mitkof6 / brahms

BRAHMS is the simulation backend used by SpineCreator

BRAHMS

BRAHMS is a simulation execution engine which was written by Ben Michinson. It's often used alongside SpineML/SpineML_2_BRAHMS to provide a simulation back-end for SpineML/SpineCreator. The original author is not currently developing BRAHMS, although it is in current use within the Adaptive Behaviour Research Group at the Department of Psychology of the University of Sheffield. As we expect to continue its use, I have created this fork of the software on github.

This fork of the Brahms simulator starts from brahms version 0.7.3. The key improvement is a standard build system to replace the custom makefiles in the original, because at present we are using "set in aspic" binary builds of BRAHMS and we would like to be able to easily modify and add features as our requirements of the software change. All of the old, custom-written makefiles are present with an "__" prefix. These will be removed at some point, but I'm keeping them for reference at present.

You can read all about the original at:

http://brahms.sourceforge.net/docs/What%20is%20BRAHMS.html

The original code is also available at that site.

What changed in this version?

The major change between this version and BRAHMS 0.7.3 is that the requirement to set SYSTEMML_INSTALL_PATH has been removed. Instead, when the software is compiled, a directory for the "installed SystemML Namespace" is set - that might be /usr/local/var/SystemML/Namespace. That becomes the default BRAHMS Namespace. It's in a var directory as you may decide to install your own components in there.

What may change in future?

I may re-instate SYSTEMML_INSTALL_PATH such that you can have a second installed Namespace, perhaps at $HOME/SystemML/Namespace which is referred to with SYSTEMML_INSTALL_PATH.

I may implement a scheme for saving the delayed values stored in connections which have some specified lag.

Building BRAHMS

This project uses cmake for a cross-platform friendly build process. cmake is a good choice of build system, because it is familiar to many developers.

You'll need to obtain the correct dependencies to build BRAHMS. On Linux that means a compiler, libXv, libXaw and WX windows. For example, on (current 2015) Ubuntu or Debian systems you'll want to do:

sudo apt-get install build-essential libxaw7-dev libxv-dev

These libraries are used to draw the progress box which shows while BRAHMS is running.

For Windows, you'll reportedly need GnuWin32, but this build has not yet been tested.

For optional features (MPI channel, Python and MATLAB bindings and one WXWidgets-dependent component) you'll need mpich2, python development libraries, WXwidgets and a MATLAB installation. The first three can be obtained via apt-get:

sudo apt-get install mpich2 python-dev libwxgtk2.8-dev

The recommended method for building is to create a separate build folder as follows:

mkdir build
cd build
cmake -DCMAKE_INSTALL_PREFIX=/usr/local .. # or cmake-gui ..
make -j4 # or however many processor cores you have
sudo make install

Options for cmake

Here are the options you can set. Shown here are the defaults (at time of writing):

Adding/removing bindings:

cmake -DCOMPILE_MATLAB_BINDING=OFF -DCOMPILE_PYTHON_BINDING=ON ..

MPICH2:

cmake -DCOMPILE_WITH_MPICH2=ON ..

Compiling a component which requires WX windows:

cmake -DCOMPILE_WX_COMPONENT=OFF ..

Compiling with a specific installation prefix:

cmake -DCMAKE_INSTALL_PREFIX=/usr/local ..

OLD Build Notes for reference

To build each part, enter each project directory and run "make". On Windows, you will have to have GnuWin32 installed. To build all parts, run "make" in this folder, and review the options.

The only pre-requisite (other than compiler, libraries, and all required INCLUDE and LIB paths set) is that you should have set the env var "SYSTEMML_INSTALL_PATH" to the folder you want to install to, or to the folder where your SystemML/BRAHMS installation currently is. If you installed BRAHMS using an installer, this may have been done for you. If you are building from scratch without a current install, this probably needs doing manually.

As well as the base development libraries on Linux the dev versions of libXv and libXaw are required.

You will need subversion installed, because the makefiles use it to get the revision number, even if you're building from an exported branch. We'll fix this one day.

On Linux, you will need the env var SYSTEMML_MATLAB_PATH set to your Matlab root directory. On our system, it's "/usr/local/encap/matlab-R2007B". On Windows, you will need to make sure the Matlab include and bin folders are in your INCLUDE and LIB env vars, respectively.

Compilers

We use G++ 4.2 on linux, or cl v14 on windows. Other compilers should work fine, but your binaries will be incompatible with ours. That won't usually matter, since you'll build all the binaries yourself.

-- Seb James, Feb 2015.