrapper activity|
|View on github|
|Fresh, new opensource launches 🚀🚀🚀|
Trendy new open source projects in your inbox! View examples
Andrej Karpathy 27 August 2012
This C++ library is a convenience wrapper around Matlab Engine, especially if you use Eigen matrix library for working with matrices. It should be quite easy to adapt it to other library types as well.
Consider the simplest example: You have an
Eigen::MatrixXf X; lying around
and you'd love to run k-means on its rows. But naturally, you don't feel like
re-implementing k-means for Eigen matrices from scratch! So you write:
MatlabWrapper mw; mw.sendMat("X", X); // transfer it over to the engine under name "X" mw.exec("[IDX, C] = kmeans(X, 10);"); Eigen::MatrixXf mat; mw.receiveMat("C", mat); // done!
That felt good. Now supose you've computing something complicated and would really like to explore it a bit interactively-- do a few plots, run some statistics, etc. No problem:
MatlabWrapper mw; mw.sendMat("X", X); mw.interact();
This will start an interactive session with matlab right in your shell! You get to type whatever you wish:
>> plot(X(:,1), X(:,2)); // interesting! >> close; >> hist(X); >> close;
Entering qq will resume the c++ program right where it left off. This is quite great for debugging as well.
Eigen library is a dependency, and so is Matlab, naturally. If you don't have eigen, install it with
$ sudo apt-get install libeigen3-dev
Also make sure you have tcsh, which is require for matlab engine to work
sudo apt-get install tcsh
At this point you may want to get the official MATLAB demo to run (see below)
then to install this library.
$ git clone firstname.lastname@example.org:karpathy/MatlabWrapper.git $ cd MatlabWrapper $ mkdir build $ cd build $ cmake .. $ make $ ./test1
where last line runs a small demo. Naturally, the above will probably not run right away because these things never do. Have a look inside CMakeLists.txt and modify the include/library paths according to your system preferences. I tried to document it as best I could.
If you haven't used Matlab Engine yet, try to get their demo running first. Here is the main documentation: http://www.mathworks.com/help/techdoc/matlab_external/f29148.html
In particular of interest is this link on getting their engdemo.c running: http://www.mathworks.com/help/techdoc/matlab_external/bsq776w-1.html#bsq78dr-9
If you can get that running, you're halfway there! Problems I ran across:
sudo apt-get install tcsh