Quantum computing devices require exceptional control of their experimental parameters to prepare quantum states and simulate other quantum systems. Classical optimization procedures used to find such optimal control parameters have further been shown in idealized settings to exhibit different regimes of learning. Of interest in this work is the overparametrization regime, where for systems with a sufficient number of parameters, global optima for prepared state and compiled unitary fidelities may potentially be reached exponentially quickly. Here, we study the robustness of overparametrization phenomena in the presence of experimental constraints on the controls, such as bounding or sharing parameters across operators, as well as in the presence of noise inherent to experimental setups. We observe that overparametrization phenomena are resilient in these realistic settings at short times, however fidelities decay to zero past a critical simulation duration due to accumulation of either quantum or classical noise. This critical depth is found to be logarithmic in the scale of noise, and optimal fidelities initially increase exponentially with depth, before decreasing polynomially with depth, and with noise. Our results demonstrate that parametrized Ans\"atze can mitigate entropic effects from their environment, offering tantalizing opportunities for their application and experimental realization in near-term quantum devices.
Quantum computing devices require exceptional control of their experimental parameters to prepare quantum states and simulate other quantum systems. Classical optimization procedures used to find such optimal control parameters have further been shown in idealized settings to exhibit different regimes of learning. Of interest in this work is the overparametrization regime, where for systems with a sufficient number of parameters, global optima for prepared state and compiled unitary fidelities may potentially be reached exponentially quickly. Here, we study the robustness of overparametrization phenomena in the presence of experimental constraints on the controls, such as bounding or sharing parameters across operators, as well as in the presence of noise inherent to experimental setups. We observe that overparametrization phenomena are resilient in these realistic settings at short times, however fidelities decay to zero past a critical simulation duration due to accumulation of either quantum or classical noise. This critical depth is found to be logarithmic in the scale of noise, and optimal fidelities initially increase exponentially with depth, before decreasing polynomially with depth, and with noise. Our results demonstrate that parametrized Ans"atze can mitigate entropic effects from their environment, offering tantalizing opportunities for their application and experimental realization in near-term quantum devices.
Overparametrization is one of the most surprising and notorious phenomena in machine learning. Recently, there have been several efforts to study if, and how, quantum neural networks (QNNs) acting in the absence of hardware noise can be overparametrized. In particular, it has been proposed that a QNN can be defined as overparametrized if it has enough parameters to explore all available directions in state space. That is, if the rank of the quantum Fisher information matrix (QFIM) for the QNN's output state is saturated. Here, we explore how the presence of noise affects the overparametrization phenomenon. Our results show that noise can “turn on” previously zero eigenvalues of the QFIM. This enables the parametrized state to explore directions that were otherwise inaccessible, thus potentially turning an overparametrized QNN into an underparametrized one. For small noise levels, the QNN is quasioverparametrized, as large eigenvalues coexists with small ones. Then, we prove that as the magnitude of noise increases all the eigenvalues of the QFIM become exponentially suppressed, indicating that the state becomes insensitive to any change in the parameters. As such, there is a pull-and-tug effect where noise can enable new directions but also suppress the sensitivity to parameter updates. Finally, our results imply that current QNN capacity measures are ill-defined when hardware noise is present. Published by the American Physical Society 2024
We simulate the effects of different types of noise in state preparation circuits of variational quantum algorithms. We first use a variational quantum eigensolver to find the ground state of a Hamiltonian in the presence of noise and adopt two quality measures in addition to the energy, namely, fidelity and concurrence. We then extend the task to the one of constructing, with a layered quantum circuit ansatz, a set of general random target states. We determine the optimal circuit depth for different types and levels of noise, and observe that the variational algorithms mitigate the effects of noise by adapting the optimized parameters. We find that the inclusion of redundant parameterized gates makes the quantum circuits more resilient to noise. For such overparameterized circuits, different sets of parameters can result in the same final state in the noiseless case, which we denote as parameter degeneracy. Numerically, we show that this degeneracy can be lifted in the presence of noise, with some states being significantly more resilient to noise than others. We also show that the average deviation from the target state is linear in the noise level, as long as this is small compared to a circuit-dependent threshold. In this region, the deviation is well described by a stochastic model. Above the threshold, the optimization can converge to states with largely different physical properties from the true target state, so for practical applications it is critical to ensure that noise levels are below this threshold.
