arXiv subjects

Yubo Yang

Publications and source records attributed to Yubo Yang.

28 records · Page 2Linked to original sources

arXiv2023

Training models using forces computed by stochastic electronic structure methods

Quantum Monte Carlo (QMC) can play a very important role in generating accurate data needed for constructing potential energy surfaces. We argue that QMC has advantages in terms of a smaller systematic bias and an ability to cover phase space more completely. The stochastic noise can ease the training of the machine learning model. We discuss how stochastic errors affect the generation of effective models by analyzing the errors within a linear least squares procedure, finding that there is an advantage to having many relatively imprecise data points for constructing models. We then analyze the effect of noise on a model of many-body silicon finding that noise in some situations improves the resulting model. We then study the effect of QMC noise on two machine learning models of dense hydrogen used in a recent study of its phase diagram. The noise enable us to estimate the errors in the model. We conclude with a discussion of future research problems.

cond-mat.mtrl-sci↗

arXiv2023

Stable solid molecular hydrogen above 900K from a machine-learned potential trained with diffusion Quantum Monte Carlo

We survey the phase diagram of high-pressure molecular hydrogen with path integral molecular dynamics using a machine-learned interatomic potential trained with Quantum Monte Carlo forces and energies. Besides the HCP and C2/c-24 phases, we find two new stable phases both with molecular centers in the Fmmm-4 structure, separated by a molecular orientation transition with temperature. The high temperature isotropic Fmmm-4 phase has a reentrant melting line with a maximum at higher temperature (1450K at 150GPa) than previously estimated and crosses the liquid-liquid transition line around 1200K and 200GPa.

cond-mat.mtrl-sci↗

arXiv2020

Analysis of a Backward Euler-type Scheme for Maxwell's Equations in a Havriliak-Negami Dispersive Medium

For the Maxwell's equations in a Havriliak-Negami (H-N) dispersive medium, the associated energy dissipation law has not been settled at both continuous level and discrete level. In this paper, we rigorously show that the energy of the H-N model can be bounded by the initial energy and the model is well-posed. We analyse a backward Euler-type semi-discrete scheme, and prove that the modified discrete energy decays monotonically in time. Such a strong stability ensures that the scheme is unconditionally stable. We also introduce a fast temporal convolution algorithm to alleviate the burden of the history dependence in the polarisation relation involving the singular kernel with the Mittag-Leffler function with three parameters. We provide ample numerical results to demonstrate the efficiency and accuracy of a full-discrete scheme via a spectra-Galerkin method in two dimensions. Finally, we consider an interesting application in the recovery of complex relative permittivity and some related physical quantities.

math.NA↗

arXiv2019

Quantum Monte Carlo Compton profiles of solid and liquid lithium

We computed the Compton profile of solid and liquid lithium using quantum Monte Carlo (QMC) and compared with recent experimental measurements obtaining good agreement. Importantly, we find it crucial to account for proper core-valence orthogonalization and to address density differences when comparing with experiment. To account for disorder effects, we sampled finite-temperature configurations using molecular dynamics (MD), then performed diffusion Monte Carlo (DMC) simulations on each configuration. We used Slater-Jastrow wavefunctions and grand-canonical twist-averaged boundary conditions. A QMC pseudopotential correction, derived from an all-electron DMC simulation of the perfect crystal was also used. Our calculations provide the first all-electron QMC benchmark for the Compton profile of lithium crystal and pseudopotential-corrected QMC Compton profiles for both the liquid and solid.

cond-mat.mtrl-sci↗

arXiv2019

Electronic band gaps from Quantum Monte Carlo methods

We develop a method for calculating the fundamental electronic gap of semiconductors and insulators using grand canonical Quantum Monte Carlo simulations. We discuss the origin of the bias introduced by supercell calculations of finite size and show how to correct the leading and subleading finite size errors either based on observables accessible in the finite-sized simulations or from DFT calculations. Our procedure is applied to solid molecular hydrogen and compared to experiment for carbon and silicon crystals.

cond-mat.mtrl-sci↗

arXiv2019

Numerical analysis of linear and nonlinear time-fractional subdiffusion equations

