Modellansatz - English episodes only

In this episode Gudrun speaks with Nadja Klein and Moussa Kassem Sbeyti who work at the Scientific Computing Center (SCC) at KIT in Karlsruhe.
Since August 2024, Nadja has been professor at KIT leading the research group Methods for Big Data (MBD) there. She is an Emmy Noether Research Group Leader, and a member of AcademiaNet, and Die Junge Akademie, among others. In 2025, Nadja was awarded the Committee of Presidents of Statistical Societies (COPSS) Emerging Leader Award (ELA). The COPSS ELA recognizes early career statistical scientists who show evidence of and potential for leadership and who will help shape and strengthen the field. She finished her doctoral studies in Mathematics at the Universität Göttingen before conducting a postdoc at the University of Melbourne as a Feodor-Lynen fellow by the Alexander von Humboldt Foundation. Afterwards she was a Professor for Statistics and Data Science at the Humboldt-Universität zu Berlin before joining KIT.
Moussa joined Nadja's lab as an associated member in 2023 and later as a postdoctoral researcher in 2024. He pursued a PhD at the TU Berlin while working as an AI Research Scientist at the Continental AI Lab in Berlin. His research primarily focuses on deep learning, developing uncertainty-based automated labeling methods for 2D object detection in autonomous driving. Prior to this, Moussa earned his M.Sc. in Mechatronics Engineering from the TU Darmstadt in 2021.
The research of Nadja and Moussa is at the intersection of statistics and machine learning. In Nadja's MBD Lab the research spans theoretical analysis, method development and real-world applications. One of their key focuses is Bayesian methods, which allow to incorporate prior knowledge, quantify uncertainties, and bring insights to the “black boxes” of machine learning. By fusing the precision and reliability of Bayesian statistics with the adaptability of machine and deep learning, these methods aim to leverage the best of both worlds. The KIT offers a strong research environment, making it an ideal place to continue their work. They bring new expertise that can be leveraged in various applications and on the other hand Helmholtz offers a great platform in that respect to explore new application areas. For example Moussa decided to join the group at KIT as part of the Helmholtz Pilot Program Core-Informatics at KIT (KiKIT), which is an initiative focused on advancing fundamental research in informatics within the Helmholtz Association. Vision models typically depend on large volumes of labeled data, but collecting and labeling this data is both expensive and prone to errors. During his PhD, his research centered on data-efficient learning using uncertainty-based automated labeling techniques. That means estimating and using the uncertainty of models to select the helpful data samples to train the models to label the rest themselves. Now, within KiKIT, his work has evolved to include knowledge-based approaches in multi-task models, eg. detection and depth estimation — with the broader goal of enabling the development and deployment of reliable, accurate vision systems in real-world applications.
Statistics and data science are fascinating fields, offering a wide variety of methods and applications that constantly lead to new insights. Within this domain, Bayesian methods are especially compelling, as they enable the quantification of uncertainty and the incorporation of prior knowledge. These capabilities contribute to making machine learning models more data-efficient, interpretable, and robust, which are essential qualities in safety-critical domains such as autonomous driving and personalized medicine. Nadja is also enthusiastic about the interdisciplinarity of the subject — repeatedly changing the focus from mathematics to economics to statistics to computer science. The combination of theoretical fundamentals and practical applications makes statistics an agile and important field of research in data science.
From a deep learning perspective, the focus is on making models both more efficient and more reliable when dealing with large-scale data and complex dependencies. One way to do this is by reducing the need for extensive labeled data. They also work on developing self-aware models that can recognize when they're unsure and even reject their own predictions when necessary. Additionally, they explore model pruning techniques to improve computational efficiency, and specialize in Bayesian deep learning, allowing machine learning models to better handle uncertainty and complex dependencies. Beyond the methods themselves, they also contribute by publishing datasets that help push the development of next-generation, state-of-the-art models. The learning methods are applied across different domains such as object detection, depth estimation, semantic segmentation, and trajectory prediction — especially in the context of autonomous driving and agricultural applications. As deep learning technologies continue to evolve, they’re also expanding into new application areas such as medical imaging.
Unlike traditional deep learning, Bayesian deep learning provides uncertainty estimates alongside predictions, allowing for more principled decision-making and reducing catastrophic failures in safety-critical application. It has had a growing impact in several real-world domains where uncertainty really matters. Bayesian learning incorporates prior knowledge and updates beliefs as new data comes in, rather than relying purely on data-driven optimization. In healthcare, for example, Bayesian models help quantify uncertainty in medical diagnoses, which supports more risk-aware treatment decisions and can ultimately lead to better patient outcomes. In autonomous vehicles, Bayesian models play a key role in improving safety. By recognizing when the system is uncertain, they help capture edge cases more effectively, reduce false positives and negatives in object detection, and navigate complex, dynamic environments — like bad weather or unexpected road conditions — more reliably. In finance, Bayesian deep learning enhances both risk assessment and fraud detection by allowing the system to assess how confident it is in its predictions. That added layer of information supports more informed decision-making and helps reduce costly errors. Across all these areas, the key advantage is the ability to move beyond just accuracy and incorporate trust and reliability into AI systems.
Bayesian methods are traditionally more expensive, but modern approximations (e.g., variational inference or last layer inference) make them feasible. Computational costs depend on the problem — sometimes Bayesian models require fewer data points to achieve better performance. The trade-off is between interpretability and computational efficiency, but hardware improvements are helping bridge this gap.
Their research on uncertainty-based automated labeling is designed to make models not just safer and more reliable, but also more efficient. By reducing the need for extensive manual labeling, one improves the overall quality of the dataset while cutting down on human effort and potential labeling errors. Importantly, by selecting informative samples, the model learns from better data — which means it can reach higher performance with fewer training examples. This leads to faster training and better generalization without sacrificing accuracy. They also focus on developing lightweight uncertainty estimation techniques that are computationally efficient, so these benefits don’t come with heavy resource demands. In short, this approach helps build models that are more robust, more adaptive to new data, and significantly more efficient to train and deploy — which is critical for real-world systems where both accuracy and speed matter.
Statisticians and deep learning researchers often use distinct methodologies, vocabulary and frameworks, making communication and collaboration challenging. Unfortunately, there is a lack of Interdisciplinary education: Traditional academic programs rarely integrate both fields. It is necessary to foster joint programs, workshops, and cross-disciplinary training can help bridge this gap.
From Moussa's experience coming through an industrial PhD, he has seen how many industry settings tend to prioritize short-term gains — favoring quick wins in deep learning over deeper, more fundamental improvements.
To overcome this, we need to build long-term research partnerships between academia and industry — ones that allow for foundational work to evolve alongside practical applications. That kind of collaboration can drive more sustainable, impactful innovation in the long run, something we do at methods for big data.
Looking ahead, one of the major directions for deep learning in the next five to ten years is the shift toward trustworthy AI. We’re already seeing growing attention on making models more explainable, fair, and robust — especially as AI systems are being deployed in critical areas like healthcare, mobility, and finance. The group also expect to see more hybrid models — combining deep learning with Bayesian methods, physics-based models, or symbolic reasoning. These approaches can help bridge the gap between raw performance and interpretability, and often lead to more data-efficient solutions. Another big trend is the rise of uncertainty-aware AI. As AI moves into more high-risk, real-world applications, it becomes essential that systems understand and communicate their own confidence. This is where uncertainty modeling will play a key role — helping to make AI not just more powerful, but also more safe and reliable.
The lecture "Advanced Bayesian Data Analysis" covers fundamental concepts in Bayesian statistics, including parametric and non-parametric regression, computational techniques such as MCMC and variational inference, and Bayesian priors for handling high-dimensional data. Additionally, the lecturers offer a Research Seminar on Selected Topics in Statistical Learning and Data Science.
The workgroup offers a variety of Master's thesis topics at the intersection of statistics and deep learning, focusing on Bayesian modeling, uncertainty quantification, and high-dimensional methods. Current topics include predictive information criteria for Bayesian models and uncertainty quantification in deep learning. Topics span theoretical, methodological, computational and applied projects. Students interested in rigorous theoretical and applied research are encouraged to explore our available projects and contact us for further details.
The general advice of Nadja and Moussa for everybody interested to enter the field is: "Develop a strong foundation in statistical and mathematical principles, rather than focusing solely on the latest trends.
Gain expertise in both theory and practical applications, as real-world impact requires a balance of both. Be open to interdisciplinary collaboration. Some of the most exciting and meaningful innovations happen at the intersection of fields — whether that’s statistics and deep learning, or AI and domain-specific areas like medicine or mobility. So don’t be afraid to step outside your comfort zone, ask questions across disciplines, and look for ways to connect different perspectives. That’s often where real breakthroughs happen. With every new challenge comes an opportunity to innovate, and that’s what keeps this work exciting. We’re always pushing for more robust, efficient, and trustworthy AI. And we’re also growing — so if you’re a motivated researcher interested in this space, we’d love to hear from you."

