Hi FEFLOW forum,
I am developing a module that carries out Markov Chain Monte Carlo inversion of observed drained volumes of water from columns of sediment. The model utilizes the Van Genuchten model for the variably saturated zone. In the inversion scheme I need to pertubate model parameters randomly and then set the updated parameter values for all elements in the model. I do this successfully for all parameters except the "fitting exponent" n in the VG model. I use the IfmSetMatUnsatFittingExponent() function and [u]using the inspector I can see that the value is set correctly[/u] however simulation results are non consistent (negative drainage). Omitting the line of code containing IfmSetMatUnsatFittingExponent(...) solves all problems. I use the function like this:
for(j = 0; j < NELE; j++)
{
..
IfmSetMatUnsatFittingExponent(pDoc,j,p_try[6]); //Bad!
..
}
p_try is the trial parameter vector and NELE is the total number of elements in the model.
It works like a charm with other similar functions like e.g. IfmSetMatUnsatFittingCoefficient() but perhaps I am using it wrong? Or could there be a bug in the API function?
Thanks in advance and best regards :)
Søren
