Multiple zeta values are special numbers built by summing infinite series involving products of whole numbers raised to powers. They appear across mathematics and physics and carry deep structure, including relationships between seemingly different-looking sums. Multiple T-values are a related family of numbers introduced by Kaneko and Tsumura, defined by a similar summing process but with a restriction that alternates signs in a specific pattern. Like their zeta cousins, T-values obey many elegant identities, but their theory is less developed and many of those identities remain conjectural.
A major open problem in the field, posed by Kaneko and Zagier, asks whether two different ways of making multiple zeta values "finite" or "symmetric" always produce the same algebraic structure. Roughly speaking, one approach truncates the infinite sums and works modulo prime numbers, while the other averages over a symmetry group in a formal algebraic setting. The conjecture says these two procedures yield isomorphic spaces of numbers, meaning there is a perfect dictionary between them. This paper investigates whether the same phenomenon holds for multiple T-values, extending earlier work by Zhao who showed it works for a broader class called Euler sums.
The authors prove several new identities among finite and symmetric T-values, confirming conjectures made by one of the paper's own authors and uncovering entirely new families of relations. Their proofs draw on a toolkit assembled from recent advances: a theory of mixed zeta-like values developed by Xu and Zhao, identities among double zeta values due to Gangl, Kaneko, and Zagier, and summation formulas for double Euler sums. They also use generating functions tied to integral representations of T-values to transfer results between the symmetric and finite settings. By computing the dimensions of the relevant spaces of numbers, the paper provides strong evidence that the Kaneko-Zagier isomorphism conjecture extends cleanly to the world of multiple T-values.