© 2026 Aqora Quantum S.A.S.

Terms Privacy Legal Notice

arXiv:2608.07207

Existence of Kraus decomposition in infinite dimension via strongly-convergent direct process tomography

Paul E. Lammert

quant-phmath-phSubmitted Aug 7, 2026• Announced Aug 10, 2026• 0 citations• Revised Oct 6, 2026 (v2)

v1 · Aug 7, 2026 v2 · Oct 6, 2026

Abstract

An algorithm is presented for Kraus decomposition of a completely positive operator over separable (countably-infinite-dimensional) Hilbert spaces, together with an elementary proof that the generated sum convergences in strong-operator topology. This improves on the standard, nonconstructive, proof by fusing the abstract problem with practical process tomography. Kraus operators are generated one-by-one, each having one more guaranteed zero matrix entry than the previous one. In this way, the stream of outputs of the algorithm provides a coherent family of Kraus decompositions of restrictions of the target CP map to ever-larger subspaces.

Comments: The equality in (7c) should have been an inclusion. This has been fixed, but it had no serious consequences. Proof of Lemma 1 has been rewritten, and explanation in Section IIIC expanded

Related papers

Choi’s proof as a recipe for quantum process tomography

Debbie Leung

Quantum process tomography is a procedure by which an unknown quantum operation can be fully experimentally characterized. We reinterpret Choi’s proof [Linear Algebr. Appl. 10, 285 (1975)] of the fact that any completely positive linear map has a Kraus representation as a method for quantum process tomography. The analysis for obtaining the Kraus operators is extremely simple. We discuss the systems in which this tomography method is particularly suitable.

arXiv:quant-ph/0201119•Journal of Mathematical Physics•Jan 17, 2003•100 citations

A Simple Proof of the Jamiołkowski Criterion for Complete Positivity of Linear Maps

David Cienfuegos Salgado, J. L. Sánchez‐Gómez, Miguel Ferrero

We give a simple direct proof of the Jamiołkowski criterion to check whether a linear map between matrix algebras is completely positive or not. This proof is more accessible for physicists than other ones found in the literature and provides a systematic method to give any set of Kraus matrices of the Kraus decomposition.

Open Systems & Information Dynamics•Mar 1, 2005•12 citations

From the Choi formalism in infinite dimensions to unique decompositions of generators of completely positive dynamical semigroups

Frederik vom Ende

Given any separable complex Hilbert space, any trace-class operator [Formula: see text] which does not have purely imaginary trace, and any generator [Formula: see text] of a norm-continuous one-parameter semigroup of completely positive maps we prove that there exists a unique bounded operator [Formula: see text] and a unique completely positive map [Formula: see text] such that (i) [Formula: see text], (ii) the superoperator [Formula: see text] is trace class and has vanishing trace, and (iii) [Formula: see text] is a real number. Central to our proof is a modified version of the Choi formalism which relates completely positive maps to positive semi-definite operators. We characterize when this correspondence is injective and surjective, respectively, which in turn explains why the proof idea of our main result cannot extend to non-separable Hilbert spaces. In particular, we find examples of positive semi-definite operators which have empty pre-image under the Choi formalism as soon as the underlying Hilbert space is infinite-dimensional.

arXiv:2401.14344•Infinite Dimensional Analysis Quantum Probability and Related Topics•Jul 19, 2024•0 citations

A simple proof of the Jamiolkowski criterion for complete positivity of linear maps

Salgado, D., J. L. Sánchez‐Gómez, Miguel Ferrero

We give a simple direct proof of the Jamiolkowski criterion to check whether a linear map between matrix algebras is completely positive or not. This proof is more accesible for physicists than others found in the literature and provides a systematic method to give any set of Kraus matrices of its Kraus decomposition.

arXiv:math-ph/0406010•Jun 4, 2004•0 citations

Gradient-Descent Quantum Process Tomography by Learning Kraus Operators

Shahnawaz Ahmed, Fernando Quijandría, Anton Frisk Kockum

We perform quantum process tomography (QPT) for both discrete- and continuous-variable quantum systems by learning a process representation using Kraus operators. The Kraus form ensures that the reconstructed process is completely positive. To make the process trace preserving, we use a constrained gradient-descent (GD) approach on the so-called Stiefel manifold during optimization to obtain the Kraus operators. Our ansatz uses a few Kraus operators to avoid direct estimation of large process matrices, e.g., the Choi matrix, for low-rank quantum processes. The GD-QPT matches the performance of both compressed-sensing (CS) and projected least-squares (PLS) QPT in benchmarks with two-qubit random processes, but shines by combining the best features of these two methods. Similar to CS (but unlike PLS), GD-QPT can reconstruct a process from just a small number of random measurements, and similar to PLS (but unlike CS) it also works for larger system sizes, up to at least five qubits. We envisage that the data-driven approach of GD-QPT can become a practical tool that greatly reduces the cost and computational effort for QPT in intermediate-scale quantum systems.

