Bayesian Inversion Analysis

In the Beam Example the MEASUREMENT DATA ‘Y’ consist of 5 measurements with only a single value data. However , How about a single measurement and an array of data. How can it be modeled ? Please see the attachment .Thanks

Let me be more clear , I’m using a forward model which not just give single value , rather an array of data [a waveform]

Good morning an welcome to UQWorld!

What you are trying to do should be pretty straightforward. The measurement matrix should be supplied as a N\times N_{out} matrix where N is the number of measurements and N_{out} is the number of model outputs. In your case of N=1 measurement you should therefore provide the data as a row vector.

I also suggest you have a look at Section 2.3 of the Bayesian inversion manual where a similar case is described.

Thanks for the reply , I tried the row vector as N =1 [Measurement Data], but I see the error as attached for refeneceError


This happens because you tried to let UQLab infer the bounds on the prior distribution for the discrepancy variance, while at least one of your supplied data values is 0. We know about this issue and will add a more descriptive error message in the upcoming next version of UQLab.

To get around this problem, you will have to manually specify a discrepancy variance for your problem. I suggest you consult Section 2.2.6 of the Bayesian inversion manual and get back to me if you run into additional problems.