Kodiak is a software library that integrates numeric and symbolic computation into a generic framework for the solution of numerical, possibly non-linear, problems over hyper-rectangular variable and parameter domains. At the basis of Kodiak there is a formally verified branch and bound algorithm and self-validating enclosure methods based on interval arithmetic and, for polynomials and rational functions, Bernstein expansions. Kodiak provides tools for paving a system of constraints, for solving constrained global optimization problems, and for computing the equilibria and bifurcation sets for systems of ordinary differential equations.

Kodiak is available under NASA's Open Source Agreement from GitHub.

Publications

• Andrew Smith, César Muñoz, Anthony Narkawicz, and Mantas Markevicius, A Rigorous Generic Branch and Bound Solver for Nonlinear Problems, Proceedings of the 17th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing, IEEE Computer Society Conference Publishing Services, 2015. BibTeX Reference.
• Andrew Smith, César Muñoz, Anthony Narkawicz, and Mantas Markevicius, Kodiak: An Implementation Framework for Branch and Bound Algorithms, Technical Memorandum, NASA/TM-2015-218776, July 2015. BibTeX Reference.
• Andrew Smith, Luis Crespo, César Muñoz, and Mark Lowenberg, Bifurcation Analysis Using Rigorous Branch and Bound Methods, IEEE Multi-Conference on Systems and Control (MSC 2014), October 2014. BibTeX Reference.
• Anthony Narkawicz and César Muñoz, A Formally Verified Generic Branching Algorithm for Global Optimization, Proceedings of the Fifth Working Conference on Verified Software: Theories, Tools and Experiments (VSTTE 2013), Lecture Notes in Computer Science, Vol. 8164, pp. 326-343, 2014. BibTeX Reference.
• Luis Crespo, César Muñoz, Anthony Narkawicz, Sean Kenny, and Daniel Giesy, Uncertainty Analysis via Failure Domain Characterization: Polynomial Requirement Functions, European Safety and Reliability Conference, September 2011. BibTeX Reference.

*The Kodiak logo was designed by Mahyar Malekpour (NASA).

The tag identifies links that are outside the NASA domain