Hi,
I have been using the uq_createAnalysis funtions a lot lately, and noticed that I run into errors now. This is a strange behaviour, since I was able to run the exact same code just 5 days ago. Can anyone help me out/clarify this?
Kind regards,
Mathieu
Did you recently update your UQLab? If yes, what was your old version and what is your current version? Also a minimal reproducible example would be helpful.
Best regards
Styfen
Hi @styfen.schaer ,
Thanks for the swift response.I am currently using version 2.0. It’s a strange occurrence, as I don’t recall updating, yet the same script I have been using just a week ago is not functioning anymore.
I have also checked the dimensions of the surrogate model output and the validation data, and they do match.
In any case, here is a snippet from my code:
myData.y = exp_f; % (m)
myData.Name = 'Power';
%% Prior distribution of model parameters
Input.Marginals(1).Name = 'alpha';
Input.Marginals(1).Type = 'Uniform';
Input.Marginals(1).Parameters = [1.5 5];
Input.Marginals(2).Name = 'beta';
Input.Marginals(2).Type = 'Uniform';
Input.Marginals(2).Parameters = [0.15 0.45];%[0.125 0.175];
Input.Marginals(3).Name = 'ka';
Input.Marginals(3).Type = 'Uniform';
Input.Marginals(3).Parameters = [0.3 1];
Input.Marginals(4).Name = 'kb';
Input.Marginals(4).Type = 'Uniform';
Input.Marginals(4).Parameters = [-0.005 0.005];
Input.Marginals(5).Name = 'obja';
Input.Marginals(5).Type = 'Uniform';
Input.Marginals(5).Parameters = [0 1];
Input.Marginals(6).Name = 'objb';
Input.Marginals(6).Type = 'Uniform';
Input.Marginals(6).Parameters = [0 .8];
Input.Marginals(7).Name = 'objc';
Input.Marginals(7).Type = 'Uniform';
Input.Marginals(7).Parameters = [0 0.6];
Input.Marginals(8).Name = 'objd';
Input.Marginals(8).Type = 'Uniform';
Input.Marginals(8).Parameters = [-.6 -0.01];
Input.Marginals(9).Name = 'd';
Input.Marginals(9).Type = 'Uniform';
Input.Marginals(9).Parameters = [0.6 1.0];
Input.Marginals(10).Name = 'eta';
Input.Marginals(10).Type = 'Constant';
Input.Marginals(10).Parameters = 1;
Input.Marginals(11).Name = 'pp';
Input.Marginals(11).Type = 'Uniform';
Input.Marginals(11).Parameters = [1.5 2.5]; % pp = 1.50
Input.Marginals(12).Name = 'TIExp';
Input.Marginals(12).Type = 'Uniform';
Input.Marginals(12).Parameters = [1 3];
myInput = uq_createInput(Input) ;
DiscrepancyOptsKnown.Type = 'Gaussian'; % known
DiscrepancyOptsKnown.Parameters = 1e-2;
model = load([SaveDir surrogatemodelname]); % saved surrogate as .mat file
BayesOpts.ForwardModel.Model = model.mySurrogateModel;
% Solver options
Solver.Type = 'MCMC';
Solver.MCMC.Sampler = 'AIES'; % MH
Solver.MCMC.Steps = 100; % 150
Solver.MCMC.NChains = 1000; %600 % 1000
% Bayes options
BayesOpts.Type = 'Inversion';
BayesOpts.Data = myData;
BayesOpts.Prior = myInput;
BayesOpts.Solver = Solver;
BayesOpts.Discrepancy = DiscrepancyOptsKnown;
BayesOpts.Display = 'quiet';
fprintf('Starting calibration') % Run Calibration part
myBayesianAnalysis = uq_createAnalysis(BayesOpts);
Hi again,
Small clarification. There was no issue with uqlab on itself. Issues arose due to interference with loaded software packages. However, the issue is solved now, and so is this ticket.