The aim of EasyVVUQ is to facilitate verification, validation and uncertainty quantification (VVUQ) for a wide variety of simulations. While very convenient for simple cases, EasyVVUQ is particularly well suited in situations where the simulations are computationally expensive, heterogeneous computing resources are necessary, the sampling space is very large or book-keeping is prohibitively complex. It coordinates execution using an efficient database, it is fault tolerant and all progress can be saved.
Here are some examples of questions EasyVVUQ can answer about your code:
- Given the uncertainties in input parameters, what is the distribution of the output?
- What percentage of the output variance each input parameter contributes?
It also lets you construct surrogate models that are cheaper to evaluate than the complete simulation.
The high-level overview of the library is avalable at our readthedocs.
For the quick start with EasyVVUQ we reccommend to check our basic interactive tutorial available here.
Available analysis and sampling methods:
- Polynomial Chaos Expansion
- Stochastic Collocation
- Dimension-adaptive Stochastic Collocation for high-dimensional inputs (incl notebook in
./tutorialsand theoretical tutorial) - Simplex Stochastic Collocation for irregular outputs (incl notebook in
./tutorialsand article) - Monte Carlo Sensitivity Analysis
- Markov-Chain Monte Carlo
EasyVVUQ also supports building surrogate models using:
- Polynomial Chaos Expansion
- Stochastic Collocation
- Gaussian Processes
Supported computing resources:
- Traditional clusters
- Kubernetes clusters
The easiest way to get familiar with the provided functionality is to follow the tutorials (*.ipynb files) in our Binder.
To use the library you will need Python 3.7+.
If you are unsure of the version of python your default pip works for type:
pip --version
If the output ends with (python 2.7) you should replace pip with pip3 in the following commands.
The following should fully install the library:
pip install easyvvuq
To upgrade the library use:
pip install easyvvuq --upgrade
Alternatively, you can manually install EasyVVUQ. First clone the repository to your computer:
git clone https://github.com/UCL-CCS/EasyVVUQ.git
Note: As above, you need to be sure you are installing for Python 3 - if necessary replace pip with pip3 and python with python3 in the commands below.
We are trying to keep dependencies at a minimum but a few are inevitable, to install these, install the EasyVVUQ library itself and build a test case use:
cd EasyVVUQ/
bash install_EasyVVUQ.sh
You can find the EasyVVUQ API documentation on our GitHub Pages.
Richardson, R A, Wright, D W, Edeling, W, Jancauskas, V, Lakhlili, J and Coveney, P V. 2020 EasyVVUQ: A Library for Verification, Validation and Uncertainty Quantification in High Performance Computing. Journal of Open Research Software, 8: 11. DOI: 10.5334/jors.303.
Wright, D.W., Richardson, R.A., Edeling, W., Lakhlili, J., Sinclair, R.C., Jancauskas, V., Suleimenova, D., Bosak, B., Kulczewski, M., Piontek, T., Kopta, P., Chirca, I., Arabnejad, H., Luk, O.O., Hoenen, O., Weglarz, J., Crommelin, D., Groen, D. and Coveney, P.V. (2020), Building Confidence in Simulation: Applications of EasyVVUQ. Adv. Theory Simul., 3: 1900246. DOI: 10.1002/adts.201900246.
Development was funded by the EU Horizon 2020 project VECMA.