Hi
I am trying to figure out how to represent my problem in UQLab. My model consists of two random variables and a parameter leading to a single output. It’s given as:
where \tau \in \mathbb{R}^1, \tau_0\sim \mathcal{N}(\mu = 50, \sigma = 0.61), C\sim \mathcal{LN}(\mu_{\ln C} = -2.73, \sigma_{\ln C} = 0.47), discrepancy \varepsilon \sim \mathcal{N}(0, \sigma_{\varepsilon}), , \sigma_{\varepsilon} \sim \mathcal{LN}(\mu = 1, \sigma = 5), and t is a deterministic parameter.
I have two data groups for t = 10 and 15, respectively. If I create two data groups then I end up having two different discrepancies, which I don’t want in my problem.
For this problem, I specified the model as a single model with two outputs with t = 10 and 15 as follows:
func = @(X) [X(1)-X(2)*10, X(1)-X(2)*15];
ModelOpts.Name = 'Forward model';
ModelOpts.mHandle= func;
myForwardModel = uq_createModel(ModelOpts);
Based on some suggestions in this forum, I converted my data into a single datagroup and specified MOMap
as follows:
% group 1 -- measurements at year 10 and 15
Data(1).y = [ut_10(:,2)', ut_15(:,2)'];
Data(1).Name = 'Thickness measurements at year 10';
Data(1).MOMap = [ones(1, l_10+l_15); ones(1, l_10), ones(1,l_15)*2]; % Model Output Map
DiscrepancyPriorOpts.Name = 'Prior of sigma at year 10 and 15';
DiscrepancyPriorOpts.Marginals(1).Name = 'Sigma1_2';
DiscrepancyPriorOpts.Marginals(1).Type = 'Lognormal';
DiscrepancyPriorOpts.Marginals(1).Parameters = [-3.26, sqrt(2)*1.81]; % Product of two logn rvs: sigma * sigma
DiscrepancyPrior10 = uq_createInput(DiscrepancyPriorOpts);
DiscrepancyOpts(1).Type = 'Gaussian';
DiscrepancyOpts(1).Prior = DiscrepancyPrior10;
But in the output of the analysis I think there are some problems. Firstly, I would expect that the model outputs are 2. Secondly, I expect that the total number of observations should be l_10 + l_15 = 1418
. The analysis output is given below:
%----------------------- Inversion output -----------------------%
Number of calibrated model parameters: 2
Number of non-calibrated model parameters: 0
Number of calibrated discrepancy parameters: 1
%------------------- Data and Discrepancy
% Data-/Discrepancy group 1:
Number of independent observations: 1
Discrepancy:
Type: Gaussian
Discrepancy family: Scalar
Discrepancy parameters known: No
Associated outputs:
Model 1:
Output dimensions: 1
to
1
2
to
2
What am I doing wrong?