Near-term quantum computation with neutral atoms:From electronic states, to gates and algorithms
Year: 2025
Type: other
Abstract: The study of quantum many-body systems, and their static and dynamic properties, is one of the most challenging problems of modern physics. Its field of application lies for instance in quantum chemistry, condensed matter physics and pharmaceutical sciences. Investigating the physics of such systems is, however, a significant challenge on present computers. The problem arises from an exponential scaling of computational resources required with respect to the problem size, hence imposing a constraint on the systems that can be studied. This motivated the development of quantum simulators, controlled quantum systems that can be programmed to carry out such calculations – and enable simulations with a linear scaling in the computational resources. Currently, these devices, more specifically quantum computers, are in the noisy intermediate scale quantum (NISQ) era – meaning that various noise mechanisms and limited resources in terms of quantum bits (qubits) prevent large-scale computation. Among the physical systems underlying these computers, those of superconducting electronic circuits and trapped ions were adopted early on and remain promising platforms. In the last few years however, quantum computers based on neutral atoms trapped in optical tweezers have developed rapidly – and offer solutions to the challenge of scalability and limited connectivity faced by other platforms. In this thesis, we investigate the challenges faced in the implementation of useful large-scale quantum algorithms – and present various approaches to move towards this goal from the current NISQ era, by leveraging the unique features of neutral atom devices. In the first two chapters we will introduce quantum computing with neutral atoms. We will further elaborate on computational methods and theoretical techniques that are required to study and control such systems. In the following chapter of the thesis, we model and analyse the various error mechanisms associated with neutral atom quantum computers. This includes performing quantum defect calculations to compute decay times of the Rydberg states used to mediate entanglement, as well as modelling the errors arising from experimental sources such as laser intensity and frequency noise. Our model allows us to compute the fidelity of quantum operations – the distance between the ideal operation and the realized operation in a noisy system. With this model, we then investigate the problem of devising laser controls to implement two-qubit gates that are robust to control deviations. For this purpose, we adapt an augmented Lagrangian numerical method that allows us to incorporate experimental constraints. We find pulses that retain a high fidelity F~0.999 in the presence of significant experimental noise, hence enhancing the capabilities of current neutral atom devices. Using our model, we then propose a strategy to select a suitable Rydberg state, further minimizing the impact of the error mechanisms. In the following chapter, we investigate the problem of implementing k-controlled phase gates, CkP (for k=1, 2), that are of importance in quantum compilation and variational quantum algorithms – two crucial applications for near-term quantum computers. Such gates implement unitary transformations that are continuous with respect to input parameters, meaning we must find a family of pulses – one for each value of the input parameter(s). This is a challenging problem that traditional numerical methods often struggle with. Here we adopt a neural network-based optimization approach, as such networks can approximate any general continuous function. Using our method, we devise pulse families to implement these gates on a time scale close to the quantum speed limit, and for a finite interaction strength – hence mitigating the losses that propagate in time. In contrast to previous such numerical approaches applied to neutral atoms that return a pulse for a fixed angle, our approach returns a piecewise continuous pulse family, which has crucial advantages for near-term algorithms. Compared to decomposition-based approaches, these families are (on average) 2.2 and 4.6 times faster for k=1,2 respectively. In the final research chapter before the conclusion, we explore the co-design of quantum algorithms and hardware-native gate sets for neutral atom quantum computers, focusing on variational quantum algorithms for quantum chemistry. We introduce an analytical scheme for implementing k-controlled, l-target unitary gates and apply it to the unitary coupled cluster with singles and doubles (UCCSD) ansatz. By leveraging native multiqubit gates and long-range interactions, we significantly reduce the total entangling gate time and Rydberg state occupation compared to decomposition-based approaches. We further propose a tensor network-assisted state preparation subroutine using matrix product states (MPS) as high-quality reference states to mitigate barren plateaus and accelerate convergence in near-term and fault-tolerant algorithms. Our results demonstrate that neutral atom platforms offer a distinct advantage in resource efficiency and algorithmic performance for quantum chemistry applications.
Source: TU/e Research Portal
Author: Mohan, Madhav
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7112230318
