Code café

Selected work projects

as a main or the solo contributor

agent_oost

Description: Temporally aware reasoning test of language model agents for completing scientific abstracts.

Code repository: agent_oost

Associated manuscript: 2503.04188

qstab

Description: Adversarial attack on biomedical knowledge in large language models using powerscaled distance-weighted sampling.

Code repository: qstab

Associated manuscript: 2402.10527

crystmorph

Description: Structural parsing and analysis of atomic octahedral networks--connected graphs with octahedra as nodes. In a wide variety of materials, the octahedra are formed by atoms. The software includes computational methods for parsing and graph featurization of the atomic structure data for machine learning.

Code repository: crystmorph

Associated manuscript: 2306.12272

pesfit

Description: Distributed multicomponent spectral analysis tool (a computing cluster-friendly alternative for lmfit).

Code repository: pesfit

Associated manuscript: 2102.05604

fuller

Description: The electronic band structure (BS) contains essential information about various physical properties of the material. Angle-resolved photoemission spectroscopy (ARPES) and its associated techniques provide the closest measurements of the electronic BS, but interpretation of the resulting data often requires extensive fitting routines and scales poorly as the complexity of data increases. Since reconstructing the band dispersion that constitutes the BS is a routine task, we have developed this software package (arguably the first) for BS reconstruction and digitization from experimental BS mapping data. The reconstructed bands can be used to compare with theory at various levels and complement existing computational materials science databases. The software package is part of the mpes-kit toolkit and has the following features.

Pre- and post-processing routines for photoemission band mapping data.
Machine learning-based BS reconstruction based on Markov random field.
Functionalities to generate BS using designated multidimensional polynomials basis.
Numerical metrics to quantify the difference between BS data.

The software package is developed in collaboration with V. Stimper.

Code repository: mpes-kit/fuller

Associated manuscript: 2005.10210

mpes

Description: Multidimensional photoemission spectroscopy (MPES) is an emerging technique for comprehensive characterization of the electronic band structure of materials. MPES instruments detect million-to-billion count photoelectron single events that require to be artifact-corrected and binned to convert to multidimensional image data for downstream analysis. The mpes package achieves this feat using the distributed processing package dask. It also includes associated algorithms for data calibration and artifact correction for the single-event as well as binned data. It is the first and the main part of the toolkit mpes-kit. The software package is developed in collaboration with L. Rettig.

Code repository: mpes-kit/mpes

Associated manuscript: 1909.07714

symmetrize

Description: Symmetrization of 2D pattern using nonrigid landmark registration with symmetry constraints. The method is currently implemented with established nonlinear transforms for registration including the projective transform and thin plate spline.

Code repository: symmetrize

Associated manuscript: 1901.00312

planar2

Description: 2D geometric primitive generator for regular shape recognition in imaging data.

The software package is based on the previous development of planar by C. Duncan.

Code repository: planar2

Associated manuscript: None

peakfind2D

Description: (Yet another) electron diffraction pattern peak detection routine. The output includes manifold information about peak statistics in addition to peak location. Multiple rounds of detection my be run to clean up false positive peaks.

Code repository: peakfind2D

Associated manuscript: None

multifit

Description: Signal generation and signal fitting routine for arbitrary number of (co)sine-modulated exponential decays, a prevalent type of temporal signals in transient impulse response experiments, such as transient absorption spectroscopy. The routine can also perform the so-called global fitting to simultaneously determine the parameters of multispectral (or broadband) signals governed by the same set of time constants.

Code repository: multifit

Associated manuscript: None

Hobby projects

ptw

Description: Parametric time warping (PTW) for nonlinear 1D function alignment. The application ranges from mass spectroscopy-based proteomics to liquid chromatography.

Code repository: ptw

Associated manuscript: P. H. C. Eilers, Parametric time warping, Anal. Chem. 76 404 (2004).

ellipsefit

Description: Least-squares fitting of an ellipse using the coordinates of a few known points.

Code repository: ellipsefit

Associated manuscript: A. W. Fitzgibbon, M. Pilu, R. B. Fisher, Direct least squares fitting of ellipses, ICPR (1996).

Page updated

Google Sites

Report abuse