The Exact Uncertainty Relation and Geometric Speed Limits in Krylov Space
Mohsen Alishahiha, Souvik Banerjee · 2025
Read The Exact Uncertainty Relation and Geometric Speed Limits in Krylov Space by Mohsen Alishahiha, Souvik Banerjee (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-2512-20359.txt.
Introduction
Mohsen Alishahiha and Souvik Banerjee give quantum speed limits a clean shape-based form. Hall's exact uncertainty relation is an equality, stronger than the usual inequality, but its meaning for operator growth was obscure. The authors rewrite it in the Krylov basis that the Liouvillian generates for an evolving operator.
In that frame, the operator's amplitude vector moves on a unit sphere at constant speed. Remarkably, the speed is set only by the first Lanczos coefficient, the opening entry of the Krylov chain. Higher coefficients drop out entirely. The resulting linear bound on such evolution holds for any Hamiltonian, integrable or chaotic. Operator growth research gains a single built-in speed scale for quantum dynamics.
Catalog record
| CREATOR | Mohsen Alishahiha, Souvik Banerjee |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-20359
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The Exact Uncertainty Relation and Geometric Speed Limits in Krylov Space Mohsen Alishahiha, Souvik Banerjee Abstract We show that Hall’s exact uncertainty relation acquires a simple geometric form in the Krylov basis generated by the Liouvillian. In this canonical operator frame, the uncertainty equality implies that the operator amplitude vector evolves on the unit Krylov sphere with constant speed fixed solely by the first Lanczos coefficient. This yields an exact linear bound on geometric operator evolution, independent of higher Lanczos coefficients and valid for arbitrary Hamiltonians, integrable or chaotic. Our results provide the first unified geometric interpretation of exact quantum speed limits and operator growth, identifying the first Lanczos coefficient as the intrinsic speed scale of quantum dynamics. This work is dedicated to Sudipta Mukherji on the occasion of his 60th birthday. I Introduction The uncertainty principle is a defining feature of quantum mechanics, limiting the precision with which certain pairs of observables can be simultaneously measured. Beyond the canonical position–momentum relation, energy–time uncertainty relations Mandelstam and Tamm 1945 ; Margolus and Levitin 1998 imply fundamental limits on how rapidly a quantum state may evolve, known as quantum speed limits (QSLs). Such limits play a central role in quantum control, quantum information processing, and quantum computation. Most uncertainty relations take the form of inequalities and therefore yield lower bounds on the time required for a given evolution. In contrast, Hall derived a state-dependent exact uncertainty relation Hall 2001 for two observables A and B in a pure state \rho=|\psi\rangle\langle\psi| , \delta_{B}A\,\Delta B_{nc}=\frac{1}{2}, (1) where \Delta B_{nc} denotes the nonclassical component of B , and \delta_{B}A is a Fisher-information-like functional of A . Although explicitly state dependent, this relation yields an equality-type QSL once the nonclassical generator is identified in a suitable basis. Indeed, Ref. Pati et al. 2023 used Hall’s relation to construct an exact QSL for quantum dynamics. In this work, we show that operator dynamics admits an exact geometric formulation in Krylov space, in which quantum speed limits arise as kinematic identities. Specifically, we demonstrate that Hall’s exact uncertainty relation acquires a particularly simple and universal form in the Krylov basis, where the nonclassical part of the Liouvillian coincides with the full generator. As an immediate consequence, Lanczos representation of the Liouvillian Lanczos 1950 ; Viswanath and Müller 2008 reveals operator evolution on the unit Krylov sphere with a constant speed fixed solely by the first Lanczos coefficient. This could be identified as a geometric Lieb–Robinson–type bound intrinsic to Krylov dynamics which remains exact even when the Lanczos coefficients grow without bound, as in typical chaotic systems. In our work, we demonstrate this by introducing the motion of amplitude fronts in Krylov index space which is distinct from the geometric motion of the operator state on the Krylov sphere. We show that even when the operator front accelerate rapidly in Krylov index space, the actual geometric motion of the operator state remains strictly bounded, thereby resolving the apparent tension between fast Krylov growth and finite causal structure. Recent work has shown that Krylov space provides a powerful framework for characterizing operator growth and information spreading in many-body quantum systems Roberts et al. 2018 ; Barbón et al. 2019 . Our results place these developments in a concrete geometric setting