Hello,
First of all, I am very grateful to uqlab for providing great convenience to my research.
Recently, I am using the PCE accelerated Bayesian inference module in uqlab to conduct research on my subject（Inversion of dam dynamic parameters), and I ran into some problems during this process.
Here is a description of my analysis:

I have four inputs, as follows：

My model comes from a finite element software (ANSYS), with six outputs, which are the first six measured natural frequencies of the dam. It is surrogated with a PCE metamodel (Eloo ~ 6.26e06 / 3.90e06 / 5.57e06 / 5.74e06 / 0.0014 / 0.0165 for six outputs).
I use the calculation results from ANSYS and the input parameters from LHS sampling to construct the PCE, then use the measured data for inversion.
The complete code is as follows：
The calculation results in uqlab are as follows：
From the calculation results, we can see that the variance is very huge, But I don’t know what caused this result.
In addition, I would like to ask whether the posterior margins calculated by uqlab can be used to obtain specific posterior estimates，because what I have obtained is the specific posterior estimated value based on the program I wrote, as follows:
Thank you very much for your guidance.
@olaf.klein @paulremo