Perron-Frobenius matrix completion result is wrong

[utotch]

https://www.cvxpy.org/examples/dgp/pf_matrix_completion.html

This may be because Convex.jl lambdamax only appicable for Symmetric matrix.

lambda_max in Convex.jl is restricted Symmetric Matrix (since impl minimize(t, t*Matrix(1.0I, n, n) - A ⪰ 0))
https://github.com/JuliaOpt/Convex.jl/blob/master/src/atoms/sdp_cone/lambda_min_max.jl

but pf_eigenvalue can take element-wise positive matrix.
https://github.com/cvxgrp/cvxpy/blob/master/cvxpy/atoms/pf_eigenvalue.py

pf_eigenvalue: lambda_max(exp(X)) (X is positive element, square) is a NEW convex function that is not listed in applicable operations.