Stochastic Krylov Dynamics: Revisiting Operator Growth in Open Quantum Systems

Arpan Bhattacharyya a, S. Shajidul Haque b,c , Jeff Murugan b,c , Mpho Tladi b and Hendrik J.R. Van Zyl b · 2026

Read Stochastic Krylov Dynamics: Revisiting Operator Growth in Open Quantum Systems by Arpan Bhattacharyya a, S. Shajidul Haque b,c , Jeff Murugan b,c , Mpho Tladi b and Hendrik J.R. Van Zyl b (2026) 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-2604-20619.txt.

Introduction

Bhattacharyya and four colleagues show that Krylov complexity keeps its shape when a closed quantum system couples to an environment, though altered. Using a Schwinger-Keldysh account of counting statistics of the Krylov position, they derive an action for operator growth under Lindblad dynamics. Even for minimal dephasing, phase-space dynamics stops being Hamiltonian: coupling makes diffusion in the variable conjugate to Krylov depth, turning set paths random. Dissipation acts as a key perturbation of the chaotic Krylov fixed point.

Catalog record

CREATORArpan Bhattacharyya a, S. Shajidul Haque b,c , Jeff Murugan b,c , Mpho Tladi b and Hendrik J.R. Van Zyl b
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE88552
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.20619

Browse the full library · About the Tower of Babel

Full text

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

Stochastic Krylov Dynamics: Revisiting Operator Growth in Open Quantum Systems Arpan Bhattacharyya a, S. Shajidul Haque b,c , Jeff Murugan b,c , Mpho Tladi b and Hendrik J.R. Van Zyl b Abstract In closed quantum systems, Krylov complexity admits a geometric description: operator growth is equivalent to Hamiltonian flow in an emergent phase space whose structure is fixed by the Lanczos coefficients. We show that this picture survives, albeit in a fundamentally altered form, once the system is coupled to an environment. Using a Schwinger–Keldysh formulation of the full counting statistics of the Krylov position, we derive an effective action for operator growth under Lindblad dynamics. Even for the minimal case of pure dephasing, the phase-space dynamics ceases to be Hamiltonian: environmental coupling generates diffusion in the variable conjugate to Krylov depth, converting deterministic trajectories into stochastic ones. The hyperbolic mechanism underlying exponential complexity growth is therefore broadened and, beyond a parametrically controlled scale, destroyed. This identifies dissipation as a relevant perturbation of the chaotic Krylov fixed point and reveals operator growth in open systems as a problem of stochastic dynamics in an emergent phase space. 1 Introduction Understanding how operators evolve and spread in quantum many-body systems has become a central theme in modern theoretical physics. This question lies at the heart of quantum chaos, thermalization, and information scrambling, and underpins phenomena ranging from the eigenstate thermalization hypothesis (ETH) to the fast scrambling conjecture and holographic duality 15 ; 35 ; 31 ; 33 ; 24 . Traditionally, operator growth has been characterized using out-of-time-order correlators (OTOCs), which diagnose the sensitivity of quantum dynamics to perturbations and provide a notion of a quantum Lyapunov exponent 20 ; 33 ; 24 . Complementary approaches based on spectral statistics and random matrix theory have further clarified the relationship between chaos, ergodicity, and universality in many-body systems 10 ; 19 . More recently, Krylov complexity has emerged as a powerful and conceptually distinct diagnostic of operator growth 28 ; 16 ; 4 ; 3 ; 29 ; 1 ; 30 . Rather than focusing on specific correlation functions, Krylov complexity tracks how an operator spreads in the space generated by repeated commutation with the Hamiltonian. This construction makes use of the computationally efficient Lanczos algorithm to measure operator growth and is closely tied to the structure of the Liouvillian. Given a seed operator O_{0} , the Krylov construction generates an orthonormal sequence \displaystyle\mathcal{K}=\text{span}\{O_{0},\mathcal{L}O_{0},\mathcal{L}^{2}O_{0},\dots\}\,, (1) where the Liouvillian superoperator is defined by \mathcal{L}O=[H,O] . Applying the Lanczos algorithm produces an orthonormal basis \{|O_{n}\rangle\} in which the Liouvillian takes a tridiagonal form \displaystyle\mathcal{L}|O_{n}\rangle=b_{n+1}|O_{n+1}\rangle+b_{n}|O_{n-1}\rangle\,, (2) with non-negative coefficients b_{n} , known as Lanczos coefficients. This representation is equivalent to a tight-binding Hamiltonian \displaystyle\widehat{H}_{\textrm{TB}}=\sum_{n}b_{n+1}\Big(|n\rangle\langle n+1|+|n+1\rangle\langle n|\Big)\,, (3) acting on a semi-infinite chain, often referred to as the Krylov chain. Krylov complexity is then defined as the expectation value of the position operator on this chain, \displaystyle\widehat{n}=\sum_{n}n|n\rangle\langle n|,\qquad K(t)=\langle\widehat{n}(t)\rangle, (4) and measures the average distance an operator has propagated in this Krylov space. This quantity captures operator growth in a coarse-grained but physically transparent way and has been shown to exhibit universal features across a wide range of systems, including exponential growth associated with chaotic dynamics and slower growth in integrable or constrained systems 28 ; 4 ; 30 1 1 1 For extension to the bosonic case readers are referred to 8 . . A key recent conceptual advance is that Krylov dynamics admits a natural reformulation in terms of a real-time Schwinger–Keldysh path integral. In this approach, Krylov complexity is treated as an in–in observable generated by the full counting statistics Z(\chi,t)=\langle e^{i\chi\hat{n}(t)}\rangle , which plays the role of a dynamical generating functional. As shown in recent work 26 , this formulation maps operator growth onto classical Hamiltonian dynamics in an emergent phase space with coordinates (n,p) . The resulting effective Hamiltonian, H_{\rm eff}(n,p)=2b(n)\cos p , encodes the full information of the Lanczos coefficients and provides a geometric interpretation of operator growth. In particular, when b(n)\sim\alpha n , the phase-space flow develops hyperbolic fixed manifolds, and exponential growth of Krylov complexity emerges as a classical instability, K(t)\sim e^{2\alpha t} . This perspective not only unifies various observations about Krylov complexity but also enables a systematic classification of operator growth in terms of universality classes determined by the asymptotic scaling of b(n) . Despite this progress though, the existing framework is restricted to closed quantum systems undergoing unitary evolution. In contrast, many experimentally relevant systems are inherently open; they interact with external environments, experience decoherence, and are subject to measurement backaction. Open quantum systems are typically described by Lindblad dynamics 21 ; 18 ; 11 , which introduces dissipation and noise at the level of the density matrix. These effects are known to qualitatively alter dynamical behavior, leading to phenomena such as decoherence, localization, and measurement-induced phase transitions 34 ; 14 2 2 2 Interested readers are referred to 13 for a proposal of detecting measurement-induced phase transitions by observing dynamics of certain observable in Krylov space. . The extension of Krylov complexity to such settings raises several conceptual challenges. The first, and most pressing, of which is that the Liouvillian governing operator evolution is no longer anti-Hermitian, and the Lanczos construction becomes non-unique or ill-conditioned. More fundamentally, it is not clear how the geometric phase-space picture of operator growth is modified in the presence of dissipation. The purpose of this work is to address these questions by extending the Schwinger–Keldysh formulation of Krylov complexity in 26 to open quantum systems. Working directly with the generating functional Z(\chi,t) , we derive a contour formulation of operator growth under Lindblad dynamics and identify the corresponding effective action in Krylov phase space. We show that environmental coupling generically converts the emergent Hamiltonian flow of the closed system into a stochastic dynamical system, with noise and dissipation encoded in the structure of the Keldysh action. This leads to a qualitatively new picture of operator growth in which hyperbolic instability competes with environmental fluctuations, providing a unified framework for studying complexity, scrambling, and dynamical phase transitions in open quantum systems. 2 Linblad Dynamics and Open Systems Open quantum systems are generically described by completely positive, trace-preserving dynamics. In the Markovian limit, this evolution is governed by the Lindblad master equation 21 ; 18 ; 11 \dot{\rho}=-i[H,\rho]+\sum_{\mu}\left(L_{\mu}\rho L_{\mu}^{\dagger}-\frac{1}{2}\left\{L_{\mu}^{\dagger}L_{\mu},\rho\right\}\right), (5) where H is the system Hamiltonian and the operators L_{\mu} encode the coupling to the environment. These jump operators describe processes such as dephasing, dissipation, particle loss, or continuous monitoring, and give rise to irreversible dynamics even in otherwise isolated systems. While Eq. ( 5 ) is naturally formulated in the Schrödinger picture, operator growth is most naturally studied in the Heisenberg picture. In this case, observables evolve according to the adjoint Liouvillian \frac{d}{dt}O(t)=\mathcal{L}_{\rm Lind}(O), (6) with \mathcal{L}_{\rm Lind}(O)=i[H,O]+\sum_{\mu}\left(L_{\mu}^{\dagger}OL_{\mu}+\frac{1}{2}\left\{L_{\mu}^{\dagger}L_{\mu},O\right\}\right)\,. (7) This equation generalizes the familiar Heisenberg evolution of closed systems by incorporating both coherent dynamics and dissipative processes. 2.1 Structure of the Lindbladian and loss of unitarity A crucial structural difference between closed and open systems lies in the properties of the generator of dynamics. For closed systems, the Liouvillian \mathcal{L}=[H,\cdot] is anti-Hermitian with respect to the Hilbert–Schmidt inner product \langle A|B\rangle=\frac{1}{\mathcal{N}}\mathrm{Tr}(A^{\dagger}B)\,, (8) and therefore generates unitary evolution in operator space. This

Read the complete text