An ultimate goal of quantum computing is to perform calculations beyond the reach of any classical computer.It is therefore imperative that useful quantum computers be very difficult to simulate classically, otherwise classical computers could be used for the applications envisioned for the quantum ones.Perfect quantum computers are unarguably exponentially difficult to simulate: the classical resources required grow exponentially with the number of qubits N or the depth D of the circuit.This difficulty has triggered recent experiments on deep, random circuits that aim to demonstrate that quantum devices may already perform tasks beyond the reach of classical computing.These real quantum computing devices, however, suffer from many sources of decoherence and imprecision which limit the degree of entanglement that can actually be reached to a fraction of its theoretical maximum.They are characterized by an exponentially decaying fidelity F ∼ ð1 -ϵÞ ND with an error rate ϵ per operation as small as ≈1% for current devices with several dozen qubits or even smaller for smaller devices.In this work, we provide new insight on the computing capabilities of real quantum computers by demonstrating that they can be simulated at a tiny fraction of the cost that would be needed for a perfect quantum computer.Our algorithms compress the representations of quantum wave functions using matrix product states, which are able to capture states with low to moderate entanglement very accurately.This compression introduces a finite error rate ϵ so that the algorithms closely mimic the behavior of real quantum computing devices.The computing time of our algorithm increases only linearly with N and D in sharp contrast with exact simulation algorithms.We illustrate our algorithms with simulations of random circuits for qubits connected in both one-and twodimensional lattices.We find that ϵ can be decreased at a polynomial cost in computing power down to a minimum error ϵ ∞ .Getting below ϵ ∞ requires computing resources that increase exponentially with ϵ ∞ =ϵ.For a two-dimensional array of N ¼ 54 qubits and a circuit with control-Z gates, error rates better than state-of-the-art devices can be obtained on a laptop in a few hours.For more complex gates such as a SWAP gate followed by a controlled rotation, the error rate increases by a factor 3 for similar computing time.Our results suggest that, despite the high fidelity reached by quantum devices, only a tiny fraction ð∼10 -8 Þ of the system Hilbert space is actually being exploited.
For several decades it has been appreciated that quantum computers hold incredible promise to perform calculations intractable to classical computation. However, this promise has be slow to realize. Dozens of quantum systems are currently being investigated for use in quantum information processing - none of which have yet demonstrated algorithms involving more than a handful of qubits and it remains unclear which, if any, of these systems will ultimately compose a scalable, robust quantum information processing architecture. In this thesis we employ analytical, optimal and algebraic control techniques to evaluate various quantum systems for their potential use in quantum information processing. In doing so, we have additionally identified several novel characterization procedures capable of probing both the coherent and incoherent dynamics of quantum systems. The first part of this thesis discusses work motivated by attempts to utilize donor qubits in silicon as quantum bits. We first propose a measurement of the state of a single donor electron spin using two-dimensional electron gas of a field-effect transistor and electrically detected magnetic resonance. We analyze the potential sensitivity of this measurement and show that it is a quantum nondemolition measurement of an electron-encoded state. We then present the first of two novel qubit characterization procedures. We consider the problem of rapidly characterizing a large number of similarly prepared qubits using techniques from optimal experiment design. All qubits are assumed to evolve according to the same physical processes, though the Hamiltonian parameters may vary from device to device - an inevitability in solid state qubits. We use the Cram\\'er-Rao bound on the variance of a point estimator to construct the optimal series of experiments to estimate these free parameters, and present a complete analysis of the optimal experimental configuration. Though applied to dipole- and exchange-coupled qubits, this technique is widely applicable to other systems. The second part of the thesis discusses the role that control can play in measuring and mitigating noise in qubit systems. Our first result describes a method for quickly simulating the effects of arbitrary markovian noise on qubit systems through the use of a numerically optimized, multi-state Markovian fluctuator. This ability to rapidly simulate the noisy qubit evolution allows us to compute control sequences capable of maximally decoupling the qubit from the noise source. We then introduce the second characterization procedure of the these, showing that a single measurable and controllable qubit may act as a spectrometer of dephasing noise. We show that the formalism of dynamical decoupling can be used to estimate the short-time correlation function of the noise source, while long time correlations may be estimated by a very simple series of free evolution experiments. This technique is applicable to the wide range of physical implementations which suffer from dephasing noise.The final part of this thesis demonstrates that trapped neutral atoms may be utilized for the robust simulation of complex systems exhibiting a topological phase. We present a method to simulate the toric code Hamiltonian stroboscopically, and demonstrate that our technique preserves the ground state degeneracy . Furthermore, we introduce a dissipative mechanism allowing for thermalization of the system to a finite temperature or direct cooling to the ground state manifold.
