Simplex Pro

What is Simplex Pro?

The simplex was constructed for optimizing the numerical parameters of chemical experiments such as temperature, pressure, concentration, and flow rate to achieve the best result (objective function) e.g. conversion, yield or selectivity. The numerical values of the reaction parameters are input into the software along with the results. The Simplex will then calculate another set of parameters that will yield a better result.

Operation of the algorithm

The software implements the widespread Nelder-Mead (or Downhill Simplex) algorithm. Based on experiments performed at Thalesnano, by using this method the number of optimization experiments can be significantly decreased.

The operation of the algorithm is a simple 2 step process and is performed as follows:

  • The user performs N+1 experiments (where N is equal to the number of variables entered into the optimization)
  • The user then makes an additional experiment for every optimization cycle with parameter values defined by the algorithm.

The advantage of this method over others is that after performing the initial measurements the Simplex converges to an optimum faster, so the optimum value of the objective function can be found by making only a few experiments. If the values of objective function differ only slightly after 2-3 consecutive experiments, then it can be presumed that the nearly optimal parameter settings have been found.

Special Features

Initial simplex

The current version of the system generates a regular simplex into the geometrical center of parameter space defined by the parameter boundaries.

Rounding values

The system displays the experiment parameters with an accuracy defined by the user.

Considering boundaries

If the parameters given by the algorithm exceed the boundaries previously defined by the user, the software modifies the exceeding value to the maximum boundary limit value.

Repeated measurement points

If the algorithm suggests an experimental point, which is included in the previously measured experiment, the system will give a warning message.