Monash Complex Computational Modelling Laboratory
We build computational models of complex systems in economics, finance, climate, health and society, and we run them on hardware we control.
Our members work on problems where the data are large, the structure is dynamic and the standard toolkit runs out: high-dimensional panels, high-frequency series, cortical signal records, text corpora in several languages, and frontier models of production under climate stress.
Methods come from econometrics, computational statistics, machine learning and natural language processing. Most of our output carries a replication package in R or Python, and we prefer methods whose behaviour we can characterise over methods that only score well.
The laboratory also functions as a shared compute and methods resource for colleagues across the School and the campus. Researchers whose work exceeds the memory available on a standard machine, or whose ethics approvals prevent sending data to commercial cloud services, can run that work here.
Two co-directors lead the laboratory: Professor Erniel B. Barrios and Dr Nazirul Hazim A. Khalim.
Financial and neuro-econometrics
Volatility modelling applied to both financial markets and cortical signal data, including cross-sector transmission between AI and electricity markets.
2 projectsClimate, agriculture and environment
Stochastic frontier estimation of agricultural production under extreme weather, emissions, environmental disclosure and food security.
3 projectsInformation, narrative and disinformation
Computational social science on narrative change and on the market conditions under which verified information is supplied.
2 projectsComputational heritage and culture
Phylogenetic and multi-model methods applied to the material culture of Southeast Asia.
1 projectHousehold finance and well-being
Longitudinal and explainable-AI modelling of household financial resilience, quality of life and mental health.
2 projectsMethods and platforms
The estimators, pipelines and open tooling that carry across the project portfolio.
View methodsOn-premise processing
Clinical signal data, licensed corpora and partner records are analysed on machines inside the building. Nothing is routed to a commercial cloud service to be computed.
Local model inference
Open-weight language models run on laboratory GPUs, so text under confidentiality obligations is never sent to a third-party API and no external token is exposed.
Governed access
Access to restricted datasets is mediated, with a record of who ran what and when. Data at rest is encrypted.
Working with the laboratory
We take on partnerships with industry, health services, government and other research groups where the problem is genuinely computational and the data are real.
Typical arrangements: hosted compute and model inference, migration of existing analytics from proprietary packages to reproducible R or Python, joint grant development, co-supervision of higher-degree research candidates, and executive training.