arXiv:2609.25660
Schrödingerization for quantum linear systems problems
Yin Yang, Yue Yu, Long Zhang
Abstract
We develop a Schrödingerization algorithm for quantum linear systems problems in two higher dimensions. The earlier LC-Schrödingerization approach represents the solution as a time integral of a homogeneous convection response and implements this integral by a linear combination of unitaries (LCU) over evolution times. We instead use Duhamel's principle to incorporate the integral into an inhomogeneous convection equation with zero initial data. Fourier projection in the convection variable and Schrödingerization in a second auxiliary variable then realize the solution without the separate LCU step. A joint choice of the kernel and recovery procedure gives an evolution time independent of the target accuracy, a uniformly bounded $L^2$ kernel normalization, and accurate recovery on a fixed interval. We establish the periodization and discretization bounds and analyze the interval recovery probability. With block preconditioning, we retain a single logarithmic precision factor and obtain linear condition-number dependence without variable-time amplitude amplification. Under exact oracle access and given a valid constant-factor solution-norm estimate, the algorithm uses $\mathcal O(κ_A\log(1/\varepsilon))$ queries to each original input oracle, with constant success probability and $\ell^2$ state error at most $\varepsilon$. The matrix-query bound matches the standard worst-case scaling when the supplied norm bounds are tight. UnitaryLab simulations on positive-definite and indefinite systems demonstrate the solution accuracy and probability gain from interval recovery.
AI Summary
The paper studies whether quantum linear-system inversion can be implemented through Schrödingerization without the separate linear combination of unitaries over evolution times used in LC-Schrödingerization. It replaces the time-integrated homogeneous convection response with the forced equation $\partial_t v=A\partial_p v+\zeta(p)b$, initialized at zero. Fourier projection in $p$ produces an inhomogeneous ODE; augmentation and Schrödingerization introduce a second auxiliary variable $q$ and enable Hamiltonian simulation. The representation accommodates both spectral signs, with Hermitian dilation handling general invertible matrices. A Gaussian-smoothed signed-interval kernel gives recovery uniformly on $p\in[-1,1]$ at time $T=5\alpha_{A^{-1}}$, independent of target accuracy. Its $L^2$ norm remains bounded, and the continuous relative recovery error is at most $e^{-r}/\operatorname{erf}(1/\sqrt2)$. Periodization and Gaussian Fourier-tail estimates establish discretization bounds. Coherent interval recovery cancels the grid-normalization penalty. Combining the evolution solver with an existing block preconditioner yields $O(\kappa_A\log(1/\varepsilon))$ queries to each original input oracle, constant success probability, and state error at most $\varepsilon$, without variable-time amplitude amplification. This matches the matrix-query scaling already available through LC-Schrödingerization rather than improving it. The bounds assume exact block encodings, exact input and auxiliary-state preparation, and a supplied valid constant-factor solution-norm estimate. Auxiliary preparation gate costs and obtaining that estimate are not included. Statevector circuit simulations on two- and four-dimensional systems test recovery and discretization, not the asymptotic query bound; preconditioning and amplitude amplification are not executed.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.