tfg-H16b

The code for my bachelor's thesis (TFG) in Informatics Engineering, carried out under the supervision of Professor Grigori Astrakharchik.

A numerical exploration of the second part of Hilbert's 16th problem.

Requirements

Install dependencies with Poetry

The project runs on Python 3.6.9. You may want to install it with pyenv.

It uses Poetry for dependency management. Poetry declares package dependencies in pyproject.toml, locks them to a specific version in poetry.lock and installs them on an isolated virtual environment for the project. Installation instructions for Poetry are here.

With Python 3.6.9 and Poetry installed, just run the following commands in the project's root directory:

poetry install
poetry shell

The application will probably work on later versions of Python and the dependencies, but I can't make any guarantees.

Build the Docker image

Alternatively, the project provides a Dockerfile to build an environment with all the requirements installed.

You can execute script/spawn_docker_shell.sh to run the container with the necessary volumes.

Usage

Command-line interface

The application has a single entry point: app/cli.py. It provides two commands: compute for running the jobs and view for visualizing their results.

For compute, you are required to provide a job configuration file, the output path is optional:

python -m app.cli compute -j jobs/sign_change_test.yaml -o results/example.json

For view, you are required to provide an input file:

python -m app.cli view -i results/example.json

If you run this in an environment without GUI, like the Docker shell, the plot is saved to results/example.png.

Find more information about the commands and their options in docs/cli-reference.txt.

Job configuration files

Job configuration files define the parameters that inform a compute execution.

Here's an example:

# jobs/sign_change_test.yaml
job_type: scan_sign_change
logging:
  level: DEBUG
  file:
dynamical_system:
  a1: 1
  b1: 1
  c1: 0
  alpha1: 0
  beta1: 1
  a2: -10.0
  b2: 2.2
  c2: 0.7
  alpha2: -72.7778
  beta2: 0.0015
trajectory_integration:
  precision: -3
  limit: 10.0
phase_space_scan:
  resolution: -2
  min_x: -1.0
  max_x: 1.0
  min_y: -0.5
  max_y: 0.5

The application accepts the following job types: compute_trajectory, scan_heatmap, scan_sign_change, scan_interloop_v2 and scan_interloop_v3.

For more information on writing configurations for a particular job type, refer to docs/job-config-reference/.

Utility scripts

Directory scripts/ provides a number of utilities that streamline interactions with the application.

For instance, there is less_last_result.sh to display the most recent results file in human-readable format, tail_last_log.sh to watch the log for a running job, or test.sh to perform a visual test for a given job type.

You can also $ source script/functions to load functions like last_h16b_json, which returns the path to the latest compute results file.

Lastly, there are a few scripts (prefixed 2mt and mt) to access and operate within the MinoTauro GPU cluster.

All scripts should be executed from the project's root directory. Those that view results require Poetry.

License

MIT