3
$\begingroup$

I am working in a project to assist an experimental team in optimizing reaction conditions. The problem involves a large number of dimensions, i.e. 30+ reactants which we are trying out different concentrations to achieve the highest yield of a certain product.

I am familiar with stochastic optimization methods such as simulated annealing, genetic algorithms, which seemed like a good approach to this problem. The experimental team proposes using design of experiments (DoE), which I'm not too familiar with.

So my question is, what are the advantages/disadvantages of DoE (namely fractional factorial and response surface method I believe) versus stochastic optimization methods, and are there use cases where one is preferred over the other?

$\endgroup$
2
  • $\begingroup$ I'm not too familiar with either approach, but I have some idea about how DoE works. Could you elaborate on how the stochastic optimization methods work to assist with experimental design? $\endgroup$
    – mkt
    Jun 28 at 9:33
  • $\begingroup$ Hi @mkt, let's say I am using genetic algorithm, I will generate a random set of initial starting points in the entire search space and evaluate the outcomes of each through the experiments. After which, the k-best performing points goes through "reproduction" to produce offspring candidate points by taking some linear combinations of their values. The poor performing points are eliminated. Some new points are also generated randomly, and in the next iteration, all of these points are evaluated and this process goes on for n number of iterations until convergence. $\endgroup$ Jun 28 at 9:40
0
$\begingroup$

Stochastic optimization (SO) and the design of experiments (DOE) are entirely differ from each other, in terms of why using it, they are not orthogonal and simply you might use both of them at the same time. You might consider DOE as a measurement of how much changes in genes (variation) affect the production, (ex. we might have n-production comes from SO), which implies that DOE is convenient for optimizing multiple-productions, but SO aims to optimize an initial production X.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.