Liang Fu’s research while affiliated with Massachusetts Institute of Technology and other places

The attention mechanism has transformed artificial intelligence research by its ability to learn relations between objects. In this work, we explore how a many-body wavefunction ansatz constructed from a large-parameter self-attention neural network can be used to solve the interacting electron problem in solids. By a systematic neural-network variational Monte Carlo study on a moir\'e quantum material, we demonstrate that the self-attention ansatz provides an accurate, efficient, and unbiased solution. Moreover, our numerical study finds that the required number of variational parameters scales roughly as $N^2$ with the number of electrons, which opens a path towards efficient large-scale simulations.

Spin moiré superlattices (SMSs) have been proposed as a magnetic analog of crystallographic moiré systems and a source of electron minibands offering vector-field moiré tunability and Berry curvature effects. However, it has proven challenging to realize an SMS in which a large exchange coupling J is transmitted between conduction electrons and localized spins. Furthermore, most systems have carrier mean free paths l mfp shorter than their spin moiré lattice constant a spin , inhibiting miniband formation. Here, we discover that the layered magnetic semimetal EuAg 4 Sb 2 overcomes these challenges by forming an interface with J ~ 100 milli–electron volts transferred between a Eu triangular lattice and anionic Ag 2 Sb bilayers hosting a two-dimensional electron band in the ballistic regime ( l mfp >> a spin ). The system realizes an SMS with a spin commensurate with the Fermi momentum, leading to a marked quenching of the transport response from miniband formation. Our findings demonstrate an approach to magnetically engineering moiré superlattices and a potential route to an emergent spin-driven quantum Hall state.

A Josephson diode passes current with zero resistance in one direction but is resistive in the other direction. While such an effect has been observed in several platforms, a large and tunable Josephson diode effect has been rare. Here we report that a simple device consisting of a topological-insulator (TI) nanowire side-contacted by superconductors to form a lateral Josephson junction presents a large diode effect with the efficiency $\eta$ reaching 0.3 when a parallel magnetic field $B_{||}$ is applied. Interestingly, the sign and the magnitude of $\eta$ is found to be tunable not only by $B_{||}$ but also by the back-gate voltage. This novel diode effect can be understood by modeling the system as a nano-SQUID, in which the top and bottom surfaces of the TI nanowire each form a line junction and $B_{||}$ creates a magnetic flux to thread the SQUID loop. This model further shows that the observed diode effect is accompanied by the emergence of topological superconductivity in TI nanowire based Josephson junction.

Probing ground-state quantum geometry and topology through optical responses is not only of fundamental interest, but it can also offer several practical advantages. Here, using first-principles calculations on thin films of the antiferromagnetic topological insulator MnBi 2 Te 4 , we demonstrate how the generalized optical weight arising from the absorptive part of the optical conductivity can be used to probe the ground-state quantum geometry and topology. We show that three-septuple-layer MnBi 2 Te 4 film exhibit an enhanced, almost-perfect magnetic circular dichroism for a narrow photon energy window in the infrared region. We calculate the quantum weight in this MnBi 2 Te 4 film and show that it far exceeds the lower bound provided by the Chern number. Our results suggest that the well-known optical methods are powerful tools for probing the ground-state quantum geometry and topology.

A 3D topological insulator (TI) is wrapped by a metallic surface that is topologically protected. While this surface is promising for generating non-Abelian Majorana zero-modes (MZMs) for topological quantum computing, it is a challenge to gap out the whole surface to get rid of low-energy quasiparticles to realize a well-defined fermion parity. Here we propose a novel TI platform that solves this problem and gives rise to a robust topological phase. It consists of a bulk-insulating rectangular TI nanowire laterally sandwiched by two superconductors. In this structure, the top and bottom surfaces individually work as SNS line junctions, forming a nanometer-scale columnar SQUID in which the nanowire cross-section defines the threading flux $\Phi$. We demonstrate that a TI device of this structure indeed presents SQUID-type oscillations of the critical current $I_c$ as a function of the axial magnetic field with period $\Phi_{0}^s=\frac{h}{2e}$ and can be tuned by a back gate to attain vanishing $I_c$ minima, an indication that the supercurrent is completely surface-dominated. Our theory shows that, when the two junctions are asymmetric, a robust topological phase hosting MZMs occurs periodically for $(n-\frac{1}{2})\Phi_{0}^s < \Phi < (n+\frac{1}{2})\Phi_{0}^s$ with odd-integer n.

Ultrafast photoexcitation offers a novel approach to manipulating quantum materials. One of the long-standing goals in this field is to achieve optical control over topological properties. However, the impact on their electronic structures, which host gapless surface states, has yet to be directly observed. Here, using time- and angle-resolved photoemission spectroscopy, we visualize the photoinduced evolution of the band structure in Bi_{y}(Pb_{1-x}Sn_{x})_{1-y}Se(111) films from topological to trivial insulators. Following near-infrared ultrafast laser excitation, we observe that the topological surface state opens a substantial gap of up to 0.1 eV. Considering the topological phase diagram associated with lattice distortion and atomic displacement, we show that a uniaxial strain generated by the ultrafast optical pulse is sufficiently effective and strong for the observed topological phase transition. Our Letter highlights the potential of optical tuning of materials through laser excitation to control topological properties on ultrafast timescales.

We show that quantum geometry sets a bound on the $q^4$ term in the static structure factor S(q). Bands that saturate this bound satisfy a form of Laplace's equation, leading us to refer to them as \textit{harmonic bands}. We provide some examples of harmonic bands in one- and two-dimensional systems.

We introduce an attention-based fermionic neural network (FNN) to variationally solve the problem of two-dimensional Coulomb electron gas in magnetic fields, a canonical platform for fractional quantum Hall (FQH) liquids, Wigner crystals and other unconventional electron states. Working directly with the full Hilbert space of N electrons confined to a disk, our FNN consistently attains energies lower than LL-projected exact diagonalization (ED) and learns the ground state wavefunction to high accuracy. In low LL mixing regime, our FNN reveals microscopic features in the short-distance behavior of FQH wavefunction beyond the Laughlin ansatz. For moderate and strong LL mixing parameters, the FNN outperforms ED significantly. Moreover, a phase transition from FQH liquid to a crystal state is found at strong LL mixing. Our study demonstrates unprecedented power and universality of FNN based variational method for solving strong-coupling many-body problems with topological order and electron fractionalization.

We show that electron crystals compete closely with non-Abelian fractional Chern insulators in the half-full second moir\'e band of twisted bilayer MoTe$_2$. Depending on the twist angle and microscopic model, these crystals can have non-zero or zero Chern numbers. The latter relies on cancellation between contributions from the full first miniband (+1) and the half-full second miniband (-1). For this reason, we call it an anti-topological crystal. Surprisingly, it occurs despite the lowest two non-interacting bands in a given valley having the same Chern number of +1. The anti-topological crystal is a novel type of electron crystal that may appear in systems with multiple Chern bands at filling factors $n > 1$.

We introduce the notion of nonreciprocal superconductors where inversion and time-reversal symmetries are broken, giving rise to an asymmetric energy dispersion. We demonstrate that nonreciprocal superconductivity can be detected by Andreev reflection. In particular, a transparent junction between a normal metal and a nonreciprocal superconductor generally exhibits an asymmetric current-voltage characteristic, which serves as a defining feature of nonreciprocal superconductivity. Unlike the superconducting diode effects, our detection scheme has the advantage of avoiding large critical currents that turn the superconducting state to normal. Last, we discuss candidates for nonreciprocal superconductivity, including graphene, UTe 2 , as well as engineered platforms.