MADS (Model Analysis & Decision Support)
MADS is an integrated high-performance computational framework for data/model/decision analyses.
MADS can be applied to perform:
- Sensitivity Analysis
- Parameter Estimation
- Model Inversion and Calibration
- Uncertainty Quantification
- Model Selection and Averaging
- Model Reduction and Surrogate Modeling
- Decision Analysis and Support
MADS utilizes adaptive rules and techniques which allow the analyses to be performed efficiently with minimum user input.
MADS provides a series of alternative algorithms to execute various types of data-based and model-based analyses.
MADS can efficiently utilize available computational resources.
MADS has been extensively tested and verified.
Documentation
MADS documentation, including description of all modules, functions, and variables, is available at:
- GitHub (always up-to-date)
- ReadtheDocs (outdated)
- LANL (outdated).
MADS information is also available at mads.gitlab.io and madsjulia.github.io
Detailed demontrative data ananlysis and model diagnostics problems are availble as Julia scripts and Jupyter notebooks. See also below.
Installation
In Julia REPL, execute:
import Pkg; Pkg.add("Mads")
To utilize the latest code updates use:
import Pkg; Pkg.add(Pkg.PackageSpec(name="Mads", rev="master"))
Testing
Execute:
import Mads; Mads.test()
or
import Pkg; Pkg.test("Mads")
Getting started
To explore getting-started instructions, execute:
import Mads; Mads.help()
Examples
Various examples located in the examples
directory of the Mads
repository.
A list of all the examples is provided by:
Mads.examples()
A specific can be executed using:
Mads.examples("contamination")
or
include(joinpath(Mads.dir, "examples", "contamination", "contamination.jl"))
This example will demonstrate various analyses related to groundwater contaminant transport.
To perform Bayesian Information Gap Decision Theory (BIG-DT) analysis, execute:
Mads.examples("bigdt")
or
include(joinpath(Mads.dir, "examples", "bigdt", "bigdt.jl"))
Notebooks
To explore evailable notebooks, execute:
Mads.notebooks()
Docker
docker run --interactive --tty montyvesselinov/madsjulia
Related Julia Packages
- SmartTensors: Unsupervised and Physics-Informed Machine Learning based on Matrix/Tensor Factorization
- RegAE: Regularization with a variational autoencoder for inverse analysis
- Geostatistical Inversion with randomized + sketching optimization