tfp.substrates.numpy.vi.triangular

The Triangular Csiszar-function in log-space.

tfp.substrates.numpy.vi.triangular(
    logu, name=None
)

A Csiszar-function is a member of,

F = { f:R_+ to R : f convex }.

The Triangular Csiszar-function is:

f(u) = (u - 1)**2 / (1 + u)

This Csiszar-function induces a symmetric f-Divergence, i.e., D_f[p, q] = D_f[q, p].

logu float-like Tensor representing log(u) from above.
name Python str name prefixed to Ops created by this function.

triangular_of_u float-like Tensor of the Csiszar-function evaluated at u = exp(logu).