Hybrid Simulations and Learning MethodsModule description #51120 / Version 1
- 6 CP
- Written exam
- Graded
- Winter semester
- English
This module description is not valid in the current semester (Winter semester 2026/27).
Current version (v2)
| Module / Version | #51120 / #1 |
|---|---|
| Validity | Winter semester 2024/25 – Summer semester 2025 |
| Available language(s) | English |
| Credit points | 6 |
| Person responsible for the module | Stender, Merten |
| Grading | Graded |
| Exam format | Written exam |
| Language(s) of instruction | English |
Affiliation
| Faculty | Fakultät V |
|---|---|
| Institute | Institut für Maschinenkonstruktion und Systemtechnik (IMS) |
| Academic chair | 35352200 FG Cyber-Physical-Systems in Mechanical Engineering |
Contact
| Office | H 66 |
|---|---|
| Contact person | Stender, Merten |
| E-mail address | merten.stender@tu-berlin.de |
| Website | http://www.tu.berlin/cpsme |
Learning outcomes
Students will first study the fundamental modeling approaches in physics-based simulations (built on the discretization of governing first principles) and data-driven models (built on the universal approximation theorem). Based on the underlying assumptions, constraints, and individual advantages, students will learn about hybrid approaches that combine physics-based and data-driven models at different levels of integration and penetration. The theoretical input is accompanied by in-depth and hands-on implementations in Python, targeting synthetical and experimental data. After participating in this class, students can make an informed choice on modeling approaches in situations of limited physical understanding and/or limited access to data. The course has a strong research focus on the latest cutting-edge scientific findings.
Content
- Numerical discretization techniques for ordinary and partial differential equations
- Basics of deep learning: universal approximation theorem, bias-variance-tradeoff, error backpropagation
- Automatic differentiation and differentiable models
- Data-driven discovery of governing equations from data: compressive sensing; sparse identification of nonlinear dynamics (SINDy, weak and strong)
- Constraining deep learning methods: soft and hard constraints
- Principles of physics-informed neural networks (PINNs)
- Advanced optimization and training strategies
- Neural operators and deep operator networks (DeepONet)
- Neural differential equations and universal differential equations
- Graph neural networks for graph-structured numerical computation meshes
- Geometry-informed neural networks (GINNs)
- Applications in science and engineering
- Basics of deep learning: universal approximation theorem, bias-variance-tradeoff, error backpropagation
- Automatic differentiation and differentiable models
- Data-driven discovery of governing equations from data: compressive sensing; sparse identification of nonlinear dynamics (SINDy, weak and strong)
- Constraining deep learning methods: soft and hard constraints
- Principles of physics-informed neural networks (PINNs)
- Advanced optimization and training strategies
- Neural operators and deep operator networks (DeepONet)
- Neural differential equations and universal differential equations
- Graph neural networks for graph-structured numerical computation meshes
- Geometry-informed neural networks (GINNs)
- Applications in science and engineering
Module components
Compulsory area
The following courses are mandatory for this module:
| Courses | Type | Number | Frequency | Language | Hours per week | Course catalog |
|---|---|---|---|---|---|---|
| Hybrid Simulations and Learning Methods | Lecture | Winter semester | [en] | 2 | WS 2026/27 (1) ISIS | |
| Hybrid Simulations and Learning Methods | Exercise | Winter semester | [en] | 2 | WS 2026/27 (1) ISIS |
Courses in the course catalog
Courses created from the teaching units of this module.
Winter semester 2026/27 (2)
Hybrid Simulations and Learning Methods (Lecture) 1
- Hybrid Simulations and Learning Methods (Vorlesung) Termingruppe 1
Hybrid Simulations and Learning Methods (Exercise) 1
- Hybrid Simulations and Learning Methods (Übung) Termingruppe 1
Workload and credit points
Hybrid Simulations and Learning Methods (Lecture)
| Workload description | Multiplier | Hours | Total |
|---|---|---|---|
| Attendance | 15 | 2 h | 30 h |
| Pre/post processing | 15 | 4 h | 60 h |
| 90 h | |||
Hybrid Simulations and Learning Methods (Exercise)
| Workload description | Multiplier | Hours | Total |
|---|---|---|---|
| Attendance | 15 | 2 h | 30 h |
| Pre/post processing | 15 | 4 h | 60 h |
| 90 h | |||
The workload of the module adds up to 180 hours. The module therefore comprises 6 credit points.
Description of teaching and learning methods
Lecture (theory) and exercise (practical programming)
Requirements for participation and examination
Desirable prerequisites for participation in the courses:
No formal prerequisites. Prior knowledge of neural networks and profound programming skills in Python will be very helpful.
Mandatory requirements for the module examination registration:
This module has no examination requirements.
Completion of the module
| Grading | Graded |
|---|---|
| Exam format | Written exam |
| Language | English |
| Duration/Extent | E-Exam (incl. programming tasks). 90 min duration. |
Duration of the module
The following number of semesters is estimated for taking and completing the module: 1 semester(s).
This module can be started in the following semesters: Winter semester.
Maximum number of participants
This module is not limited to a number of students.
Registration procedures
The course is organized through ISIS. Exam registration through MOSES.
Recommended reading, lecture notes
Lecture notes
unavailable
Electronic lecture notes
available
Recommended literature
No recommended literature given
Assigned programs
This module version is used in the following programs:
| Program / program version | Semester(s) | Curriculum component |
|---|---|---|
| Computational Engineering Science (Informationstechnik im Maschinenwesen) StuPO 2018 (17.01.2018) | SS 2025 | 2. Kernbereich: Simulation und Optimierung |
| Engineering Science StuPO 2007 (19.12.2007) | SS 2025 | 2.1 Numerik und Simulation › 2.1a Kernbereich |
| Engineering Science StuPO 2020 | SS 2025 | 2.1 Numerik und Simulation › 2.1a Kernbereich |
Miscellaneous
Not specified