arXiv subjects
Publications and source records attributed to Yayun Zhang.
Tensor products of quiver representations have been extensively studied; typical examples include the pointwise tensor product and the tensor product induced by the coalgebra structure of path algebras. In this paper, we investigate the growth rates of the number of indecomposable direct summands in tensor powers of quiver representations with respect to these two typical tensor products.
Quantification is a fundamental component of everyday language use, yet little is known about how speakers decide whether and how to quantify in naturalistic production. We investigate quantification in Mandarin Chinese using a picture-based elicited description task in which speakers freely described scenes containing multiple objects, without explicit instructions to count or quantify. Across both spoken and written modalities, we examine three aspects of quantification: whether speakers choose to quantify at all, how precise their quantification is, and which quantificational strategies they adopt. Results show that object numerosity, animacy, and production modality systematically shape quantificational behaviour. In particular, increasing numerosity reduces both the likelihood and the precision of quantification, while animate referents and modality selectively modulate strategy choice. This study demonstrates how quantification can be examined under unconstrained production conditions and provides a naturalistic dataset for further analyses of quantity expression in language production.
Infants develop complex visual understanding rapidly, even preceding the acquisition of linguistic skills. As computer vision seeks to replicate the human vision system, understanding infant visual development may offer valuable insights. In this paper, we present an interdisciplinary study exploring this question: can a computational model that imitates the infant learning process develop broader visual concepts that extend beyond the vocabulary it has heard, similar to how infants naturally learn? To investigate this, we analyze a recently published model in Science by Vong et al., which is trained on longitudinal, egocentric images of a single child paired with transcribed parental speech. We perform neuron labeling to identify visual concept neurons hidden in the model's internal representations. We then demonstrate that these neurons can recognize objects beyond the model's original vocabulary. Furthermore, we compare the differences in representation between infant models and those in modern computer vision models, such as CLIP and ImageNet pre-trained model. Ultimately, our work bridges cognitive science and computer vision by analyzing the internal representations of a computational model trained on an infant visual and linguistic inputs. Project page is available at https://kexueyi.github.io/webpage-discover-hidden-visual-concepts.
Machine learning models of visual action recognition are typically trained and tested on data from specific situations where actions are associated with certain objects. It is an open question how action-object associations in the training set influence a model's ability to generalize beyond trained situations. We set out to identify properties of training data that lead to action recognition models with greater generalization ability. To do this, we take inspiration from a cognitive mechanism called cross-situational learning, which states that human learners extract the meaning of concepts by observing instances of the same concept across different situations. We perform controlled experiments with various types of action-object associations, and identify key properties of action-object co-occurrence in training data that lead to better classifiers. Given that these properties are missing in the datasets that are typically used to train action classifiers in the computer vision literature, our work provides useful insights on how we should best construct datasets for efficiently training for better generalization.
ReaxFF provides a method to model reactive chemical systems in large-scale molecular dynamics simulations. Here, we developed ReaxFF parameters for phosphorus and hydrogen to give a good description of the chemical and mechanical properties of pristine and defected black phosphorene. ReaxFF for P/H is transferable to a wide range of phosphorus and hydrogen containing systems including bulk black phosphorus, blue phosphorene, edge-hydrogenated phosphorene, phosphorus clusters and phosphorus hydride molecules. The potential parameters were obtained by conducting unbiased global optimization with respect to a set of reference data generated by extensive ab initio calculations. We extend ReaxFF by adding a 60° correction term which significantly improves the description of phosphorus clusters. Emphasis has been put on obtaining a good description of mechanical response of black phosphorene with different types of defects. Compared to nonreactive SW potential [1], ReaxFF for P/H systems provides a huge improvement in describing the mechanical properties the pristine and defected black phosphorene and the thermal stability of phosphorene nanotubes. A counterintuitive phenomenon is observed that single vacancies weaken the black phosphorene more than double vacancies with higher formation energy. Our results also show that mechanical response of black phosphorene is more sensitive to defects for the zigzag direction than for the armchair direction. Since ReaxFF allows straightforward extensions to the heterogeneous systems, such as oxides, nitrides, ReaxFF parameters for P/H systems build a solid foundation for the reactive force field description of heterogeneous P systems, including P-containing 2D van der Waals heterostructures, oxides, etc.
Based on extensive evolutionary algorithm driven structural search, we propose a new diphosphorus trisulfide (P2S3) 2D crystal, which is dynamically, thermally and chemically stable as confirmed by the computed phonon spectrum and ab initio molecular dynamics simulations. This 2D crystalline phase of P2S3 corresponds to the global minimum in the Born-Oppenheimer surface of the phosphorus sulfide monolayers with 2:3 stoichiometries. It is a wide band gap (4.55 eV) semiconductor with P-S σ bonds. The electronic properties of P2S3 structure can be modulated by stacking into multilayer P2S3 structures, forming P2S3 nanoribbons or rolling into P2S3 nanotubes, expanding its potential applications for the emerging field of 2D electronics.
The hydration of ions in nanoscale hydrated clusters is ubiquitous and essential in many physical and chemical processes. Here we show that the hydrolysis reaction is strongly affected by relative humidity. The hydrolysis of CO32- with n = 1-8 water molecules is investigated by ab initio method. For n = 1-5 water molecules, all the reactants follow a stepwise pathway to the transition state. For n = 6-8 water molecules, all the reactants undergo a direct proton transfer to the transition state with overall lower activation free energy. The activation free energy of the reaction is dramatically reduced from 10.4 to 2.4 kcal/mol as the number of water molecules increases from 1 to 6. Meanwhile, the degree of the hydrolysis of CO32- is significantly increased compared to the bulk water solution scenario. The incomplete hydration shells facilitate the hydrolysis of CO32- with few water molecules (especially for n = 6) to be not only thermodynamically favorable but also kinetically favorable. This discovery provides valuable insights for designing efficient sorbents including that for CO2 air capture.
Two-dimensional materials have attracted tremendous attention for their fascinating electronic, optical, chemical and mechanical properties. However, the band gaps of most 2D materials reported are smaller than 2.0 eV, which greatly restricted their optoelectronic applications in blue and ultraviolet range of the spectrum. Here, we propose a new stable sulfur nitride (S3N2) 2D crystal that is a covalent network composed solely of S-N σ bonds. S3N2 crystal is dynamically stable as confirmed by the computed phonon spectrum and ab initio molecular dynamics simulations in the NPT ensemble. Hybrid density functional calculations show that 2D S3N2 crystal is a wide, direct band-gap (3.17 eV) semiconductor with good hole mobility. These fascinating electronic properties could pave the way for its optoelectronic applications such as blue or ultra-violet light-emitting diodes (LEDs) and photodetectors.
Today, 2D semiconductor materials have been extended into the nitrogen group: phosphorene, arsenene, antimonene and even nitrogene. Motivated by them, based upon first-principles density functional calculations, we propose a new two-dimensional phosphorus nitride (PN) structure that is stable well above the room temperature, due to its extremely high cohesive energy. Unlike phosphorene, PN structure is resistant to high temperature oxidation. The structure is predicted to be a semiconductor with a wide, indirect band gap of 2.64 eV. More interestingly, the phosphorus nitride monolayer experiences an indirect-to-direct band-gap transition at a relatively small tensile strain. Such dramatic transformation in the electronic structure combined with structural stability and oxidation resistance at high temperature could pave the way for exciting innovations in high-speed ultrathin transistors, power electronic modules, ultra-high efficiency LEDs and semiconductor lasers.