Experimental Optimization by the Simplex Method

This simple search method allows multidimensional optimization to be carried out experimentally the functional dependence of the target function on the individual parameters need not be known. The value of the target function (e.g., the yield of a 

1) Determination of the target quantity at the comers of the simplex (n + 1) experiments 

3) Generation of a new corner x + by reflection of the triangle about the side opposite to the worst corner 

6) If y is worse than all other results, go to point 7, otherwise point 2 

7) Selection of the second worst corner as then go to point 3 

---

In an experimental optimization by the simplex method, the yield of a process was determined as a function of two variables xi and X2- The following table shows the results of the initial simplex with three experiments ... 

---