Heterogeneous relational message passing networks for molecular dynamics simulations

With many frameworks based on message passing neural networks proposed to predict molecular and bulk properties, machine learning methods have tremendously shifted the paradigms of computational sciences underpinning physics, material science, chemistry, and biology. While existing machine learning models have yielded superior performances in many occasions, most of them model and process molecular systems in terms of homogeneous graph, which severely limits the expressive power for representing diverse interactions. In practice, graph data with multiple node and edge types is ubiquitous and more appropriate for molecular systems. Thus, we propose the heterogeneous relational message passing network (HermNet), an end-to-end heterogeneous graph neural networks, to efficiently express multiple interactions in a single model with {it ab initio} accuracy. HermNet performs impressively against many top-performing models on both molecular and extended systems. Specifically, HermNet outperforms other tested models in nearly 75%, 83% and 94% of tasks on MD17, QM9 and extended systems datasets, respectively. Finally, we elucidate how the design of HermNet is compatible with quantum mechanics from the perspective of the density functional theory. Besides, HermNet is a universal framework, whose sub-networks could be replaced by other advanced models.

Chiral Phonons in Moire Superlattices

We theoretically demonstrate that moire phonons at the lowest-energy bands can become chiral. A general symmetry analysis reveals that they originate from stacking configurations leading to an asymmetric interlayer binding energy that breaks the $C_{2z}$ symmetry on the moire length scale. Within elastic theory, we provide a complete classification of van der Waals heterostructures in respect to hosting moire chiral phonons and discuss their emergence in twisted bilayer MoS$_2$ as an example. The formation of the chiral phonons can be qualitatively understood using an effective model, which emphasizes their origin in the energy difference between stacking domains. Since moire chiral phonons are highly tunable, with excitation energies in only a few meV, and moire scale wavelengths, they might find potential applications in phononic twistronic devices.

Witten-Morse functions and Morse inequalities on digraphs

In this paper, we prove that discrete Morse functions on digraphs are flat Witten-Morse functions and Witten complexes of transitive digraphs approach to Morse complexes. We construct a chain complex consisting of the formal linear combinations of paths which are not only critical paths of the transitive closure but also allowed elementary paths of the digraph, and prove that the homology of the new chain complex is isomorphic to the path homology. On the basis of the above results, we give the Morse inequalities on digraphs.

On The Discrete Morse Functions for Hypergraphs

A hypergraph can be obtained from a simplicial complex by deleting some non-maximal simplices. In this paper, we study the embedded homology as well as the homology of the (lower-)associated simplicial complexes for hypergraphs. We generalize the discrete Morse functions on simplicial complexes. We study the discrete Morse functions on hypergraphs as well as the discrete Morse functions on the (lower-)associated simplicial complexes of the hypergraphs.

Dynamic Relevance Learning for Few-Shot Object Detection

Expensive bounding-box annotations have limited the development of object detection task. Thus, it is necessary to focus on more challenging task of few-shot object detection. It requires the detector to recognize objects of novel classes with only a few training samples. Nowadays, many existing popular methods based on meta-learning have achieved promising performance, such as Meta R-CNN series. However, only a single category of support data is used as the attention to guide the detecting of query images each time. Their relevance to each other remains unexploited. Moreover, a lot of recent works treat the support data and query images as independent branch without considering the relationship between them. To address this issue, we propose a dynamic relevance learning model, which utilizes the relationship between all support images and Region of Interest (RoI) on the query images to construct a dynamic graph convolutional network (GCN). By adjusting the prediction distribution of the base detector using the output of this GCN, the proposed model can guide the detector to improve the class representation implicitly. Comprehensive experiments have been conducted on Pascal VOC and MS-COCO dataset. The proposed model achieves the best overall performance, which shows its effectiveness of learning more generalized features. Our code is available at https://github.com/liuweijie19980216/DRL-for-FSOD.

Characteristics and Challenges of Low-Code Development: The Practitioners Perspective

Background: In recent years, Low-code development (LCD) is growing rapidly, and Gartner and Forrester have predicted that the use of LCD is very promising. Giant companies, such as Microsoft, Mendix, and Outsystems have also launched their LCD platforms. Aim: In this work, we explored two popular online developer communities, Stack Overflow (SO) and Reddit, to provide insights on the characteristics and challenges of LCD from a practitioners perspective. Method: We used two LCD related terms to search the relevant posts in SO and extracted 73 posts. Meanwhile, we explored three LCD related subreddits from Reddit and collected 228 posts. We extracted data from these posts and applied the Constant Comparison method to analyze the descriptions, benefits, and limitations and challenges of LCD. For platforms and programming languages used in LCD, implementation units in LCD, supporting technologies of LCD, types of applications developed by LCD, and domains that use LCD, we used descriptive statistics to analyze and present the results. Results: Our findings show that: (1) LCD may provide a graphical user interface for users to drag and drop with little or even no code; (2) the equipment of out-of-the-box units (e.g., APIs and components) in LCD platforms makes them easy to learn and use as well as speeds up the development; (3) LCD is particularly favored in the domains that have the need for automated processes and workflows; and (4) practitioners have conflicting views on the advantages and disadvantages of LCD. Conclusions: Our findings suggest that researchers should clearly define the terms when they refer to LCD, and developers should consider whether the characteristics of LCD are appropriate for their projects.

