MultilayerPy (v1.0): a Python-based framework for building, running and optimising kinetic multi-layer models of aerosols and films

Kinetic multi-layer models of aerosols and films have become the state-of-the-art method of describing complex aerosol processes at the particle and film level. We present MultilayerPy: an open-source framework for building, running and optimising kinetic multi-layer models – namely the kinetic mult...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Geoscientific Model Development Jg. 15; H. 18; S. 7139 - 7151
Hauptverfasser: Milsom, Adam, Lees, Amy, Squires, Adam M., Pfrang, Christian
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Katlenburg-Lindau Copernicus GmbH 22.09.2022
Copernicus Publications
Schlagworte:
Abstracts
Aerosols
Air pollution
Algorithms
Atmospheric aerosols
Atmospheric chemistry
Bayesian analysis
Chemistry
Global optimization
Human error
Markov chains
Mathematical models
Mathematical optimization
Modelling
Monte Carlo method
Multilayers
Ordinary differential equations
Outdoor air quality
Ozone
Particle interactions
Probability theory
Random sampling
Software
Source code
Statistical inference
Statistical methods
Statistical sampling
ISSN:1991-9603, 1991-959X, 1991-962X, 1991-9603, 1991-962X
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Kinetic multi-layer models of aerosols and films have become the state-of-the-art method of describing complex aerosol processes at the particle and film level. We present MultilayerPy: an open-source framework for building, running and optimising kinetic multi-layer models – namely the kinetic multi-layer model of aerosol surface and bulk chemistry (KM-SUB) and the kinetic multi-layer model of gas–particle interactions in aerosols and clouds (KM-GAP). The modular nature of this package allows the user to iterate through various reaction schemes, diffusion regimes and experimental conditions in a systematic way. In this way, models can be customised and the raw model code itself, produced in a readable way by MultilayerPy, is fully customisable. Optimisation to experimental data using local or global optimisation algorithms is included in the package along with the option to carry out statistical sampling and Bayesian inference of model parameters with a Markov chain Monte Carlo (MCMC) sampler (via the emcee Python package). MultilayerPy abstracts the model building process into separate building blocks, increasing the reproducibility of results and minimising human error. This paper describes the general functionality of MultilayerPy and demonstrates this with use cases based on the oleic- acid–ozone heterogeneous reaction system. The tutorials in the source code (written as Jupyter notebooks) and the documentation aim to encourage users to take advantage of this tool, which is intended to be developed in conjunction with the user base.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1991-9603
1991-959X
1991-962X
1991-9603
1991-962X
DOI:10.5194/gmd-15-7139-2022