MaskedPrior.create_variable#
- MaskedPrior.create_variable(name)[source]#
Create a deterministic variable with full dims using the active subset.
Creates an underlying variable over the active entries only and expands it back to the full masked shape, filling inactive entries with zeros.
- Parameters:
- name
str
Base name for the created variables.
- name
- Returns:
pt.TensorVariable
Deterministic variable with the original dims, zeros on inactive entries.