yale-phys-382l / ising Goto Github PK

# Ising Lattice Simulation Implementation

The main program here is ising.py which takes a variety of inputs and generates outputs of 
the 2-D Ising model. The Ising lattice simulation is provided by two codes, one in Python
and the other in C++. The C++ code is *much* faster, particularly for larger lattice sizes,
but requires a compiled library to work.

The files present are:

* ising.py:
    You will most likely only need to modify this file.
    This contains the main program. It requires the following:
    1. Python3 (sorry, Python2 will not work)
    2. several modules:
        - logging
        - numpy
        - matplotlib
    It will interface with either the Python or C++ lattice simulations.

    It contains many named inputs from the command line. Each input is
    made with [word]:[value] where value is (case insensitive for boolean values) 
    T/F/True/False/number/string.
    
    Numbers without decimal points are considered integers.

    The inputs options are:
    value           default   use
    t_min           1.6       minimum temperature
    t_max           3.6       maximum temperature (will not be included)
    t_step          0.1       temperatures will be analyzed from t_min to t_max
                              in increments of t_step and exlude t_max 
    t_top           4.0       temperature parameter used by default temperature annealing
    N               10        side of lattice, for a total of N*N sites
    n_steps         10000     number of steps to evolve the ising lattice
    n_burnin        2000      parameter for default annealing process
    n_analyze       5000      number of steps measured for mean and std statistics
    B               0         magnetic field strength
    flip_perc       0.1       ratio of lattice-site that may flip in a step
                              (note: this is exact in the C++ implementation,
                              but only the average in the Python implementation)
    dir_out         data      name of output directory
    plots           False     option to distplay E and M plots after finishign running
                              note: note available for multiprocess
    print_inp       False     echo all of the input options to the command line
    use_cpp         True      use C++ implementation; will switch off if the
                              shared library is not present
    date_output     False     add YYYYMMDD-HHMMSS postfix to output directory name
    file_prefix     ''        string to prefix output files
    multiprocess    False     use multiple processes
    skip_prog_print False     do not provide progress output information
                              note: progress info is not avialable in multiprocess


* IsingLattice_python.py:
    This is the Python implementation of the Ising lattice simulation.

* IsingLattice_cpp.py:
    This is a Python wrapper to talk to the compiled C++ library
    for the IsingLattice

* ising_lattice_lib.cxx:
    This is the C++ library to compile for the ising simulation
    It compiles ot ising_lattice_lib.so ('.so' for 'shared object')

* linux_compile_c_library.sh:
    This is a script to compile ising_lattice_lib.cxx in *NIX systems
    It should work on the cluster and on your PC

* slurm.input:
    This is a sample slurm input script for running concurrent jobs on 
    the grace cluster

* grace_environment.sh
    On the grace gluster, checkout your own node (unless you are just
    submitting jobs to slurm) and then run
    `source grace_environment.sh`

To be done:
    There are at least two major improvements to made in the code.

    1. Add a 'step_SwendsenWang' or a 'step_Wolf' (see lab handout for refernce)
       These algorithms overcome a large degree of the critical slowing that
       occures as the lattice approaches critical temperature, and would 
       allow some much better measurements.

    2. The spin correlations are calculated only for the final step. It
       would likely be *much* better to caclulate the avege and std of
       these values, as is done for the E and M parameter.