In this paper, a new type of the discrete fractional Gr{ö}nwall inequality is developed, which is applied to analyze the stability and convergence of a Galerkin spectral method for a linear time-fractional subdiffusion equation. Based on the temporal-spatial error splitting argument technique, the discrete fractional Gr{ö}nwall inequality is also applied to prove the unconditional convergence of a semi-implicit Galerkin spectral method for a nonlinear time-fractional subdiffusion equation.

math.NA↗

arXiv2018

QMCPACK : An open source ab initio Quantum Monte Carlo package for the electronic structure of atoms, molecules, and solids

QMCPACK is an open source quantum Monte Carlo package for ab-initio electronic structure calculations. It supports calculations of metallic and insulating solids, molecules, atoms, and some model Hamiltonians. Implemented real space quantum Monte Carlo algorithms include variational, diffusion, and reptation Monte Carlo. QMCPACK uses Slater-Jastrow type trial wave functions in conjunction with a sophisticated optimizer capable of optimizing tens of thousands of parameters. The orbital space auxiliary field quantum Monte Carlo method is also implemented, enabling cross validation between different highly accurate methods. The code is specifically optimized for calculations with large numbers of electrons on the latest high performance computing architectures, including multicore central processing unit (CPU) and graphical processing unit (GPU) systems. We detail the program's capabilities, outline its structure, and give examples of its use in current research calculations. The package is available at http://www.qmcpack.org .

physics.comp-ph↗

arXiv2016

Interpolated wave functions for nonadiabatic simulations with the fixed-node quantum Monte Carlo method

Simulating nonadiabatic effects with many-body wave function approaches is an open field with many challenges. Recent interest has been driven by new algorithmic developments and improved theoretical understanding of properties unique to electron-ion wave functions. Fixed-node diffusion Monte Caro is one technique that has shown promising results for simulating electron-ion systems. In particular, we focus on the CH molecule for which previous results suggested a relatively significant contribution to the energy from nonadiabatic effects. We propose a new wave function ansatz for diatomic systems which involves interpolating the determinant coefficients calculated from configuration interaction methods. We find this to be an improvement beyond previous wave function forms that have been considered. The calculated nonadiabatic contribution to the energy in the CH molecule is reduced compared to our previous results, but still remains the largest among the molecules under consideration.

physics.chem-ph↗

arXiv2015

How Large are Nonadiabatic Effects in Atomic and Diatomic Systems?

With recent developments in simulating nonadiabatic systems to high accuracy, it has become possible to determine how much energy is attributed to nuclear quantum effects beyond zero-point energy. In this work we calculate the non-relativistic ground-state energies of atomic and molecular systems without the Born-Oppenheimer approximation. For this purpose we utilize the fixed-node diffusion Monte Carlo method, in which the nodes depend on both the electronic and ionic positions. We report ground-state energies for all systems studied, ionization energies for the first-row atoms and atomization energies for the first-row hydrides. We find the ionization energies of the atoms to be nearly independent of the Born-Oppenheimer approximation, within the accuracy of our results. The atomization energies of molecular systems, however, show small effects of the nonadiabatic coupling between electrons and nuclei.

physics.chem-ph↗

arXiv2010

A Study of Bandwidth-Perception Management Mechanisms in IEEE 802.16 Networks

Bandwidth request-grant mechanisms are used in 802.16 networks to manage the uplink bandwidth needs of subscriber stations (SSs). Requests may be sent by SSs to the base station (BS) by means of several mechanisms defined in the standard. Based on the incoming requests, the BS (which handles most of the bandwidth scheduling in the system) schedules the transmission of uplink traffic, by assigning transmission opportunities to the SSs in an implementation-dependent manner. In this paper we present a study of some bandwidth allocation issues, arising from the management of the perception of subscriber stations' bandwidth needs at the base station. We illustrate how the bandwidth perception varies depending on the policy used to handle requests and grants. By means of ns-2 simulations, we evaluate the potential impact of such policies on the system's aggregate throughput when the traffic is composed of Best-Effort TCP flows.

cs.NI↗