Difference between revisions of "Data Driven Modelling Special Interest Group"
Jump to navigation
Jump to search
Andre Weiner (talk | contribs) |
Andre Weiner (talk | contribs) |
||
| (10 intermediate revisions by the same user not shown) | |||
| Line 4: | Line 4: | ||
* machine learning (ML): combining CFD and ML, e.g., using ML models in CFD or deriving ML models from CFD data | * machine learning (ML): combining CFD and ML, e.g., using ML models in CFD or deriving ML models from CFD data | ||
| − | * data science: analyzing CFD data to guide modeling and decision making | + | * data science: analyzing CFD data to guide modeling and decision-making |
* data engineering: aggregating, storing, and processing CFD data | * data engineering: aggregating, storing, and processing CFD data | ||
The following list briefly summarizes our current objectives: | The following list briefly summarizes our current objectives: | ||
* short term objectives | * short term objectives | ||
| − | ** reduce the technical barrier to | + | ** reduce the technical barrier to getting started with data-driven modeling in OpenFOAM to enable more people to include data-driven workflows in their applications or research |
** promote data-driven techniques that are already available in OpenFOAM, e.g., dynamic mode decomposition (DMD) | ** promote data-driven techniques that are already available in OpenFOAM, e.g., dynamic mode decomposition (DMD) | ||
** promote third-party libraries for data-driven modeling that are based on or built for OpenFOAM | ** promote third-party libraries for data-driven modeling that are based on or built for OpenFOAM | ||
| Line 17: | Line 17: | ||
** establish tested data-driven techniques as a natural element of CFD simulations to improve accuracy and/or speed | ** establish tested data-driven techniques as a natural element of CFD simulations to improve accuracy and/or speed | ||
| − | + | We organize joint work and community events on [https://github.com/OFDataCommittee Github]. There, you will find examples of OpenFOAM-ML coupling, reduced-order modeling, Bayesian optimization, reinforcement learning, and more. | |
| − | |||
| − | |||
== Meetings == | == Meetings == | ||
| − | We | + | The SIG is currently co-chaired by [mailto:[email protected] Andre Weiner] and [mailto:[email protected] Tomislav Marić]. We organize virtual meetings every '''second week'''. We also have a Slack channel to coordinate joint projects and events. If you would like to join the channel or the meetings, please get in touch with one of the chairs. |
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | + | We also organize OpenFOAM+ML [https://github.com/OFDataCommittee/OFMLHackathon hackathons]: | |
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | + | * next hackathon: '''June 16-18, 2025, in-person''' [https://github.com/OFDataCommittee/OFMLHackathon/blob/main/ofml_june2025.md (learn more)] | |
| + | * July 16-19, 2024, virtual | ||
| + | * Nov 08-10, 2023, virtual | ||
| + | * July 24-26, 2023, virtual | ||
| + | * Jan 23-25, 2023, virtual | ||
| + | * July 23-25, 2022, virtual | ||
Latest revision as of 15:50, 17 March 2025
Scope
For us, data-driven modeling comprises a variety of different topics, some of which are listed below:
- machine learning (ML): combining CFD and ML, e.g., using ML models in CFD or deriving ML models from CFD data
- data science: analyzing CFD data to guide modeling and decision-making
- data engineering: aggregating, storing, and processing CFD data
The following list briefly summarizes our current objectives:
- short term objectives
- reduce the technical barrier to getting started with data-driven modeling in OpenFOAM to enable more people to include data-driven workflows in their applications or research
- promote data-driven techniques that are already available in OpenFOAM, e.g., dynamic mode decomposition (DMD)
- promote third-party libraries for data-driven modeling that are based on or built for OpenFOAM
- long term objectives
- aid the understanding of when and how to use data-driven modeling in the CFD workflow
- accelerate developments and applications of data-driven approaches around OpenFOAM
- establish tested data-driven techniques as a natural element of CFD simulations to improve accuracy and/or speed
We organize joint work and community events on Github. There, you will find examples of OpenFOAM-ML coupling, reduced-order modeling, Bayesian optimization, reinforcement learning, and more.
Meetings
The SIG is currently co-chaired by Andre Weiner and Tomislav Marić. We organize virtual meetings every second week. We also have a Slack channel to coordinate joint projects and events. If you would like to join the channel or the meetings, please get in touch with one of the chairs.
We also organize OpenFOAM+ML hackathons:
- next hackathon: June 16-18, 2025, in-person (learn more)
- July 16-19, 2024, virtual
- Nov 08-10, 2023, virtual
- July 24-26, 2023, virtual
- Jan 23-25, 2023, virtual
- July 23-25, 2022, virtual
