What is MPI?

MPI,[mpi-using][mpi-ref] theMessage Passing Interface, is astandardized and portable message-passing system designed to functionon a wide variety of parallel computers. The standard defines thesyntax and semantics of library routines and allows users to writeportable programs in the main scientific programming languages(Fortran, C, or C++).

Since its release, the MPI specification[mpi-std1][mpi-std2] hasbecome the leading standard for message-passing libraries for parallelcomputers. Implementations are available from vendors ofhigh-performance computers and from well known open source projectslikeMPICH[mpi-mpich] andOpen MPI[mpi-openmpi].

What is Python?

Python is a modern, easy to learn, powerful programming language. Ithas efficient high-level data structures and a simple but effectiveapproach to object-oriented programming with dynamic typing anddynamic binding. It supports modules and packages, which encouragesprogram modularity and code reuse. Python’s elegant syntax, togetherwith its interpreted nature, make it an ideal language for scriptingand rapid application development in many areas on most platforms.

The Python interpreter and the extensive standard library areavailable in source or binary form without charge for all majorplatforms, and can be freely distributed. It is easily extended withnew functions and data types implemented in C or C++. Python is alsosuitable as an extension language for customizable applications.

Python is an ideal candidate for writing the higher-level parts oflarge-scale scientific applications[Hinsen97] and drivingsimulations in parallel architectures[Beazley97] like clusters ofPC’s or SMP’s. Python codes are quickly developed, easily maintained,and can achieve a high degree of integration with other librarieswritten in compiled languages.

Related Projects

As this work started and evolved, some ideas were borrowed from wellknown MPI and Python related open source projects from the Internet.

• OOMPI

  • It has no relation with Python, but is an excellent objectoriented approach to MPI.

  • It is a C++ class library specification layered on top of the Cbindings that encapsulates MPI into a functional class hierarchy.

  • It provides a flexible and intuitive interface by adding someabstractions, likePorts andMessages, which enrich andsimplify the syntax.

• Pypar

  • Its interface is rather minimal. There is no support forcommunicators or process topologies.

  • It does not require the Python interpreter to be modified orrecompiled, but does not permit interactive parallel runs.

  • General (picklable) Python objects of any type can becommunicated. There is good support for numeric arrays,practically full MPI bandwidth can be achieved.

• pyMPI

  • It rebuilds the Python interpreter providing a built-in modulefor message passing. It does permit interactive parallel runs,which are useful for learning and debugging.

  • It provides an interface suitable for basic parallel programming.There is not full support for defining new communicators or processtopologies.

  • General (picklable) Python objects can be messaged betweenprocessors. There is native support for numeric arrays.

• Scientific Python

  • It provides a collection of Python modules that areuseful for scientific computing.

  • There is an interface to MPI and BSP (Bulk Synchronous Parallelprogramming).

  • The interface is simple but incomplete and does not resemblethe MPI specification. There is support for numeric arrays.

Additionally, we would like to mention some available tools forscientific computing and software development with Python.

• NumPy is a package that provides array manipulation andcomputational capabilities similar to those found in IDL, MATLAB, orOctave. Using NumPy, it is possible to write many efficientnumerical data processing applications directly in Python withoutusing any C, C++ or Fortran code.

• SciPy is an open source library of scientific tools for Python,gathering a variety of high level science and engineering modulestogether as a single package. It includes modules for graphics andplotting, optimization, integration, special functions, signal andimage processing, genetic algorithms, ODE solvers, and others.

• Cython is a language that makes writing C extensions for thePython language as easy as Python itself. The Cython language isvery close to the Python language, but Cython additionally supportscalling C functions and declaring C types on variables and classattributes. This allows the compiler to generate very efficient Ccode from Cython code. This makes Cython the ideal language forwrapping for external C libraries, and for fast C modules that speedup the execution of Python code.

• SWIG is a software development tool that connects programswritten in C and C++ with a variety of high-level programminglanguages like Perl, Tcl/Tk, Ruby and Python. Issuing header filesto SWIG is the simplest approach to interfacing C/C++ libraries froma Python module.