Skip to content
New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

Definition of psi0 in vardtc #16

Open
Hebbalali opened this issue Jul 30, 2018 · 0 comments
Open

Definition of psi0 in vardtc #16

Hebbalali opened this issue Jul 30, 2018 · 0 comments

Comments

@Hebbalali
Copy link

Hello,
Firstly i would like to thank you for sharing this implementation online.
I have one question about the definition of the stastistics in vardtc for uncertain inputs.
In the code it's written:

        if uncertain_inputs:
            psi0 = kern.psi0(Z, X)
            psi1 = kern.psi1(Z, X)*beta
            psi2 = kern.psi2(Z, X)*beta

Then in the computation of the likelihood:

logL = -(output_dim*(num_data*log_2_pi+logL_R+psi0-np.trace(LmInvPsi2LmInvT))+YRY- bbt)/2.-output_dim*logdet_L/2.
My question is: shouldn't it be psi0= kern.psi0(Z, X)*beta or in the logL be psi0*beta ?? because in the logL psi0 should be scaled with beta

# for free to join this conversation on GitHub. Already have an account? # to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant