arXiv:2609.40254
Quantum oblique eigenprojection
Alexander M. Dalzell, Yuan Su
Abstract
Every square matrix decomposes its underlying Hilbert space into generalized, nonorthogonal eigensubspaces. We show that a quantum computer can perform such an oblique eigenprojection $Π$ given block encoding access to the input matrix. Our approach has a query complexity nearly linear in the inverse gap and a normalization factor close to $\lVertΠ\rVert$ under a spectral-set condition on the input. This covers common assumptions on the numerical range or diagonalizability and matches known results for orthogonal eigenprojections. We achieve this with a two-sided block preconditioning that uses a discrete Fourier transform of the matrix resolvent. We describe applications to: (i) preparing eigenstates of matrices with complex eigenvalues, extending the quantum eigenvalue transformation algorithm of Low and Su beyond real spectra; (ii) solving continuous-time algebraic Riccati equations, cubically speeding up a prior solver of Rodenas-Ruiz, Zhao, and Lee; and (iii) solving ordinary Sylvester equations, quadratically improving a direct augmented method of Wang and Liu. Our result suggests a promising route to applying nonanalytic matrix functions on quantum computers.
AI Summary
The paper studies how to implement the Riesz projector $\Pi$ onto generalized eigenspaces of a non-normal matrix selected by a region of the complex plane, given block-encoding access to the matrix. Standard discretized resolvent integration produces a projector block encoding whose normalization can scale as $\|\Pi\|/\delta$, introducing extra inverse-gap costs when used in subsequent algorithms. The construction Fourier-transforms the discretized resolvent, explicitly characterizes its block-circulant structure, and applies two-sided diagonal block preconditioning before QSVT inversion. For the unit disk, assuming spectral sets for the inner restriction and inverse outer restriction with constant $c$ and gap $\delta$, it obtains normalization $O(c\|\Pi\|)$ using $\widetilde O(c\alpha_A\|\Pi\|/\delta)$ queries, where $\alpha_A$ is the input normalization. The assumptions include separated per-subspace numerical ranges with constant $c$, or diagonalizable restrictions with $c$ given by their eigenbasis condition numbers. Analytic separating maps extend the construction to other regions; an explicit rational map handles half-planes. Applications improve inverse-gap dependence from $\widetilde O(\delta^{-2})$ to $\widetilde O(\delta^{-1})$ for complex-spectrum eigenstate preparation and a direct augmented Sylvester solver, and from $\widetilde O(\delta^{-3})$ to $\widetilde O(\delta^{-1})$ for a Riccati solver, holding other parameters fixed. Costs still depend on projector norms, spectral-set constants, and application-specific conditioning or overlap. The method does not directly cover a product-of-resolvents matrix-sign construction. General regions require constructing a separating map and controlling its cost and transformed gap; direct preconditioning on arbitrary contours remains open.
Error found: The Sylvester construction directly yields solution normalization proportional to sqrt(1+||X||²), not ||X||, so its claimed constant-factor-optimal normalization requires additional amplification or a lower bound on ||X||.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 41 pages, 3 figures