Comment on 'Multiparty quantum mutual information: An alternative definition'

We show that, contrary to the claim by Kumar [Phys. Rev. A 96, 012332 (2017)], the quantum dual total correlation of an $n$-partite quantum state cannot be represented as the quantum relative entropy between $n-1$ copies of the quantum state and the product of $n$ different reduced quantum states for $n \geq 3$. Specifically, we argue that the latter fails to yield a finite value for generalized $n$-partite Greenberger-Horne-Zeilinger states.
Comment: 2 pages, close to published version