| name | quantum-riccati-solver |
| description | Quantum algorithms for solving algebraic Riccati equations via Riesz projectors and quantum singular value transformations (QSVT). Applications to quantum chemistry RPA and coupled-cluster theory. Activation: quantum riccati, quantum nonlinear matrix, quantum chemistry algorithm, RPA quantum, algebraic riccati equation quantum, quantum singular value transformation chemistry. |
Quantum Solvers for Nonlinear Matrix Equations
Quantum algorithm for solving algebraic Riccati equations using Riesz projectors onto invariant subspaces, implemented via contour-integral resolvents and QSVT, with applications to quantum-chemical RPA and coupled-cluster theory. (arXiv:2605.16189)
Metadata
- Source: arXiv:2605.16189
- Authors: Pablo Rodenas-Ruiz, Andrew Zhao, Joonho Lee
- Published: 2026-05-15
- Categories: quant-ph, physics.chem-ph
Core Methodology
Key Innovation
First quantum algorithm framework for solving algebraic Riccati equations - a fundamental class of nonlinear matrix equations ubiquitous in control theory, quantum chemistry, and many-body physics. The method achieves linear scaling in system size and polynomial scaling in excitation rank, suggesting exponential advantage over classical local-correlation heuristics.
Technical Framework
-
Block-Encoding via Riesz Projectors
- Riccati solutions are encoded as Riesz projectors onto invariant subspaces of an associated non-normal matrix
- The projector P = (1/2πi) ∮_Γ (zI - A)^(-1) dz captures the stabilizing solution
- Contour Γ encloses eigenvalues corresponding to the stable invariant subspace
-
Contour-Integral Resolvent Implementation
- The resolvent (zI - A)^(-1) is approximated using quantum linear system algorithms (QLSA)
- Contour integration discretized via quadrature: sum of weighted resolvent evaluations
- Each resolvent evaluation implemented as a block-encoded linear system solve
-
Quantum Singular Value Transformation (QSVT)
- Polynomial transformations of block-encoded matrices extract the projector
- QSVT provides optimal gate complexity for matrix function approximation
- Enables efficient estimation of correlation-energy density from the solution
-
Application to RPA (Random-Phase Approximation)
- m-particle, m-hole RPA reduces to a Riccati equation for cluster amplitudes
- Under localized-orbital sparsity: O(N) system-size scaling, poly(m) excitation-rank scaling
- Exponential advantage in m over classical local-correlation methods
Implementation Guide
Prerequisites
- QSVT / quantum signal processing library
- Quantum linear system solver (HHL variant or QSVT-based)
- Block-encoding primitives for the problem matrix A
Step-by-Step
-
Formulate the Riccati Equation
A^T X + X A - X B X + C = 0
Identify the associated Hamiltonian matrix H = [[A, -B], [-C, -A^T]]
-
Construct Block-Encoding of H
- Decompose H into efficiently implementable unitary blocks
- Ensure normalization ||H|| ≤ 1 for QSVT compatibility
-
Define Contour and Quadrature
- Choose contour Γ enclosing stable eigenvalues
- Discretize via trapezoidal or Gauss-Legendre quadrature
- Number of quadrature points O(log(1/ε)) for ε precision
-
Implement QSVT-Based Projector
- Design polynomial approximation of the indicator function
- Apply QSVT to block-encoded H
- Extract projector onto stable subspace
-
Estimate Observables
- Compute correlation-energy density: E_corr = Tr(X · C)
- Use amplitude estimation for quadratic speedup in sampling
Code Example (Pseudocode)
H_block = block_encode(H, precision=epsilon)
z_points, weights = contour_quadrature(stable_eigenvalues, n_points=O(log(1/eps)))
projector = sum(w * qlsolve(z_i * I - H_block) for z_i, w in zip(z_points, weights))
X = extract_solution(projector)
E_corr = amplitude_estimate(lambda X: trace(X @ C))
Applications
- Quantum chemistry: RPA, higher-order RPA, coupled-cluster theory
- Control theory: optimal control, Kalman filtering, LQR
- Many-body physics: Green's function calculations, self-consistent field methods
- Machine learning: discrete-time Riccati equations in filtering/smoothing
Pitfalls
- Non-normal matrices: Standard QSVT assumes normal/Hermitian matrices; non-normal H requires careful block-encoding design
- Sparsity assumptions: Linear scaling relies on localized-orbital sparsity; dense systems revert to polynomial scaling
- Contour selection: Poor contour choice leads to ill-conditioned resolvents; eigenvalue gap must be bounded away from zero
- QSVT degree: Polynomial degree scales with condition number; ill-conditioned problems require high-degree polynomials
- State preparation: Input state quality directly affects solution accuracy
Related Skills
- quantum-block-encoding-linear-algebra
- quantum-linear-algebra-block-encoding
- qsvt-quantum-signal-processing
- quantum-chemistry-algorithms
- quantum-algorithm-framework-designer