This is a Reproduction Package as described in the manuscript "Three Empirical Principles for Computational Reproducibility and their Implementation: The Reproduction Package" by M. S. Krafczyk, A. Shi, A. Bhaskar, D. Marinov, & V. Stodden.
This Reproduction Package does not contain the run_all.sh script because all computations run extremely quickly, and there is no need to run a reduced set of experiments.
"A modification of Newton's method for nondifferentiable equations" is an implementation of the techniques presented in A modification of Newton's method for nondifferentiable equations
Instructions were tested using Docker version 18.06.0-ce, build 0ffa825, on Ubuntu 16.04.5 LTS.
docker build -t ${DOCKER_IMAGE_NAME} .
To start a container for the Docker image:
docker run -it --rm -v $(pwd):/Scratch ${DOCKER_IMAGE_NAME}
Within the Docker container, to run everything, computational scripts for experiments and visualization scripts, run
./run.sh
Please be aware of computational efforts for the scripts. More details can be found here.
See sections below provide for details about the individual steps.
Within the Docker container, run
./computation.sh
Output will be computed_results.txt.
Expected results is `expected_results.txt'.
The script also runs
python check_results.py expected_results.txt computed_results.txt
to automatically check the computed_results.txt against expected_results.txt
We kept track of our progress and issues inside notes.txt. We also have an
jupyter notebook showing this progress over time ReproducibilityPlot.ipynb.
We want acknowledge the authors for their fine work on this experiment. We succeeded with this project where many others had failed. The authors should be commended on putting together high quality work.
This work was partially funded by NSF grants OAC-1839010 and CNS-1646305.