The *Simulation.R files only create one .init.R file, but should create one PER CHAIN. Currently each chain begins at the same point in the parameter space, which is problematic and doesn't allow us to properly test for convergence.
We need a scheme to create random initial estimates for each chain used when running Stan.
The
*Simulation.Rfiles only create one.init.Rfile, but should create one PER CHAIN. Currently each chain begins at the same point in the parameter space, which is problematic and doesn't allow us to properly test for convergence.We need a scheme to create random initial estimates for each chain used when running Stan.