PROF. TYLL KRÜGER, MOCOS GROUP
WROCLAW UNIVERSITY OF SCIENCE AND TECHNOLOGY
PROF. TYLL KRÜGER, MOCOS GROUP
WROCLAW UNIVERSITY OF SCIENCE AND TECHNOLOGY
Agent modelling for the MOCOS group – Modelling Coronavirus Spread
PROJECT OBJECTIVE
Challenge
Using an agent-based model to model the COVID-19 epidemic.
Project objective
Analysing the risk, making forecasts for the following days and weeks, providing recommendations for COVID-19 eradication. Supporting public institutions in bringing the epidemic to a state of extinction (sub-critical), i.e. reducing the scale of virus reproduction to a level that is negligible from the point of view of public safety.
PROJECT OBJECTIVE
Challenge
Using an agent-based model to model the COVID-19 epidemic.
Project objective
Analysing the risk, making forecasts for the following days and weeks, providing recommendations for COVID-19 eradication. Supporting public institutions in bringing the epidemic to a state of extinction (sub-critical), i.e. reducing the scale of virus reproduction to a level that is negligible from the point of view of public safety.
TASKS FOR THE SUPERCOMPUTER
01.
Type of computing: computationally intensive tasks and queuing of parallel tasks in order to
(a) identify critical parameters describing the current state of the epidemic,
(b) simulate development of an outbreak in a proprietary programme written using Julia programming language (originally Python).
02.
User requirements: a parallel environment that allows the use of large computing power and large disk space to run a task that is solved by the author’s programme.
BENEFITS OF COOPERATION WITH WCSS
Grid simulations to find parameters and trajectories best describing the course of the epidemic so far and enabling weekly forecasts of the development of the epidemic made available in graphic form at www.mocos.pl, as well as at hdps://covid19forecasthub.eu/visualisaHon.html and hdps://kitmetricslab.github.io/forecasthub.
Enabling the results to be received on time.
A more detailed overview of the parameters.
Enabling large-scale simulation results (on a simulated population of the whole country).
Disk space for storing key results.
Software used:
qsub queuing system, an in-house
program written using Julia programming language, whose repository is public: hdps://github.com/MOCOS-COVID19/MocosSim and https://github.com/MOCOS-COVID19/MocosSimLauncher,
Bash,
Python.
EFFECTS
Publication of modelling results using the agent system on the websites www.mocos.pl, hdps://covid19forecasthub.eu/visualisaHon.html and hdps://kitmetricslab.github.io/forecasthub/.
Influence of modelling results on the decisions of experts attached to the Ministry of Health.
The highest quality scores of the model results in predicting diagnosed infections as well as deaths due to Covid-19 (for Poland) in the European Covid-19 Forecast Hub project.
Numerical analyses and algorithms are versatile enough to be applied in the lower scientific and industrial field related to the design and optimisation of mixers, e.g. paints, pharmaceuticals, building materials, food products.
Participation in the SaxoCov project for Saxony (also contribution to reports for the SaxoCov project).
Bracher, J., Wolffram, D., Deuschel, J. et al. A pre-registered short-term forecasHng study of COVID-19 in Germany and Poland during the second wave. Nat Commun 12, 5173 (2021). hdps://doi.org/10.1038/s41467-021-25207-0.
BOCK, Wolfgang, et al. Mitigation and herd immunity strategy for COVID-19 is likely to fail. medRxiv, 2020.
Awards
MOCOS was the winner in the 30 Creative Wroclawians competition hdps://pwr.edu.pl/uczelnia/aktualnosci/kreatywni-z-politechniki-wroclawskiej-11838.html.
EFFECTS
Publication of modelling results using the agent system on the websites www.mocos.pl, hdps://covid19forecasthub.eu/visualisaHon.html and hdps://kitmetricslab.github.io/forecasthub/.
Influence of modelling results on the decisions of experts attached to the Ministry of Health.
The highest quality scores of the model results in predicting diagnosed infections as well as deaths due to Covid-19 (for Poland) in the European Covid-19 Forecast Hub project.
Numerical analyses and algorithms are versatile enough to be applied in the lower scientific and industrial field related to the design and optimisation of mixers, e.g. paints, pharmaceuticals, building materials, food products.
Participation in the SaxoCov project for Saxony (also contribution to reports for the SaxoCov project).
Bracher, J., Wolffram, D., Deuschel, J. et al. A pre-registered short-term forecasting study of COVID-19 in Germany and Poland during the second wave. Nat Commun 12, 5173 (2021). hdps://doi.org/10.1038/s41467-021-25207-0.
BOCK, Wolfgang, et al. Mitigation and herd immunity strategy for COVID-19 is likely to fail. medRxiv, 2020.
Awards
MOCOS was the winner in the 30 Creative Wroclawians competition hdps://pwr.edu.pl/uczelnia/aktualnosci/kreatywni-z-politechniki-wroclawskiej-11838.html.
RECOMMENDATION
RECOMMENDATION
“First and foremost, the supercomputer is an ‘enabler’: it has enabled us to get results
on the large-scale simulated population required by the design assumptions. Without
of this infrastructure, we would not have been able to run such a large number of simultaneous for the whole of Poland.
The WCSS team has been very helpful: among other things, we could count on assistance in order to optimally run lists of related tasks through qsub, the team efficiently solved any difficulties we reported”.
Prof. Tyll Krüger, MOCOS Group,
created, among others, by researchers from the Faculty of Electronics at Wroclaw University of Science and Technology, Wrocław University of Science and Technology
