2024/25 Taught Postgraduate Module Catalogue

COMP5455M Data-Driven Fluid Dynamics

15 creditsClass Size: 30

Module manager: Prof Peter Jimack
Email: p.k.jimack@leeds.ac.uk

Taught: Semester 2 (Jan to Jun) View Timetable

Year running 2024/25

Pre-requisite qualifications

Experience in programming.

Pre-requisites

MATH5453M

Foundations of Fluid Dynamics

Module replaces

COMP5454M Fluid-Structure Interactions

This module is not approved as an Elective

Module summary

Machine learning (ML) techniques are becoming ever more important in fluid dynamics. This module, which is aimed at students specialising in fluid dynamics, will provide an introduction to some of the key concepts of ML within this context.

Objectives

The overall aim of this module is to provide students who already have Masters’ level expertise in fluid dynamics (mathematical and numerical models, computational simulation and experimental techniques) with an introduction to ML, with a specific emphasis on those techniques that may be used most naturally with fluid dynamics problems.

The objectives are to provide students with:
• A high-level overview of common ML approaches and their strengths and their limitations.
• Hands-on experience of training, testing and modifying ML models using one or more standard python frameworks.
• The opportunity to explore a number of in-depth case studies based upon popular ML techniques in fluid dynamics.
• Underpinning mathematical and statistical tools to allow them to independently learn about ML techniques not covered by this module.

Learning outcomes
On successful completion of the module students will have demonstrated the following learning outcomes relevant to the subject:
1. Awareness and high-level understanding of a wide range of common ML approaches used in fluid dynamics, and critical evaluation of their strengths and their limitations (SS1).
2. Knowledge and critical awareness of classical ML approaches such as principal component analysis (PCA), Gaussian process regression, support vector machines and random forest (SS2)
3. Knowledge and critical awareness of ML and deep learning (DL) techniques based upon artificial neural networks (ANNs), including for supervised, unsupervised and semi-supervised learning (SS3).
4. Understanding of training via stochastic gradient descent and the phenomenon of overfitting, with the ability to make informed judgements concerning model training and testing (SS4).
5. Knowledge of statistical tools and techniques that can aid the understanding and critical evaluation of ML applied to problems in fluid dynamics (SS5).
6. In-depth knowledge of the application of 5 different techniques from ML to practical problems in fluid dynamics (SS6).

Syllabus

There will be an introduction to machine learning concepts including:
· Supervised and unsupervised learning
· Regression
· Principal Component Analysis
· Artificial neural networks
· Stochastic gradient descent
· The breadth of possible applications in fluid dynamics
Self-study to become familiar with the use of PyTorch (or an alternative ML framework)
There will be 5 mini-projects selected from a range of contemporary topics in data-driven fluid dynamics, such as:
· Deep learning for flow classification and feature recognition
· Auto-encoders and reduced order modelling
· Model reconstruction
· Convolution neural networks
· Physics-informed neural networks
· Fourier neural operators

Teaching methods

Delivery type	Number	Length hours	Student hours
Workshop	5	3.00	15.00
Lecture	25	1.00	25.00
Independent online learning hours			85.00
Private study hours			25.00
Total Contact hours			40.00
Total hours (100hr per 10 credits)			150.00

Opportunities for Formative Feedback

There will be some short-form questions associated with the 10 introductory lectures that the students will be able to complete and receive formative feedback on.
There will be a one-week gap between the first assignment and the start of the second assignment to allow the students to receive formative (as well as summative) feedback on the way in which they completed the assignment.

Methods of assessment

Coursework

Assessment type	Notes	% of formal assessment
In-course Assessment	Coursework Portfolio of 5 specialist applications to assess practical application of knowledge and skills	100.00
Total percentage (Assessment Coursework)		100.00

This module will be reassessed by coursework.

Reading list

The reading list is available from the Library website

Last updated: 25/09/2024 09:18:38

Disclaimer

Browse Other Catalogues

• Undergraduate module catalogue
• Taught Postgraduate module catalogue
• Undergraduate programme catalogue
• Taught Postgraduate programme catalogue

Errors, omissions, failed links etc should be notified to the Catalogue Team.PROD