TOPICS
Search

Trace Inequality


A trace inequality compares expressions formed from traces of matrices or their matrix powers. For a Hermitian matrix A with eigenvalues lambda_i, the identity Tr(A^j)=sum_(i)lambda_i^j converts polynomial trace inequalities into inequalities in power sums.

Debus and Schabert (2026) proved that deciding whether a polynomial with rational number coefficients in Tr(A^2), Tr(A^4), ..., Tr(A^(46)) is nonnegative for every Hermitian matrix of every finite size is undecidable. Undecidability persists under homogeneity with respect to scaling A|->tA, giving Tr(A^(2j)) weight 2j. For positive semidefinite matrices, powers 1 through 23 already suffice for undecidability. These statements use ordinary trace, not trace divided by the matrix size. The matrix size is unbounded.


See also

Hermitian Matrix, Matrix Trace, Symmetric Function

Explore with Wolfram|Alpha

References

Debus, S. and Schabert, R. "Hilbert's 17th Problem for Symmetric Functions: Undecidability and a Counterexample." 12 Nov 2026. https://arxiv.org/abs/2611.16690v1.

Cite this as:

Weisstein, Eric W. "Trace Inequality." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/TraceInequality.html

Subject classifications