Latin hypercube sampling mixed code#

An example would be if there is larger local variation or variance in some part of the parameter or design space.ДЄn application where this is useful: A standard example for spline fitting is a motorcycle dataset. Have you seen anything about deviating from uniformly distributed design to put more points in 'interesting' local neighborhoods. So for this LHS would be similar to Halton sequences, as a numerical integration tool and not literally for A related but different question? (just an idea right now In discrete choice models like mixed multinomial logit or probit models the likelihood function needs to be numerically integrated, using gaussian quadrature, halton sequences or monte carlo, or, based on this reference modified LHS. I had thought of LHS as a method to create a design matrix for actual experiments, but it is also just another method for numerical integration I have many different topics that I need to keep track of or need to catch up with, so I cannot keep up, but I will come back to it every once in a while. It would be good if you could work on whatever you think is useful in this area. I browsed a bit in the literature but have barely some superficial ideas about this. If those are mainly for the DOE use cases, then they could also go into a new doe or experimental subfolder.Д®ssentially this is you area for now. However, I don't know about the other sequences and methods that you describe. I know internal use cases for halton in at least discrete and genmod, and most likely for all models for numerical or Monte Carlo integration, so tools would be an appropriate location.

Latin hypercube sampling mixed code#

In terms of organizing the code and it's location: (I didn't go through the details of their paper) Which would mean that we don't have to compute and throw away numbers if we want to skip numbers in the halton sequence, as proposed by Kocis and Whiten. If I read van_der_corput correctly, then we can compute the sequence for any i separately. separate prime numbers code from sequence code. Also maybe outsource the find n prime numbers loop from the halton function to a helper function would be useful, i.e. They won't change and it is not necessary to recompute them each time. I think we can hardcode the first few prime numbers as in the old halton. Even if it works this way not modifying outer loop variable is an easier to understand design. In van_der_corput use a different loop variable in the, while i > 0. Halton looks fine going through van der Corput, but I have no idea if there are numerical advantages to either version. A suggestion: put the gist functions in a new module