Joey Huchette

Papers

  • Very small algebraic representations for disjunctive sets.
    With Juan Pablo Vielma.
    Working paper.
  • Nonconvex piecewise linear functions: Advanced formulations and simple modeling tools. [arXiv]
    With Juan Pablo Vielma.
    Submitted.
  • A combinatorial approach for small and strong formulations of disjunctive constraints. [arXiv]
    With Juan Pablo Vielma.
    Under review (2nd round).
  • JuMP: A modeling language for mathematical optimization. [arXiv]
    With Iain Dunning and Miles Lubin.
    SIAM Review, 2017.
    Winner of the 2016 INFORMS Computing Society Prize.
    Winner of the 2016 MIT Operations Research Center Best Student Paper Award.
    Winner of the 2015 COIN-OR Cup.
  • Beating the SDP bound for the floor layout problem: A simple combinatorial idea. [arXiv]
    With Santanu Dey and Juan Pablo Vielma.
    To appear, Information Systems and Operational Research, 2017.
  • Strong mixed-integer formulations for the floor layout problem. [arXiv]
    With Santanu Dey and Juan Pablo Vielma.
    Information Systems and Operational Research, 2017.
  • Extended formulations in mixed integer conic quadratic programming. [arXiv]
    With Juan Pablo Vielma, Iain Dunning, and Miles Lubin.
    Mathematical Programming Computation, 2016.
  • Parallel algebraic modeling for stochastic optimization. [ACM]
    With Miles Lubin and Cosmin Petra.
    In Proceedings of HPTCDL 2014.
  • Taming parallel I/O complexity with auto-tuning. [ACM]
    With Babak Behzad, Huong Luu, Surendra Byna, Prabhat, Ruth Aydt, Quincey Koziol, and Marc Snir.
    In Proceedings of SC 2013.

Presentations

  • Nonconvex piecewise linear functions: Advanced formulations and simple modeling tools
    • INFORMS 2017 (upcoming)
    • MIP 2017 (poster)
      Best Poster Award (Honorable Mention)
  • PiecewiseLinearOpt.jl: Solving optimization problems containing piecewise linear functions
    • JuMP Developers Meetup 2017 (video)
  • Mixed-integer sum of squares optimization: Computation and application
    • SIAM Opt 2017
  • A combinatorial approach for small and strong formulations of disjunctive constraints
    • INFORMS 2016
    • MIP 2016 (poster)
  • Strong mixed-integer formulations for the floor layout problem
    • INFORMS 2015
    • Argonne National Laboratory (2015)
    • ISMP 2015
    • MIP 2015 (poster)
    • INFORMS 2014
    • MIP 2014 (poster)
  • Modeling optimization problems with JuMP in Julia
    • Carnegie Mellon (2014, joint with Miles Lubin)
    • Georgia Tech (2014)
    • Berkeley (2014, joint with Iain Dunning and Miles Lubin)
  • JuliaOpt - Optimization packages for Julia
    • JuliaCon 2015 (workshop, joint with Iain Dunning, Miles Lubin, and Madeleine Udell)
    • JuliaCon 2014 (joint with Iain Dunning)

Teaching

  • Taught one session for MIT Software Tools for Business Analytics (January 2017).
  • Co-taught one session for MIT 15.S60: Computing in Optimization and Statistics (January 2017).
  • Guest lecturer (two classes) for MIT 15.093J: Optimization Methods (Fall 2016).
  • Teaching assistant for MIT 15.083J: Integer Programming and Combinatorial Optimization (Spring 2016).
  • Organized MIT 15.S60: Software Tools for Operations Research (January 2015 and January 2016). Course materials: 2015 and 2016.
  • Co-taught a total of 6 recitations on JuMP for MIT 15.058, 15.081J, and 15.085J (2014).

Software

  • JuliaOpt - a suite of optimization software in Julia. Includes:
    • JuMP: an algebraic modeling language for linear, integer, and nonlinear optimization.
    • Convex.jl: a "disciplined convex programming" modeling language.
    • Efficient wrappers for over a dozen state-of-the-art solvers (Gurobi, Ipopt, Mosek, etc.), with a unified, abstract interface.

About Me

I'm a fifth-year PhD student in the Operations Research Center at MIT, with an expected graduation date of Spring 2018. I am advised by Juan Pablo Vielma and was previously supported by the NSF Graduate Research Fellowship. I received my B.A. in Applied Mathematics from Rice University, where I worked with Beatrice Riviere and Hadley Wickham.

Research Interests

I am broadly interested in operations research and the theory and application of optimization. Much of my current work is concerned with mathematical formulations, or how to translate a decision problem to a mathematical description we can solve efficiently.

I'm also interested in all aspects of computational optimization, especially user-facing tools for modeling and for developing advanced algorithms.

CV (Updated 8/6/2017).

GitHub.

Google Scholar.

Email