Kagome Lattices
The Quantum Advancement of Materials Shallow Variational Science Ground State of Frustrated Kagome Lattice Magnet Probed via Quantum Eigensolver
Fundamental research in quantum magnetism models is still being driven by the desire to understand complex magnetic materials. A group of researchers led by Abdellah Tounsi, Nacer Eddine Belaloui, and Abdelmouheymen Rabah Khamadja has announced a major breakthrough in simulating one such complicated system, the antiferromagnetic Heisenberg model on the kagome lattice. Primarily from Constantine Quantum Technologies and Frères Mentouri University Constantine 1 in Algeria, with partners from Purdue University in the USA and the University of Science and Technology Houari Boumediene in Algeria, the team effectively used a Variational Quantum Eigensolver (VQE) built on actual quantum hardware to precisely identify the ground state of the system.
The kagome lattice is a geometrically frustrated system that is well-known for its unusual magnetic characteristics, such as the ability to exhibit topological phases and quantum spin liquids that are pertinent to quantum computing. Because Preparing this system’s ground state is a non-trivial task because of its frustrated nature. A triangle and a star were the minimal kagome cells that the researchers concentrated on. a study marks a significant advancement in the use of near-term quantum computers (NISQ devices) to characterize the ground state of an antiferromagnetic Heisenberg model.
You can also read Reduced Density Matrix RDMs For Many-Body Systems
A Hardware-Efficient Ansatz for Robust Computation
The creation of a shallow, hardware-efficient quantum circuit (ansatz) is a key novelty of this work. It was essential for addressing the short coherence times and noise of NISQ devices. The parameter space of the ansatz was intended to be naturally Euclidean. By taking advantage of the Fubini-Study metric, this unique design was made, guaranteeing a parameter space devoid of singularities and streamlining the optimization procedure. In addition to being hardware economical, the circuit structure is naturally trainable.
The researchers created a stable optimisation landscape that makes training easier by building the ansatz so that the Fubini-Study metric is diagonal and constant. Because of this construction, the normal gradient and the quantum natural gradient coincide.
Implicit-Adaptive Quantum Natural Gradient Descent
The researchers created a novel optimization technique known as Implicit-Adaptive Quantum Natural Gradient Descent (I-AQNGD) to speed up the search for the lowest energy state.
By eliminating the need to measure the Fubini-Study metric directly at each iteration, I-AQNGD preserves the advantages of Adaptive Quantum Natural Gradient Descent (AQNGD). It adds a backtracking search for dynamic step size adaptation to the natural gradient technique. In comparison to the simultaneous perturbation stochastic approximation (SPSA), experiments showed that I-AQNGD maintains competitive runtime while achieving faster convergence in fewer iterations. The study demonstrates that the adaptive feature allows for faster convergence that is less reliant on the starting point by using the backtracking search.
You can also read SEEQC Quantum & IBM Boost DARPA Quantum Benchmarking
Accurate Ground State Determination and Resilience to Noise
The ground state energy for the systems under investigation was successfully and precisely ascertained by the VQE implementation. The VQE converged to -0.749(1) J for the triangular kagome cell, in agreement with known theoretical expectations. The ground state energy of the star-shaped kagome lattice (12 qubits) was determined to be −0.666(2) J, providing fresh information on frustrated magnetic systems. A noteworthy accomplishment in the use of noisy quantum devices for condensed matter physics was made by the team when they implemented the VQE algorithm on the IBMQ Yorktown quantum processor, achieving great gate fidelity (99.7% for single qubit gates and 94.2% for two-qubit gates).
Surprisingly, the technique proved robust against the noise present in existing quantum computers. During the VQE process, the bespoke ansatz was able to precisely recover crucial spin correlation terms without the need for intricate error mitigation strategies.
By analyzing spin-spin correlations and the static spin structure factor, the scientists were able to characterize the dimer state beyond only predicting energy. In addition to showing resilience to noise, this structural characterization offers important new information on the structure of the quantum states and its potential for unusual magnetic properties. It was possible to qualitatively characterize the dimer state using spin correlation and the spin structure factor even in the absence of error mitigation.
You can also read Model Based Optimization For Superconducting Qubit
Error Mitigation Techniques
The team used error mitigation (EM) post-optimization approaches, such as qubit-wise readout error mitigation (REM) and zero noise extrapolation (ZNE), to further improve the accuracy of the final results.
The results show that observable estimations are significantly more accurate when using EM methods. REM improved the detection of local dimers while effectively maintaining the variational principle. ZNE, on the other hand, permits the violation of the Rayleigh-Ritz variational principle, hence it does not provide an upper bound on the exact energy. Impressively, though, ZNE frequently produced the best accuracy across devices when utilized independently, especially when quadratic extrapolation was employed. There may be overlap between the effects of REM and ZNE, as seen by the occasional undershoots.
Pathway to Future Quantum Materials Research
This study develops a strong and reliable approach to investigate intricate quantum many-body systems with near-term quantum computers. A viable route forward is suggested by the ability to use VQE to prepare the ground state of kagome lattice pieces with shallow circuits, even when there is a lot of noise present.
There is a lot of promise for producing expressive, hardware-efficient, and inherently trainable ansatzes at cheap costs by carefully planning the geometry of the ansatz to have an analytically calculable, Euclidean parameter space. The work’s implications for future research include applying this method to bigger, more intricate kagome lattice, incorporating strategies like randomized compilation to deal with coherent noise, and investigating the possibility of finding new quantum phases of matter, like topological phases and quantum spin liquids.
You can also read Quantum Art Uses CUDA-Q For Fast Logical Qubit Compilation
