Dynamics and Neural Systems Group · School of Physics, The University of Sydney

I build open-source scientific software for time-series analysis, statistical inference, and machine learning.

My research is multi-disciplinary, from applying methods developed for quantum many-body physics (specifically matrix-product states) to data-driven time-series problems such as classification, imputation, and synthetic data generation, to using statistical and dynamical-systems methods to infer structure in complex, non-stationary processes. Alongside that, I maintain and develop widely-used feature-extraction libraries that let researchers compare thousands of time-series statistics in a single, reproducible pipeline.

I work mainly in Python and Julia, with a focus on performance, correctness, and libraries that other people (from hobbyists and researchers to clinicians) can actually pick up and use.

	pyhctsa  ·  Python  ·  docs The most comprehensive time-series feature-extraction package in Python — a native port of the MATLAB hctsa library, giving access to thousands of interpretable time-series features without a MATLAB licence. I am the lead developer and primary contributor; published in the Journal of Open Source Software (2026).
	MPSTime.jl  ·  Julia  ·  docs A Julia package for learning the joint probability distribution of time series directly from data using matrix-product state (MPS) methods from quantum many-body physics. Supports classification, imputation, and synthetic data generation. Co-authored from the ground up; the underlying method is published in Physical Review Research (2025).
	pyspi  ·  Python  ·  docs A library for comparative analysis of pairwise interactions in multivariate time series, evaluating hundreds of statistics of dependence side by side. I am the top contributor by commit volume, working across the estimator suite, testing, and performance.
	c22-usage-examples  ·  Jupyter A curated set of worked examples for pycatch22, the 22-feature canonical time-series characteristics set — practical recipes for feature-based time-series classification.

Using matrix-product states for time-series machine learning J. B. Moore, H. P. Stackhouse, B. D. Fulcher, S. Mahmoodian Physical Review Research 7, 043010 (2025)  ·  doi:10.1103/61h8-8qr5

Introduces MPSTime, the first matrix-product-state algorithm for learning the joint probability distribution underlying a time-series dataset, and applies it to classification and imputation at moderate bond dimension.

pyhctsa: A Python package for highly comparative time-series analysis J. B. Moore, B. D. Fulcher Journal of Open Source Software 11 (123), 10581 (2026)  ·  doi:10.21105/joss.10581

_{Full list on Google Scholar.}

_{Open to research and engineering roles in scientific computing, ML, and data science. Reach me at joshua.moore@sydney.edu.au.}