arXiv:2208.00812•Physical Review Letters•Apr 14, 2023•25 citations

From Kraus Operators to the Stinespring Form of Quantum Maps: An Alternative Construction for Infinite Dimensions

Frederik vom Ende

We present an alternative (constructive) proof of the statement that for every completely positive, trace-preserving map $Φ$ there exists an auxiliary Hilbert space $\mathcal K$ in a pure state $|ψ\rangle\langleψ|$ as well as a unitary operator $U$ on system plus environment such that $Φ$ equals $\operatorname{tr}_{\mathcal K}(U((\cdot)\otimes|ψ\rangle\langleψ|)U^*)$. The main tool of our proof is Sz.-Nagy's dilation theorem applied to isometries defined on a subspace. In our construction, the environment consists of a system of dimension "Kraus rank of $Φ$" together with a qubit, the latter only acting as a catalyst. In contrast, the original proof of Hellwig & Kraus given in the 70s yields an auxiliary system of dimension "Kraus rank plus one". We conclude by providing an example which illustrates how the constructions differ from each other.

arXiv:2301.05488•Jan 13, 2023•0 citations

An algorithm is presented for Kraus decomposition of a completely positive operator over separable (countably-infinite-dimensional) Hilbert spaces, together with an elementary proof that the generated sum convergences in strong-operator topology. This improves on the standard, nonconstructive, proof by fusing the abstract problem with practical process tomography. Kraus operators are generated one-by-one, each having one more guaranteed zero matrix entry than the previous one. In this way, the stream of outputs of the algorithm provides a coherent family of Kraus decompositions of restrictions of the target CP map to ever-larger subspaces.

Given any separable complex Hilbert space, any trace-class operator [Formula: see text] which does not have purely imaginary trace, and any generator [Formula: see text] of a norm-continuous one-parameter semigroup of completely positive maps we prove that there exists a unique bounded operator [Formula: see text] and a unique completely positive map [Formula: see text] such that (i) [Formula: see text], (ii) the superoperator [Formula: see text] is trace class and has vanishing trace, and (iii) [Formula: see text] is a real number. Central to our proof is a modified version of the Choi formalism which relates completely positive maps to positive semi-definite operators. We characterize when this correspondence is injective and surjective, respectively, which in turn explains why the proof idea of our main result cannot extend to non-separable Hilbert spaces. In particular, we find examples of positive semi-definite operators which have empty pre-image under the Choi formalism as soon as the underlying Hilbert space is infinite-dimensional.

We perform quantum process tomography (QPT) for both discrete- and continuous-variable quantum systems by learning a process representation using Kraus operators. The Kraus form ensures that the reconstructed process is completely positive. To make the process trace preserving, we use a constrained gradient-descent (GD) approach on the so-called Stiefel manifold during optimization to obtain the Kraus operators. Our ansatz uses a few Kraus operators to avoid direct estimation of large process matrices, e.g., the Choi matrix, for low-rank quantum processes. The GD-QPT matches the performance of both compressed-sensing (CS) and projected least-squares (PLS) QPT in benchmarks with two-qubit random processes, but shines by combining the best features of these two methods. Similar to CS (but unlike PLS), GD-QPT can reconstruct a process from just a small number of random measurements, and similar to PLS (but unlike CS) it also works for larger system sizes, up to at least five qubits. We envisage that the data-driven approach of GD-QPT can become a practical tool that greatly reduces the cost and computational effort for QPT in intermediate-scale quantum systems.

We present an alternative (constructive) proof of the statement that for every completely positive, trace-preserving map

Φ

there exists an auxiliary Hilbert space

\mathcal K

in a pure state

|ψ\rangle\langleψ|

as well as a unitary operator

U

on system plus environment such that

Φ

equals

\operatorname{tr}_{\mathcal K}(U((\cdot)\otimes|ψ\rangle\langleψ|)U^*)

. The main tool of our proof is Sz.-Nagy's dilation theorem applied to isometries defined on a subspace. In our construction, the environment consists of a system of dimension "Kraus rank of

Φ

" together with a qubit, the latter only acting as a catalyst. In contrast, the original proof of Hellwig & Kraus given in the 70s yields an auxiliary system of dimension "Kraus rank plus one". We conclude by providing an example which illustrates how the constructions differ from each other.