New release - SUMO Toolbox 2016a

We are proud to release the SUMO Toolbox 2016a. This new release can be downloaded through the download page. This release addresses some important bugs caused by Matlab 2015a and newer. Most importantly, crashes in the SUMO toolbox caused by changes in the Matlab plot system and user interface (guide) have been fixed. The Kriging-based optimization framework in SUMO has been updated with new state-of-the-art techniques (slice sampling, knowledge gradient for deterministic systems, better integration of the GPML toolbox, etc.).


  • Fixed bugs caused by Matlab 2015a and newer (guiPlotModel)
  • Further enhanced the Gaussian process model
  • Added slice sampling support for Kriging and Gaussian process models
  • Added knowledge gradient optimization for deterministic noise-free simulation
  • Refactored SVM models
    • LS-SVM and libSVM are now in two separate classes, change your configuration file accordingly (see default.xml)
  • Added DiscreteOptimizer for discrete optimization problems
  • Added two new measures: TrainingTimeMeasure and EvaluationTimeMeasure
  • Overhaul of the EGO modelbuilder for Bayesian optimization of the hyperparameters
  • Various bugfixes