Sequential Experimental Design (SED) Toolbox

SED ToolboxBrief description

The SED Toolbox (Sequential Experimental Design) is a powerful Matlab toolbox for sequential Design of Experiments (DoE). In traditional experimental design, all the design points are selected up front, before performing any (computer or real-life) experiment, and no additional design points are selected afterwards. This traditional approach is prone to oversampling and/or undersampling, because it is often very difficult to estimate the required number of design points in advance. The SED Toolbox solves this problem by providing the user with state-of-the-art algorithms that generate an experimental design in a sequential way, i.e. one point at a time, without having to provide the total number of design points in advance. This is called sequential experimental design. The SED Toolbox was designed to be extremely fast and easy to use, yet very powerful.


Documentation | Download instructions | Features | Screenshots |References

Key features

  • Includes several highly efficiënt space-filling sequential experimental design algorithms, which generate designs competitive with state-of-the-art one-shot experimental design techniques such as Latin hypercubes.

  • Optimized for speed: several algorithms are included, ranging from extremely fast, even in high-dimensional problems, to slightly slower (yet still relatively fast).

  • All methods support constraints of any kind (linear and nonlinear). These constraints can be specified with a Matlab function.

  • Includes the powerful LOLA-Voronoi algorithm, which distributes design points sequentially according to the nonlinearity of the problem, following the assumption that nonlinear regions are more difficult to understand than linear ones.

  • Support for rectangular input spaces: not every input necessarily equally important, so weights can be given to each input and the algorithms will take this into account when selecting design points.

  • Easy to use and configure, with examples amptly available in the documentation and in the distribution.

  • Proper Object Oriented design.

Download instructions

The SED Toolbox is available in 2 different forms:

  • Fully functional Open Source version (AGPLv3 license) for non-commercial, personal, academic use.
  • Proprietary version for commercial use.

Details can be found in the License Terms. You can download the open source version from the menu on the left, for non-commercial, personal, academic use only. For details about a commercial license please use the contact form. Flexible licences are also available for research partners (aiming at collaboration, data exchange, or joint publications). Please refer to the documentation for instructions on how to configure and run the SED Toolbox in Matlab.

Screenshots

Comparison of SED Toolbox algorithms and other popular sequential and non-sequential experimental design methods: Screenshot 1Screenshot 2

2D and 3D designs generated and plotted by the SED Toolbox:

Screenshot 1Screenshot 2

References

  • Generating Sequential Space-filling Designs Using Genetic Algorithms and Monte Carlo Methods pdf
    K. Crombecq and T. Dhaene
    Simulated Evolution And Learning (SEAL-2010), Kanpur, India, pp. 80-84
  • Space-filling Sequential Design Strategies for Adaptive Surrogate Modelling pdf
    K. Crombecq, I. Couckuyt, D. Gorissen and T. Dhaene
    Soft Computing Technology in Civil, Structural and Environmental Engineering, 2009 (CSC 2009)
  • A Novel Sequential Design Strategy for Global Surrogate Modeling pdf
    K. Crombecq, D. Gorissen, L. De Tommasi and T. Dhaene
    Proceedings of the 41th Conference on Winter Simulation, Austin, Texas, December 2009, pp. 731-742