{"id":"arxiv/2611.03859","url":"https://arxiv.org/abs/2611.03859","tags":["math.NA","quant-ph"],"score":0,"title":"Factorized approximate inverse preconditioners in the tensor train format","domain":"math.NA","source":"arxiv","authors":["Pau Torrente Badia","Daniel Pérez García","Artur Garcia-Saez","Stefano Carignano"],"reviews":0,"subject":"math.NA","summary":"We present a preconditioning pipeline that accelerates Tensor Train (TT) linear solvers for symmetric positive-definite problems. Extending the Factorized Sparse Approximate Inverse framework to tensor networks, we construct a preconditioner as the product of a TT operator and its transpose. This formulation is positive semidefinite for any bond dimension and truncation, and definite whenever the factor is non-singular, satisfying the mathematical requirements of the Conjugate Gradient method. We detail the construction of this preconditioner and its integration into Density Matrix…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-11-02","reactions":{}}