Cheuk Hin Ho

Intro

Graduate student at the University of British Columbia, working under Prof. Christoph Ortner

Reseach Interest

I am boardly interested in application of machine learning in atomistic modelling, espectically in buind symmetries aware forcefields. Recently getting interested in algorithms for accelerating molecular dynamics simulations.

and other things that interest me...

Imaging processing, Uncertainty quantification, Tensor decomposition formats, Few-shots learning

Entertainment

Outside of work I love endurance running, hiking or any outdoor activity 🏃

Work

Currently I am working under the research group leading by Prof. Christoph Ortner , mainly variational method in parameterized neural-network like wave function solution to the Schrodinger Equation and improving stability of ACE descriptor. Also working on reviewing equilvariant Graph Neural Networks.

Previously working as Research Assitant to Prof. Eric Chung and Prof. Ronald Lui. Also a software developer at DataAnt group under Prof. Kwai Lam Wong on developing C++ machine learning library and a part time software developer at OneEd Limited.

Research interest

• Invariant/Equivariant Machine Learning, including group equivariants
• Approximation Theory and Orthogonal Polynomials
• Bayesian Statistics
• Molecular Dynamics

• Image processing
• Computational geometry with Quasi-Conformal theory
• Uncertainty quantification
• Quantum Mechanics, mainly the Schrodinger Equations and Variational Methods
• Reduced Order Modelling with Machine Learning

Also love coding 😀 Serious to project management and testing, enjoy the satisfaction coming from passing the CI test and publishing a new release 🗿

Photo taken during summer research intership:

Publications

Equal contribution: *

In preparation

• Scalable Learning Framewok for Smooth Parameterization of High Dimension Mapping with Quasi-Conformal Theory, Z. Li *, C. Ho *, L. Lui, in preparation

Publications/Preprints

• Atomic Cluster Expansion without Self-Interaction, C. Ho, T. S. Gutleb, C. Ortner. ArXiv e-prints, 2401.01550, 2024.
• A Multilevel Method for Many-Electron Schrödinger Equations Based on the Atomic Cluster Expansion, D. Zhou, H. Chen, C. Ho, and C. Ortner. SIAM J. Sci. Comput. 46.1 (2024) A105-A129.
• ACEpotential.jl : A Julia Implementation of the Atomic Cluster Expansion, W. Witt, C. van der Oord, E. Gelžinytė, A. Ross, C. Ho, W. Baldwin, T. Järvinen, M. Sachs, J. Kermode, N. Bernstein, G. Csanyi, and C. Ortner. J. Chem. Phys. 159, 164101 (2023)

Invited Talks

• Lightening talk at workshop on Machine Learning Force Fields, Data-Driven Materials Informatics, Institute for Mathematical and Statistical Innovation, University of Chicago, April 4-8, 2024.

Organized workshops

Projects

• Automatic Differentiation workshop series, PACMAN, Institute for Applied Mathematics, University of British Columbia, April , 2024. With Christoph Ortner and Miranda-Holmes

Gallery

• Molecular dynamics simulation of ala-ala-ala with MACE
• Dissociation curve of H8 with ACE parameterized wavefunction optimized with variational monte carlo
• Landmark-intensity Regsitration with Quasi-Conformal theory on MRI images
• Super resolution with a paralle-structured Convolutional Neural Network

Softwares/Codes

• Polynoamials4ML.jl : A Julia package for polynoamials evaluation for machine learning, support fast evaluation, batching, differentiation and compatible with Lux.jl
• ACEpsi.jl : A Julia packge for Atomic Cluster Expansion (ACE) wave function parameterizations
• EquivariantModels.jl : A Julia package for tools for buiding equivariant models - both deep and shallow, WIP
• QCPINN : A python project on Quasi-Conformal Physically Informed neural network, WIP

• ACE1.jl : A Julia packge for Atomic Cluster Expansion (ACE) in Machine Learning of Interatomic Potentials
• ACE1pack.jl : Provides convenience functionality for the usage of ACE1.jl , ACEfit.jl and JuLIP.jl for fitting interatomic potentials using the ACE model.
• Octavian.jl: Octavian.jl is a multi-threaded BLAS-like library that provides pure Julia matrix multiplication on the CPU, built on top of LoopVectorization.jl.

Contact

• Tel: (+852) 97349338 (not in use anymore)/ (+1) 778-955-1248
• Email: cheukhinhojerry@gmail.com / jerryho528@math.ubc.ca

• Linkedin
• GitHub

Elements

i = 0;

while (!deck.isInOrder()) {
    print 'Iteration ' + i;
    deck.shuffle();
    i++;
}

print 'It took ' + i + ' iterations to sort the deck.';