Moose

File Release Notes and Changelog

Release Name: Moose Beta 1.1.0

Notes:
These are the release notes for the Beta 1.1.0 release of MOOSE, the
Multiscale Object-Oriented Simulation environment.

Introduction.
This release has three major functional changes from the previous Beta 1.0.0:
- The development of PyMOOSE as a full-fledged user environment.
- The implementation of a much more functional single-neuron modeling engine
- The implementation of array objects and messages.

The release also has a major overhaul to the release management, which has been
coordinated by a new member of the MOOSE team, Ms. Siji George. The current
release has binaries for Linux, Solaris, Mac and Windows. The Linux release
includes a .deb package for simplifying the installation. There are several
permutations for Python support and for 32/64 bits.
We expect that the source compilation should also go more smoothly for those
who wish to compile it themselves.

=============================================
Release notes for PyMOOSE (MOOSE beta 1.1) (Subhasis Ray)
=============================================

1. INTRODUCTION

PyMOOSE is MOOSE for Python. It is a module that you can import in a
Python script.  Thus use the MOOSE classes as Python classes to create
models and run simulations.

2. INSTALLATION

PyMOOSE is compiled as a shared library. We have compiled it on
various platforms and you can download one suited for you from the
sourceforge site. But in case there is no precompiled binary suited
for your environment, you can compile it for yourself.

2.1. LINUX

To compile PyMOOSE you need the Python development library
installed. You can install it using apt ( on Debian like systems ):

apt-get install python-dev

Using yum ( on Fedora )

yum install python-dev

If you want to use GSL, then install the GSL development
library also.

apt-get install libgsl0-dev

or

yum install gsl-devel

In the Makefiles we have used standard paths for these libraries. If
your system has them installed somewhere else, edit the Makefile and
do these replacements:

 /usr/include/python2.5 

 with 

 path to your python development library's include directory ( which 
contains the file Python.h )

 -lpython2.5 

 with

 -lpythonX.Y where X.Y is the version of your Python installation.

Now go to the moose directory and build PyMOOSE using:

make pymoose BUILD=release USE_GSL=true

If you want to build without GSL support:

make pymoose BUILD=release

For a debug build use BUILD=debug on the make command line.

2.2. Mac OS X

Replace
/Library/Frameworks/Python.framework/Versions/2.5/include/python2.5 in
the pymoose/Makefile with the path to the directory containing
Python.h and
/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/config/
with the path to the directory containing libpythonX.Y.a where X.Y is
your Python version number.

Now go to the moose directory and build PyMOOSE using:

make pymoose BUILD=release USE_GSL=true

If you want to build without GSL support:

make pymoose BUILD=release

If you want a debug build, use BUILD=debug instead.

2.3. MICROSOFT WINDOWS

Here we recommend that you use the ready-made binaries. To use GSL you
can install the Windows version of GSL Library
(http://gnuwin32.sourceforge.net/packages/gsl.htm). 

If you want to build PyMOOSE for yourself, you need VC++ 8 or 9 to
compile PyMOOSE on Windows. The freely available express editions are
just fine.  Open the pymoose_vcpp2005/moose/moose.vcproj in Visual C++
IDE. Select the project properties and update the preprocessor
directives (add or remove the directive USE_GSL), update C++ include
paths and Linker inputs and library directories.

2.4. SOLARIS

The procedure to compile PyMOOSE on solaris is very similar to that 
outlined for Linux. Install Python development libraries and GSL using
your favourite package manager, or from source.

It is perhaps easiest to use GNU tools to compile PyMOOSE. You will need
to set the appropriate paths for GNU Make, g++, flex and bison with the
following bash commands:
	export PATH=/usr/sfw/bin:/usr/ccs/bin:/usr/local/bin:$PATH
	alias make='/usr/sfw/bin/gmake'

If you are going to use GSL, then set the linker path:
	export LD_LIBRARY_PATH=/usr/local/lib

Finally, update the Makefile for your system, and build pymoose as
described above for Linux. 

After a successful build, you can copy pymoose/_moose.so and
pymoose/moose.py to a directory of your choice. You have to append
this directory to your PYTHONPATH variable in order to import the
module. Assuming you kept files in ~/moose directory, in bash shell
you can update PYTHONPATH by executing the command

export PYTHONPATH=$PYTHONPATH:~/moose

If you want this to be permanent, append the same to your ~/.bashrc file.

On Windows, instead of _moose.so you will have _moose.pyd. You can set
the PYTHONPATH variable by right clicking My Computer and selecting
Properties from the pop-up menu. Now go to Advanced tab and click
Environment Variables. If there is no PYTHONPATH variable defined,
create a new one. Otherwise edit the existing one and append the path
to the directory containing _moose.pyd and moose.py.


3. USAGE

To test your PyMOOSE installation, run python and in the interpreter prompt , try

import moose

If it works ( no error messages ), you can now use the moose functions
and classes in Python scriupts. There are some example scripts in
examples directory. In particular, there is a new avatar of the
GENESIS squid demo using PyQt. To run the GUI, you have to install
PyQt4(http://www.riverbankcomputing.co.uk/software/pyqt/download) and
PyQwt5(http://pyqwt.sourceforge.net/download.html). And beware that
for PyQt to work on windows, the path to PyQt4/bin must be the first
entry in your PYTHONPATH environment variable.

You can run the squid demo without GUI as well, just cd to the
pymoose/examples/squid and run:

python squid.py

That will create a bunch of .plot files containing the time series for
various parameters.


You can use most MOOSE classes like regular Python classes. However you have to access global functions using the special PyMooseContext object. You can retrieve it by:

context = moose.PyMooseBase.getContext()

And call the global MOOSE functions as methods on this object. 
Moreover, you can run GENESIS commands from Python using 

context.runG("GENESIS command string")

and load whole GENESIS scripts using

context.loadG("genesis script file")

Take a look into the example scripts and that should get you going.

=============================================
Release notes for the single-neuron modeling engine: Hsolve (Niraj Dudani)
=============================================

- The HSolver can now handle cell models with intracellular calcium pools and
calcium-dependent channels. It uses a simple single-exponential description for
the calcium dynamics. Calcium diffusion, and updates to Nernst potential are not
dealt with yet.

- A major shortcoming was found in the solution for branching compartmental
models. Due to this, the solution was incorrect for most branching cell models.
Much of the code for passive models has been completely reworked, and tested
extensively using low-level unit tests and full-fledged models. Only models
with symmetric compartments are presently supported.

- HSolve is not compatible with arrays of elements (created using createmap).

- The entire solver has been revamped. It is now cleaner, more stable, and
faster. Early benchmarks show its speed to be very close to that of hsolve in
GENESIS.

- Fields and messages belonging to a cell model are now completely served for
by its solver. This management was incomplete earlier.




Changes: