Quantum Entropy Prover

Shao-Lun Huang, Tobias Rippchen, Mario Berta · 2025

Read Quantum Entropy Prover by Shao-Lun Huang, Tobias Rippchen, Mario Berta (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-16025.txt.

Introduction

Shao-Lun Huang, Tobias Rippchen, and Mario Berta wrote this 2025 paper on proving facts about quantum entropy. Entropy bounds set the hard limits of what quantum systems can store and send, but proofs grow painful as the number of parties rises. For classic systems, a linear programming frame already helps; the authors build the quantum twin of that frame.

Their method rests on two core entropy rules for quantum states. It also handles bounds with extra limits, the kind that once revealed non-standard classic bounds. The team ships the work as a Python package called qITIP and shows its use on several worked examples for researchers in quantum information.

Catalog record

CREATORShao-Lun Huang, Tobias Rippchen, Mario Berta
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE25 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-16025

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Quantum Entropy Prover Shao-Lun Huang, Tobias Rippchen, Mario Berta Abstract Information inequalities govern the ultimate limitations in information theory and as such play an pivotal role in characterizing what values the entropy of multipartite states can take. Proving an information inequality, however, quickly becomes arduous when the number of involved parties increases. For classical systems, [Yeung, IEEE Trans. Inf. Theory (1997)] proposed a framework to prove Shannon-type inequalities via linear programming. Here, we derive an analogous framework for quantum systems, based on the strong sub-additivity and weak monotonicity inequalities for the von-Neumann entropy. Importantly, this also allows us to handle constrained inequalities, which — in the classical case — served as a crucial tool in proving the existence of non-standard, so-called non-Shannon-type inequalities [Zhang & Yeung, IEEE Trans. Inf. Theory (1998)]. Our main contribution is the Python package qITIP, for which we present the theory and demonstrate its capabilities with several illustrative examples. I Introduction Information inequalities are inequalities that can be expressed as linear combinations of the entropy of multipartite states. Since such inequalities govern what is ultimately feasible in information theory, they are also regarded as the laws of information theory [ 1 ] and play a vital role in characterizing the entropy and other information measures (e.g. [ 2 , 3 , 4 , 5 , 6 ] ). However, a challenge is that the proof of an information inequality is usually not mathematically obvious. In particular, in quantum information theory, the only inequalities currently known to be satisfied by all quantum states can be derived from the strong sub-additivity (SSA) and weak monotonicity (WM) of the quantum entropy [ 6 ] . In the literature, these are called von-Neumann-type inequalities . A quantum information inequality that is not von-Neumann-type, but nevertheless satisfied by all quantum states would accordingly be called a non-von-Neumann-type inequality . Currently, the existence of non-von-Neumann-type inequalities remains an open question [ 5 , 7 , 6 ] . In classical information theory, a linear programming approach is used to prove the classical counterparts to von-Neumann-type inequalities (cf. [ 8 , 9 , 1 ] ). Namely, the so-called Shannon-type inequalities can be derived by SSA and strong monotonicity (SM) of the classical entropy function. SM is equivalent to the non-negativity of the classical conditional entropy and it is this fact that is one of the biggest differences between classical and quantum information theory. The quantum conditional entropy can be negative; hence, a different set of basic inequalities is required here. In [ 3 ] , Pippenger found the minimal set of inequalities that generate all instances of SSA and WM. Inspired by the classical prover, our work combines Pippenger’s minimal set with the linear programming technique by Yeung. Our main contribution is a Python package based on our framework, called qITIP [ 10 ] , that automates the proof of von-Neumann-type inequalities. Moreover, the existence of non-Shannon-type inequalities was answered in the affirmative in [ 2 ] after a constrained one was found in [ 11 ] . Our framework then also includes the possibility to handle additional constraints to facilitate similar research into non-von-Neumann-type inequalities. II Quantum Entropic Cone A quantum state \hat{\rho} is a unit-trace positive semi-definite operator on a finite-dimensional Hilbert space \mathcal{H} . The entropy of \hat{\rho} , the so-called quantum entropy or von-Neumann entropy, is defined as S(\hat{\rho})\equiv-\Tr[\hat{\rho}\log{\hat{\rho}}] (1) with the trace operation \Tr[\cdot] . For an n -party quantum system, the Hilbert space factors into the tensor product \mathcal{H}=\bigotimes_{i=1}^{n}\mathcal{H}_{i} of the individual Hilbert spaces \mathcal{H}_{i} . In the following, we will often denote the parties by capital letters A,B,C,\text{etc} . The state of an n -party system will be represented as \hat{\rho}_{N_{n}} , where N_{n} abbreviates the set \{1,2,\dots,n\} . If we focus on the subsystems I\subseteq N_{n} , the marginal quantum state \hat{\rho}_{I} is given by \hat{\rho}_{I}\equiv\Tr_{N_{n}\setminus I}[\hat{\rho}_{N_{n}}], (2) where \Tr_{N_{n}\setminus I}[\cdot] means partially tracing out the parts not included in I . Therefore, given a n -party quantum state \hat{\rho}_{N_{n}} , there are k=2^{n}-1 marginal states associated with it excluding the trivial null state \hat{\rho}_{\emptyset} . The k -dimensional vector whose entries are given by these k marginal entropies is called the entropic vector . The associated vector space \mathbb{R}^{k} is called the entropic space . Consider, for example, a bipartite quantum system whose quantum state is \hat{\rho}_{AB} . The entropic vector of \hat{\rho}_{AB} is given by \mathbf{s}(\hat{\rho}_{AB})=\langle{S(\hat{\rho}_{A}),S(\hat{\rho}_{B}),S(\hat{\rho}_{AB})}\rangle. (3) We use \mathbf{s}(\hat{\rho}_{AB}) to indicate the entropic vector of the state \hat{\rho}_{AB} . For ease of notation, we will drop the symbol \hat{\rho} for the marginal entropies, leaving only the indices of the subsystems, e.g., S(A)\equiv S(\hat{\rho}_{A}) . In the entropic space, let \Gamma_{n}^{*} denote the set of all entropic vectors associated with a valid n -party quantum state. The closure of \Gamma_{n}^{*} , denoted by \overline{\Gamma_{n}^{*}} , was proven by Pippenger to form a convex cone called the entropic cone in [ 3 ] . He then initiated a program to characterize the structure of this entropic cone using information inequalities. II-A Von-Neumann Cone First, let us formally introduce the two most important information inequalities in quantum information theory – the strong sub-additivity (SSA), S(I)+S(J)\geq S(I\cup J)+S(I\cap J), (4) and the weak monotonicity (WM), S(I)+S(J)\geq S(I\setminus J)+S(J\setminus I), (5) where I,J\subseteq N_{n}\setminus\emptyset . The most familiar instances of SSA and WM are probably those of a tripartite system (see [ 12 ] for the original proof) given by S(A,B)+S(B,C)\geq S(A,B,C)+S(B) (6) and S(A,B)+S(B,C)\geq S(A)+S(C). (7) Note that equation ( 6 ) can be re-expressed as the positivity of the quantum conditional mutual information defined as I(A;C|B)\equiv S(A,B)+S(B,C)-S(A,B,C)+S(B). (8) Note that although WM can be derived from SSA, the proof requires an additional party to which we have no access in our setting. We refer to all instances of SSA and WM as basic inequalities . As we can see, the basic inequalities are given in terms of marginal entropies; therefore, we can write all of them compactly in matrix form as \mathbf{\tilde{G}}\mathbf{s}\geq\mathbf{0} , where \mathbf{0} is the null vector and \mathbf{\tilde{G}} is a matrix derived from the coefficients in front of the marginal entropies. In entropic space, we then define another set \Gamma_{n}\subset\mathbb{R}^{k} as \Gamma_{n}=\left\{\mathbf{s}\in\mathbb{R}^{k}\colon\mathbf{\tilde{G}}\mathbf{s}\geq\mathbf{0}\right\}, (9) which we call the von-Neumann cone . Since all quantum states satisfy SSA and WM, we have \Gamma_{n}^{*}\subset\overline{\Gamma_{n}^{*}}\subset\Gamma_{n}. (10) An open question is whether \Gamma_{n}=\overline{\Gamma_{n}^{*}} holds for n\geq 4 [ 3 ] . If this were not the case, it implies the existence of non-von-Neumann-type inequalities. A visualization of the argument is presented in Fig. 1 . (a) \overline{\Gamma_{n}^{*}}\subsetneq\Gamma_{n} (b) \overline{\Gamma_{n}^{*}}=\Gamma_{n} Fig. 1: Implications of (non)-existence of non-von-Neumann-type inequalities. There exists an information inequality \mathbf{b^{\top}\mathbf{s}}\geq 0 such that every element in \overline{\Gamma_{n}^{*}} satisfies the inequality. However, we see there are elements within \Gamma_{n} that do not. Since \Gamma_{n}=\overline{\Gamma_{n}^{*}} , if every element in \overline{\Gamma_{n}^{*}} satisfies \mathbf{b^{\top}\mathbf{s}}\geq 0 , so does every element in \Gamma_{n} . II-B Pippenger’s Minimal Set of Inequalities Note that some basic inequalities can be derived from other basic inequalities. For example, the inequality I(A;B,C)\equiv S(A)+S(B,C)-S(A,B,C)\geq 0 (11) can be derived from I(A;B)\geq 0 and I(A;C|B)\geq 0 combined with the chain rule I(A;B,C)=I(A;B)+I(A;C|B). (12) Pippenger [ 3 ] showed that the basic inequalities can be

Read the complete text