ChatDiff: A ChatGPT-based diffusion model for long-tailed classification.

Long-tailed data distributions have been a major challenge for the practical application of deep learning. Information augmentation intends to expand the long-tailed data into uniform distribution, which provides a feasible way to mitigate the data starvation of underrepresented classes. However, most existing augmentation methods face two significant challenges: (1) limited diversity in generated samples, and (2) the adverse effect of generated negative samples on downstream classification performance.

Stopping power of hot dense deuterium-tritium plasmas mixed with impurities to charged particles.

In this work, we studied the stopping power of deuterium-tritium (DT) plasmas mixed with impurities to the injected charged particles. Based on the Brown-Preston-Singleton model, the analytical expression for the change ratio of stopping power (denoted by η) induced by impurities in DT plasmas is developed, in which both classical short-distance collision part and quantum correction contribution are purely linear response to the impurity concentration ξ_{X}, while the classical long-range collision brings about higher-order nonlinear response to ξ_{X}. Furthermore, the expression for change ratio of deposition depth (denoted by χ) of charged particles induced by impurities in DT plasmas is also derived.

Spin-Orbit Coupling Effects on Ligand-Free Icosahedral Matryoshka Superatoms.

With the help of density functional theory, a series of matryoshka superatoms X@Y@X (X = Ge, Y = Zn; X = Sn, Y = Mg, Mn, Zn or Cd; X = Pb, Y = Mg, Mn, Cd or Hg) with icosahedral symmetry has been extensively studied, to focus on the influence of the spin-orbit coupling on geometries, stabilities, electronic structures and magnetic moments for these clusters. Generally speaking, the effect of spin-orbit coupling is highly correlated with composition elements of these clusters. Ge@Zn@Ge is little affected by the spin-orbit coupling, while clusters containing Sn atom will generally undergo a moderate influence on their atomization energy, HOMO-LUMO gap and projected density of states.

Dynamic properties of the energy loss of multi-MeV charged particles traveling in two-component warm dense plasmas.

The energy loss of multi-MeV charged particles moving in two-component warm dense plasmas (WDPs) is studied theoretically beyond the random-phase approximation. The short-range correlations between particles are taken into account via dynamic local field corrections (DLFC) in a Mermin dielectric function for two-component plasmas. The mean ionization states are obtained by employing the detailed configuration accounting model.

Transport properties of hydrogen-helium mixtures at extreme density and temperature conditions.

We perform a systematic study of hydrogen-helium mixtures using quantum molecular dynamics (QMD) with a focus on the equations of state and structural and transport properties such as electrical conductivity, diffusion, and viscosity at conditions of giant planet interiors of 0.2∼2.3 g/cm(3) and 1000∼80,000 K for a typical helium mass fraction of 0.

Generalized Lenard-Balescu calculations of electron-ion temperature relaxation in beryllium plasma.

The problem of electron-ion temperature relaxation in beryllium plasma at various densities (0.185-18.5g/cm^{3}) and temperatures [(1.

Quantum molecular dynamics study of expanded beryllium: evolution from warm dense matter to atomic fluid.

By performing quantum molecular dynamics (QMD) simulations, we investigate the equation of states, electrical and optical properties of the expanded beryllium at densities two to one-hundred lower than the normal solid density, and temperatures ranging from 5000 to 30000 K. With decreasing the density of Be, the optical response evolves from the one characteristic of a simple metal to the one of an atomic fluid. By fitting the optical conductivity spectra with the Drude-Smith model, it is found that the conducting electrons become localized at lower densities.

Ab initio molecular dynamics study of high-pressure melting of beryllium oxide.

We investigate, through first-principles molecular dynamics simulations, the high-pressure melting of BeO in the range 0 ≤ p ≤ 100 GPa. The wurtzite (WZ), zinc blend (ZB), and rocksalt (RS) phases of BeO are considered. It is shown that below 40 GPa, the melting temperature for the WZ phase is higher than that for the ZB and RS phases.

Quantum molecular dynamics simulations of the thermophysical properties of shocked liquid ammonia for pressures up to 1.3 TPa.

We investigate via quantum molecular-dynamics simulations the thermophysical properties of shocked liquid ammonia up to the pressure 1.3 TPa and temperature 120,000 K. The principal Hugoniot is predicted from the wide-range equation of state, which agrees well with the available experimental measurements up to 64 GPa.

Quantum molecular dynamic simulations of warm dense carbon monoxide.

Using quantum molecular dynamic simulations, we have studied the thermophysical properties of warm dense carbon monoxide under extreme conditions. The principal Hugoniot pressure up to 286 GPa, which is derived from the equation of state, is calculated and compared with available experimental and theoretical data. The chemical decomposition of carbon monoxide has been predicted at 8 GPa by means of pair correlation function and the charge density distribution.