Classiq and Rolls-Royce Embed Quantum Linear Solvers into Iterative CFD Workflows

Aerospace engineering is moving from abstract complexity bounds to practical, end-to-end operational performance. Through a hybrid quantum-classical computing framework, Classiq Technologies and Rolls-Royce plc have embedded a Quantum Linear Solver (QLS) directly into an industrial Computational Fluid Dynamics (CFD) pipeline.

The partnership focuses on optimizing the simulation of aerodynamic flows inside jet engine components. In standard aerospace design, the discretization of nonlinear momentum terms creates systems of linear equations where matrix dimensions (N) and condition numbers (κ) scale from 106 to 109. These scales push classical supercomputing infrastructures to their operational limits.

The technical challenge lies in the readout bottleneck. While a QLS can represent these large solution vectors using only log2(N) qubits, reading out precise state amplitudes introduces severe sampling and measurement errors. To address this, the team developed an architecture that isolates the quantum processor to handle linear corrections within an outer, classical fixed-point iteration loop modeling steady flow through a 1D nozzle.

This approach shifts the focus from theoretical asymptotic complexity to how near-term fault-tolerant quantum algorithms affect macro-level convergence. The research targets bottlenecks in simulating complex, transonic fluid fields across high-resolution spatial meshes.

The implementation faces significant hardware constraints. Matching classical double-precision tolerances (ε∼10−12) via standard Quantum Singular Value Transformation (QSVT) requires a polynomial degree (d) scaling as d∼κlog(κ/ε). This translates to tens of millions of arbitrary single-qubit rotations. Because compiling each rotation into fault-tolerant Clifford+T gate sequences incurs a circuit depth penalty, the experiment demonstrated that the outer CFD scheme remains stable against relaxed precision limits. The system converges steadily even when low-degree polynomial approximations introduce errors.

This development suggests that the path to quantum advantage in fluid dynamics does not require perfect precision on day one. Instead, value is found by integrating quantum modules into existing classical loops where the algorithm can tolerate approximation.

The industry must now determine if the reduction in circuit depth provided by these hybrid architectures is sufficient to overcome the overhead of error correction in future hardware.

Source