Seminar: Graduate Seminar
Enhancing Quantum Algorithms through Engineered Dissipation
Date:
August,14,2024
Start Time:
12:30 - 13:30
Location:
861, Meyer Building
Zoom:
Zoom link
Add to:
Lecturer:
Yigal Ilin
Research Areas:
| Despite numerous advantages promised by the quantum computers (QCs), noise in modern quantum computers presents a significant challenge, impacting their ability to perform accurate and reliable computations. There are various sources of noise in the QCs, including decoherence, operational errors, and environmental interactions, which cause qubits to lose their quantum state or introduce errors during operations. These noise factors limit the performance of quantum algorithms, especially on Noisy Intermediate-Scale Quantum (NISQ) devices, which are the current generation of quantum computers. Enhancing the abilities of quantum computers requires development of methods to characterize, mitigate and correct the errors in NISQ devices and beyond. My research aims to expand the capabilities of current and future NISQ devices by introducing new methods to study noisy discrete-time dynamics in the quantum devices, and proposing new ways to design quantum algorithms and circuits that are more resilient to noise. Specifically, as I will demonstrate, introducing dissipative operations in the form of mid-circuit measurement RESET gates into local channels enables creation of a benchmarking method that efficiently learns a local noisy quantum channel. Furthermore, incorporating such dissipative operations within variational quantum algorithms results in a framework inherently more resilient to both coherent and non-coherent errors, leading to significantly improved performance in noisy environments. First, I will introduce a scalable method for learning local quantum channels using local expectation values measured on their steady state. Inspired by algorithms for learning local Hamiltonians, our approach requires non-trivial steady states and non-unital channels, feasible to implement with mid-circuit measurements or RESET gates on current quantum computers. We demonstrate that the full structure of such channels is encoded in their steady states, and can be learned efficiently using only the expectation values of local observables on these states. Focusing on the learning and validation of the noisy dynamics, we provided multiple examples of numerical and experimental evidence (on an IBM Quantum machine) that the method proposed in our work allows for verification and learning of a full noise model from Pauli measurements on a single circuit output. The experimental demonstration of our technique on IBM Quantum superconducting devices provided a sound basis for its applicability. Second, I will introduce dissipative variational quantum algorithms (D-VQAs) by incorporating dissipative operations such as qubit RESET and stochastic gates into variational quantum circuits. We demonstrated that integrating these operations offers two significant advantages: 1) Allowing for the creation of mixed states, eliminating the need for ancilla qubits when the VQA trial state is mixed. 2) D-VQA circuits can mitigate some non-coherent errors similar to how unitary variational quantum algorithms (VQAs) mitigate coherent errors, resulting in significantly improved performance in noisy environments, particularly when targeting states with low purity. We demonstrated the advantages of our D-VQA ansatz through a simple toy model and through classical simulations of variational Gibbs state preparation, a central problem in quantum computation and information of both theoretical and practical importance. Focusing on periodic 1D models, we have shown that a simple brick-wall architecture with dissipative gates matches state-of-the-art results in the noiseless case and significantly outperforms other coherent variational algorithms in the noisy case. Ph.D. Under the supervision of Prof. Itai Arad, Prof. Guy Bartal and Prof. Uzi Pereg.
|