Running

Parallelization

MPI

Pthreads

Total number of measurements

Hybrid GPU-CPU systems

TBA

Input files

TBA

Cooldown

TBA

cooldown.py

We provide a python script that generates directories, input files and batch scripts for a cooldown. By default this script and the provided template input files are configured for DCA⁽⁺⁾ calculations of the 2D single-band Hubbard model with on-site Coulomb interaction U and fixed density d.

Usage

1. Create a work directory that contains:

   • the script cooldown.py,
   • the input templates input_sp.json.in and input_tp.json.in,
   • and the binaries main_dca and main_analysis.

2. Configure input_sp.json.in and input_tp.json.in:

   • Provide the parameter groups for your selected build options (cluster solver and lattice model).
   • Set the simulation parameters that are not specified via PLACEHOLDERS (capitalized string values) by cooldown.py.
   • If threaded Monte Carlo integration has been enabled, choose an appropriate number of walker and accumulator threads according to your system's specification.

3. Create a batch script template in your work directory.
   The file should contain the placeholder JOBS, which will be substituted with the schedule of dca/analysis runs for the cooldown (see example above). In addition, the following placeholders can be used:

   • APPLICATION: dca | analysis
   • HUBBARDU: value of the on-site Coulomb interaction U
   • DENS: value of the density d
   • SIZE: DCA cluster size

   We proivde batch script examples for SLURM (GPU, multi-core), PBS and local machines.

4. Configure the EDIT-block of cooldown.py:

   • batch_tmpl: name (including path) of your patch script template
   • run_command: the command (including options) to run (MPI) programs, e.g. 'mpirun -n 8'
   • simulation parameters

5. Execute the python script with the option dca:

    python cooldown.py dca
   

6. Revise the generated batch script for the dca runs, adjust resources if necessary, and submit/execute it.

7. When all dca jobs are finished, execute the python script with the option analysis:

    python cooldown.py analysis
   

8. Revise the generated batch script for the analysis runs, adjust resources if necessary, and submit/execute it.