Researchers Build Resilient Quantum States for Equations

Toru Fujii A new quantum circuit achieves fidelity of 0.9611 with twelve qubits in preparing smooth real-amplitude states. This single-layer approach uses Ry rotations and CZ entangling layers alongside virtual Rz frame updates to create these states; they are key components in solving partial differential equations using quantum computers. The design efficiently prepares the necessary states for tackling complex mathematical problems used across various scientific fields.

This single-layer method reduces computational steps compared with existing methods and demonstrates strong resilience against errors common in today’s quantum computers. This new design employs Ry rotations, visualised as adjusting dials to set specific values controlling probability amplitudes within each qubit, alongside CZ entangling layers that connect entangled qubits like linked gears. The team’s approach also uses ‘virtual Rz’ operations implemented as software phase adjustments, effectively compressing data representation via Matrix Product State techniques much like image file compression without losing essential detail.

High fidelity quantum computation using compressed single layer circuits

A fidelity of 0.9611 was achieved with twelve qubits, representing a strong improvement over existing methods that typically suffer accuracy loss when scaled up. Previously, maintaining such high precision required circuits whose two-qubit depth increased rapidly, rendering them impractical for current quantum computers.

The new approach circumvents those limitations through a streamlined design utilising Ry rotations and CZ entangling layers alongside virtually implemented Rz operations to prepare smooth real-amplitude states essential for solving complex partial differential equations efficiently. By employing ‘virtual’ adjustments, software phase shifts rather than physical pulses, information is effectively compressed without sacrificing performance or increasing hardware demands; this demonstrates durability against noise common in near-term devices.

Furthermore, utilising six qubits, their single-layer circuit achieved an ideal fidelity score of 0.9617, exceeding that of comparable circuits employing RealAmplitudes with CZ entangling gates which yielded only 0.9612 under identical conditions.

Virtual rotation assumptions limit current fidelity but offer routes to error reduction

Preparing these specialised quantum states promises more accurate modelling of complex systems governed by partial differential equations; however the authors acknowledge a reliance on an assumption potentially untrue in practice. Specifically, treating ‘virtual’ Rz rotations, software adjustments mimicking hardware operations, as perfect introduces a potential source of error when deployed on actual quantum devices. Imperfections within underlying technology may introduce errors when utilising these software shortcuts instead of direct hardware implementation.

Smooth real-amplitude quantum state preparation is key for quantum solvers tackling dissipative partial differential equations like LCHS, where discretized positive weights must be encoded into amplitudes. While exact state preparation can ideally reach unit fidelity, its two-qubit depth increases quickly causing fidelity loss on near-term devices. A low-depth ansatz tailored to typical damped PDE dynamics offers an effective solution; this single-layer circuit achieves high ideal fidelity with O(n) depth and greater noisy fidelity than deeper constructions. Achieving 0.9611 ideal fidelity using twelve qubits indicates comparable expressibility at this reduced depth.

The researchers developed a new method for preparing smooth real-amplitude quantum states which are important for modelling complex systems governed by partial differential equations. Their single-layer circuit demonstrated higher noisy fidelity, reaching 0.9598 compared to 0.6118 in simulations under depolarising noise, and maintained comparable performance to more complex circuits while requiring less computational depth. The approach utilises virtual Rz rotations, an assumption that currently limits achievable fidelity but may offer routes towards error reduction as the technology matures. This work suggests potential benefits when using near-term devices for solving dissipative PDEs.