Intrinsic nonlinear Hall effect in antiferromagnetic tetragonal CuMnAs

The nonlinear Hall effect is mostly studied as a Berry curvature dipole effect in nonmagnetic materials, which depends linearly on the relaxation time. On the other hand, in magnetic materials, an intrinsic nonlinear Hall effect can exist, which does not depend on the relaxation time. Here we show that the intrinsic nonlinear Hall effect can be observed in an antiferromagnetic metal: tetragonal CuMnAs, and the corresponding conductivity can reach the order of mA/V$^2$ based on density functional theory calculations. The dependence on the chemical potential of such nonlinear Hall conductivity can be qualitatively explained by a tilted massive Dirac model. Moreover, we demonstrate its strong temperature-dependence and briefly discuss its competition with the second order Drude conductivity. Finally, a complete survey of magnetic point groups are presented, providing guidelines for finding candidate materials with the intrinsic nonlinear Hall effect.

Tunable terahertz plasmons in graphite thin films

Tunable terahertz plasmons are essential for reconfigurable photonics, which have been demonstrated in graphene through gating, though with relatively weak responses. Here, we demonstrate strong terahertz plasmons in graphite thin films via infrared spectroscopy, with dramatic tunability by even a moderate temperature change or an in-situ bias voltage. Meanwhile, through magneto-plasmon studies, we reveal that massive electrons and massless Dirac holes make comparable contributions to the plasmon response. Our study not only sets up a platform for further exploration of two-component plasmas, but also opens an avenue for terahertz modulation through electrical bias or all-optical means.

UniverseMachine: Predicting Galaxy Star Formation over Seven Decades of Halo Mass with Zoom-in Simulations

We apply the empirical galaxy--halo connection model UniverseMachine to dark matter-only zoom-in simulations of isolated Milky Way (MW)--mass halos along with their parent cosmological simulations. This application extends textsc{UniverseMachine} predictions into the ultra-faint dwarf galaxy regime ($ 10^{2},mathrm{M_{odot}} leqslant M_{ast} leqslant 10^{5},mathrm{M_{odot}}$) and yields a well-resolved stellar mass--halo mass (SMHM) relation over the peak halo mass range $10^8,mathrm{M_{odot}}$ to $10^{15},mathrm{M_{odot}}$. The extensive dynamic range provided by the zoom-in simulations allows us to assess specific aspects of dwarf galaxy evolution predicted by textsc{UniverseMachine}. In particular, although UniverseMachine is not constrained for dwarf galaxies with $M_* lesssim 10^{8},mathrm{M_{odot}}$, our predicted SMHM relation is consistent with that inferred for MW satellite galaxies at $z=0$ using abundance matching. However, UniverseMachine predicts that nearly all galaxies are actively star forming below $M_{ast}sim 10^{7},mathrm{M_{odot}}$ and that these systems typically form more than half of their stars at $zlesssim 4$, which is discrepant with the star formation histories of Local Group dwarf galaxies that favor early quenching. This indicates that the current UniverseMachine model does not fully capture galaxy quenching physics at the low-mass end. We highlight specific improvements necessary to incorporate environmental and reionization-driven quenching for dwarf galaxies, and provide a new tool to connect dark matter accretion to star formation over the full dynamic range that hosts galaxies.

Stiefel Liquids: Possible Non-Lagrangian Quantum Criticality from Intertwined Orders

We propose a new type of quantum liquids, dubbed Stiefel liquids, based on $2+1$ dimensional nonlinear sigma models on target space $SO(N)/SO(4)$, supplemented with Wess-Zumino-Witten terms. We argue that the Stiefel liquids form a class of critical quantum liquids with extraordinary properties, such as large emergent symmetries, a cascade structure, and nontrivial quantum anomalies. We show that the well known deconfined quantum critical point and $U(1)$ Dirac spin liquid are unified as two special examples of Stiefel liquids, with $N=5$ and $N=6$, respectively. Furthermore, we conjecture that Stiefel liquids with $N>6$ are non-Lagrangian, in the sense that under renormalization group they flow to infrared (conformally invariant) fixed points that cannot be described by any renormalizable continuum Lagrangian. Such non-Lagrangian states are beyond the paradigm of parton gauge mean-field theory familiar in the study of exotic quantum liquids in condensed matter physics. The intrinsic absence of (conventional or parton-like) mean-field construction also means that, within the traditional approaches, it will be difficult to decide whether a non-Lagrangian state can actually emerge from a specific UV system (such as a lattice spin system). For this purpose we hypothesize that a quantum state is emergible from a lattice system if its quantum anomalies match with the constraints from the (generalized) Lieb-Schultz-Mattis theorems. Based on this hypothesis, we find that some of the non-Lagrangian Stiefel liquids can indeed be realized in frustrated quantum spin systems, for example, on triangular or Kagome lattice, through the intertwinement between non-coplanar magnetic orders and valence-bond-solid orders.