MIT Uncertainty Quantification (MUQ) library. In a nutshell, MUQ is a collection of tools for constructing models and a collection of uncertainty quantification (UQ)–focused algorithms for working on those models. Our goal is to provide an easy and clean way to set up and efficiently solve UQ problems. On the modelling side, we have a suite of tools for: Combining many simple model components into a single sophisticated model. Propagating derivative information through sophisticated models. Integrating ordinary differential equations and differential algebraic equations (via Sundials). Furthermore, on the algorithmic side, we have tools for: Performing Markov chain Monte Carlo (MCMC) sampling; Constructing polynomial chaos expansions (PCE); Computing Karhunen-Loeve expansions; Building optimal transport maps; Solving nonlinear constrained optimization problems (both internally and through NLOPT); Regression (including Gaussian process regression)