cuqi.distribution.Lognormal.enable_FD# Lognormal.enable_FD(epsilon=1e-08)# Enable finite difference approximation for logd gradient. Note that if enabled, the FD approximation will be used even if the _gradient method is implemented.