Quantum computing’s burgeoning influence extends across disciplines such as cryptography, machine learning, and optimization problems, promising computational advantages that surpass the limitations of classical approaches. Despite their theoretical potential, the fidelity of quantum algorithms is compromised by the detrimental effects of noise and decoherence, resulting in substantial errors during the execution of quantum circuits. Although Hardware Efficient Ansatz (HEA) circuits are designed to generate pure quantum states, ideally represented by state vectors, the presence of noise in physical quantum devices transforms these pure states into mixed states, necessitating the use of density matrices and thereby complicating error analysis. Computing noise induced errors in density matrices with high precision remains computationally demanding because of the exponential scaling of the underlying mathematical operations. However, the shallow depth and near term compatibility of HEA circuits make them well suited for variational quantum methods, offering a practical approach within the constraints of current hardware. This paper proposes an OpenMP based parallelization methodology designed to enhance the computational efficiency of noise error estimation in HEA circuits. The integration of classical parallel computing techniques into quantum circuit simulation presents a viable approach to improving the scalability and precision of noise aware quantum computing research. Experimental results indicate an average speedup of approximately 22.34 for pure states and 3.89 for mixed states across varying numbers of qubits.
Noisy Intermediate-Scale Quantum (NISQ) devices impose structural constraints on the training of Variational Quantum Algorithms (VQAs). In the absence of full error correction, each quantum gate introduces a non-zero probability of error that accumulates with circuit depth, while gradient estimation through finite sampling adds additional statistical variability. As a result, convergence depends on a delicate balance between physical coherence and estimation variance. In this setting, the depth of the variational ansatz not only determines model expressivity, but also its operational feasibility under noise. Without an explicit characterization of the interaction between error accumulation, number of measurements (shots), and control strategies, increasing experimental resources may fail to improve performance and can even become counterproductive. In this work, we empirically analyze how circuit depth affects the practical signal of the gradient under depolarizing noise. Across multiple configurations (q=4−7qubits), we observe a pattern consistent with an effective exponential attenuation of the coherent gradient contribution, characterized by a per-layer rateϵeff. The estimated magnitude ofϵeffdepends on circuit complexity and remains largely independent of the measurement budget, indicating that it primarily reflects architectural and physical factors rather than sampling effects. The analysis of adaptive control strategies further suggests that accuracy gains are constrained by gradient attenuation, rather than scaling significantly with circuit depth. However, a crossmode comparison across three distinct training strategies reveals that, while accuracy improvements remain modest, adaptive perturbations substantially improve the statistical detectability of the attenuation signal, acting as diagnostic probes rather than mere optimization heuristics. Taken together, these results suggest that the usable depth of a variational ansatz can be interpreted as an emergent property of the effective signal-to-noise regime, governed by a measurable structural parameter whose ranking across circuit configurations is preserved independently of the training mode. This framework enables principled comparison of circuit architectures in terms of their effective trainability under noise, providing guidance for architectural and control design in Quantum Machine Learning under NISQ conditions.