UQWorld

Problem with Bayesian linear regression

Hi all,
I encountered a problem in performing regression analysis. The model I am going to run has 9 inputs, but when I run the program, the functions that contain the first 5 inputs are executed, but the functions that include inputs 6 to 9 are not executed.
I have also defined the command |PMap| vector correctly.

I do not understand what is the reason for this problem?
Can you help me with this problem?

I attached the codes.

uq_Example_Inversion_09_test.m (4.8 KB)
uq_ArmourFailure.m (474 Bytes) uq_OvertoppingFailure.m (753 Bytes) uq_ToeFailure.m (484 Bytes)
AllData.zip (1.6 KB)

HI @Soheil_Radfar

I looked at your code and it seems like you indexed X wrongly. The point of PMap is to define which parameters are passed to the individual forward models. Only the relevant parameters are then passed to the models, looking now at uq_ToeFailure with PMap = [1 2 5 7 8] this means that you should change

V = X(:, 2) - (X(:, 5) ./ (2 + 6.2 .* (X(:, 8) .^ 2.7)) ./(X(:, 7) .^0.15)./X(:, 1));

to

V = X(:, 2) - (X(:, 3) ./ (2 + 6.2 .* (X(:, 5) .^ 2.7)) ./(X(:, 4) .^0.15)./X(:, 1));

and analogously for uq_OvertoppingFailure. Note that you could also just leave out the PMap definition in your main file. UQLab will then pass all parameters to the individual forward models, where they are correctly indexed.

2 Likes

Dear @paulremo,

Great!
Thank you and also, all UQLab team for your perfect software and perfect support.