Plham: Platform for large-scale and high-frequency artificial market

Plham is a platform for artificial market simulations, written in X10 language:

• for large-scale parallel computation
• as well as standalone sequential computation

Plham is shipped with reusable examples based on recent artirifial market studies on

• Fundamentalist-chartist agents
• Single asset simulations
• Price limit regulation
• Trading halt regulation
• Fat finger error
• Flash Crash shock transfer

The documentation is currently only written in Japanese (the English text will appear soon):

• Documentation (Japanese)

The API is written in English:

• API (English)

Checkout

This repository uses some submodules. You can recursively clone the submodules as follows:

git clone --recursive git@github.com:plham/plham.git

Single Thread Version

You can compile and run the sample programs at the top directory of Plham as follows:

compile:
% x10c -sourcepath .:cassiaX10lib samples/CI2002/CI2002Main.x10
execution:
% x10 samples.CI2002.CI2002Main samples/CI2002/config.json

Parallel Version

For parallel versions, please use the native version of X10 with MPI runtimes.

Sample Program: samples/Parallel/ParallelDistMain.x10

compile:
% x10c++ -x10rt mpi -sourcepath .:cassiaX10lib -O -o paraMain samples/Parallel/ParallelDistMain.x10
parallel execution:
% mpirun -np numHosts -host host0,host1,... ./paraMain samples/Parallel/config-a002.json

Execution Environment:

• The parallel version needs some customization of X10.
  • Please take the customized version of X10 from here and build it.
• We recommend to use thread-safe MPI.
• You can build the X10 with MPI runtimes with executing the following command at the x10.dist directory.

ant  -DX10RT_MPI=true -Doptimize=true -DTHREAD_LOCAL=true -DNO_CHECKS=true dist

• Please use native version of X10 and set an environment variable as X10RT_MPI_FORCE_COLLECTIVES=true. Please set X10_NTHREADS to specify the number of worker threads on each computing node.

License

Currently under Eclipse Public License 1.0 (read here).

History

• v0.3: April 2019
• v0.1.1: May 2016
• v0.1: April 2016

Contributors

• Takuma Torii
• Kiyoshi Izumi
• Tomio Kamada
• Hiroto Yonenoh
• Daisuke Fujishima
• Izuru Matsuura
• Masanori Hirano
• Tosiyuki Takahashi