and identify a basis-independent speed law that governs the operator dynamics. The novelty of this speed law lies in the fact that its universality remains valid across the choice of a Hamiltonian, be it chaotic or integrable. It is interesting to note that while the higher Lanczos coefficients do not affect the speed, they do govern the geometry of the trajectory—its curvature and departure from geodesic motion, leading to universal constraints such as bounds on return amplitudes. This paper is organized as follows. Section 1 reviews the Krylov construction, derives the constant-speed identity, and shows its equivalence to Hall’s relation in Krylov space. Section 2 develops the geometric picture: curvature and torsion, geodesics, the Krylov light cone, tail bounds, and quadratic invariants. Section 3 contains our conclusions. II Krylov Bases as Optimal Frames and the Exact Speed Law In a closed quantum system, understanding how an initial operator spreads under unitary time evolution is a central problem, with implications ranging from thermalization and chaos to complexity theory and information scrambling. While the full operator algebra {\cal B}({\cal H}) of a d -dimensional Hilbert space has dimension d^{2} , physical dynamics typically explores only a much smaller subspace generated by repeated commutators of the operator with the Hamiltonian. This motivates working within Krylov space, an optimally adapted basis which renders the Liouvillian tridiagonal and makes operator growth maximally transparent. In this section, we review the operator Krylov construction 1 1 1 In what follows we will consider operator growth, though it can be generalized to state spreading Balasubramanian et al. 2022 ; Alishahiha and Banerjee 2022 ; Caputa et al. 2024 ; Alishahiha 2023 (See Alishahiha and Banerjee ). and show that Krylov space is precisely the basis in which the exact uncertainty relation of Hall is saturated. This directly leads to an exact speed law, showing that the “velocity” of operator evolution in Krylov space is time-independent and given by the first Lanczos coefficient b_{1} . Consider an operator {\cal O}(0) evolving in the Heisenberg picture, {\cal O}(t)=e^{-iHt}\,{\cal O}(0)\,e^{iHt}. (2) Although {\cal O}(t) is an element of the full operator algebra, its time evolution is restricted to the Krylov subspace {\cal B}_{\mathcal{O}}(H)=\{{\cal O},[H,{\cal O}],[H,[H,{\cal O}]],\dots\}, (3) whose dimension is typically much smaller than d^{2} . Introducing the Liouvillian superoperator {\cal L}{\cal O}=[H,{\cal O}] , the Heisenberg evolution can be compactly written as {\cal O}(t)=e^{-it{\cal L}}{\cal O}(0) . which solves the operator Schrödinger equation. To construct a convenient basis for {\cal B}_{\mathcal{O}}(H) , we introduce the operator–state mapping |{\cal O}) , with inner product (A|B)=\mathrm{Tr}(A^{\dagger}B) , and apply Gram–Schmidt orthogonalization to the non-orthogonal set \{{\cal L}^{n}{\cal O}\} . The resulting orthonormal vectors \{|K_{n}\rangle\}_{n=0}^{D_{\mathcal{O}}-1} with D_{\mathcal{O}}\leq d^{2}-d+1 , satisfy the Lanczos recursion Viswanath and Müller 2008 |K_{0}\rangle=\frac{|\mathcal{O}(0))}{||\mathcal{O}(0))|},\quad{\cal L}|K_{n}\rangle=b_{n+1}|K_{n+1}\rangle+b_{n}|K_{n-1}\rangle, (4) which is tridiagonal and minimal in bandwidth. This bandwidth-minimization characterizes Krylov space as an optimal basis for describing operator growth. Here b_{0}=0 and b_{n\geq 1}\geq 0 are the Lanczos coefficients that characterize the rate of operator spreading. Let us expand the time-evolved operator in this basis, |{\cal O}(t))=\sum_{n=0}^{D_{\mathcal{O}}-1}i^{n}\phi_{n}(t)\,|K_{n}\rangle,\;\;\;\;\sum_{n=0}^{D_{\cal O}-1}\phi_{n}^{2}(t)=1. (5) Indeed, this implies that the Heisenberg evolution of the operator |{\cal O}(t)) can be viewed as a point moving along a trajectory on the unit sphere in Krylov space, dubbed Krylov sphere , with coordinates \{\phi_{n}(t),\;n=0,1,\dots,\,D_{\cal O}-1\} satisfying \dot{\phi}_{n}(t)=b_{n}\phi_{n-1}(t)-b_{n+1}\phi_{n+1}(t), (6) with boundary conditions b_{0}=0 and \phi_{-1}=0 . Taking the square of the operator Schrödinger equation i\dot{\cal O}(t)={\cal L}{\cal O}(t) , yields the fundamental identity \sum_{n=0}^{D_{\mathcal{O}}-1}\dot{\phi}_{n}^{2}(t)=\left({\mathcal{O}(t)}|{\cal L}^{2}|{\mathcal{O}(t)}\right)=b_{1}^{2}, (7) valid for all times t . This immediately sets an intrinsic Krylov velocity v_{K}(t)=\sqrt{\sum_{n}\dot{\phi}_{n}^{2}(t)}, (8) which is strictly constant in time along the trajectory on the sphere ( 7 ) and equal to the first Lanczos coefficient b_{1} . In Sec. III 2, we demonstrate the roles of higher Lanczos coefficients b_{n\geq 2} in determining the curvature and torsion of the trajectory in Krylov space, thus providing a complete description of Krylov geometry. Let us conclude the section by demonstrating the equivalence between ( 7