Use NMinimize for MonteCarloFitting???
- To: mathgroup at smc.vnet.net
- Subject: [mg131938] Use NMinimize for MonteCarloFitting???
- From: dilana78 <mail at vasso.de>
- Date: Mon, 4 Nov 2013 23:15:59 -0500 (EST)
- Delivered-to: l-mathgroup@mail-archive0.wolfram.com
- Delivered-to: l-mathgroup@wolfram.com
- Delivered-to: mathgroup-outx@smc.vnet.net
- Delivered-to: mathgroup-newsendx@smc.vnet.net
Hi, I am trying to estimate parameters from my statistical model by curve fitting. The model has several paramters which follow truncated normal distributions.The combination of the parameters within the model is non-linear.I run a MonteCarlo simulation to get a distribution of results from the model and this result I need to fit to a dataset. Coming from this problem I coded an objective Function which needs to be minimized which then gives me the optimal paramters. If I implement this in mathematica with NMinimize I get the following error: RandomVariate::posprm: Parameter ss at position 2 in NormalDistribution[mmu,ss] is expected to be positive. >> Can't NMinimize handle non analytic functions (I need to evaluate in each step a pool of data with MonteCarlo)? If it won't work with NMinimize, is there something I can fit data to a function whose parameters each follow certain distributions? (z.b. a Log[c], a and c are paramters which are both (truncated) normal distributed). thanks, Dilana