Literature and further information
Webpage of the group
G. Nuti, Lluis A.J. Rugama, A.-I. Cross: Efficient Bayesian Decision Tree Algorithm, arxiv Jan 2019
Wikipedia: Expected value of sample information
C. Howson & P. Urbach: Scientific Reasoning: The Bayesian Approach (3rd ed.). Open Court Publishing Company. ISBN 978-0-8126-9578-6, 2005.
A.Gelman e.a.: Bayesian Data Analysis Third Edition. Chapman and Hall/CRC. ISBN 978-1-4398-4095-5, 2013.
Yu, Angela: Introduction to Bayesian Decision Theory cogsci.ucsd.edu, 2013.
Devin Soni: Introduction to Bayesian Networks, 2015.
G. Nuti, L. Rugama, A.-I. Cross: Efficient Bayesian Decision Tree Algorithm, arXiv:1901.03214 stat.ML, 2019.
M. Carlan, T. Kneib and N. Klein: Bayesian conditional transformation models, Journal of the American Statistical Association, 119(546):1360-1373, 2024.
N. Klein: Distributional regression for data analysis , Annual Review of Statistics and Its Application, 11:321-346, 2024
C.Hoffmann and N.Klein: Marginally calibrated response distributions for end-to-end learning in autonomous driving, Annals of Applied Statistics, 17(2):1740-1763, 2023
Kassem Sbeyti, M., Karg, M., Wirth, C., Klein, N., & Albayrak, S. (2024, September). Cost-Sensitive Uncertainty-Based Failure Recognition for Object Detection. In Uncertainty in Artificial Intelligence (pp. 1890-1900). PMLR.
M. K. Sbeyti, N. Klein, A. Nowzad, F. Sivrikaya and S. Albayrak: Building Blocks for Robust and Effective Semi-Supervised Real-World Object Detection pdf. To appear in Transactions on Machine Learning Research, 2025
Podcasts
Learning, Teaching, and Building in the Age of AI Ep 42 of Vanishing Gradient, Jan 2025.
O. Beige, G. Thäter: Risikoentscheidungsprozesse, Gespräch im Modellansatz Podcast, Folge 193, Fakultät für Mathematik, Karlsruher Institut für Technologie (KIT), 2019.

Gudrun talks with Polyxeni Spilioti at Aarhus university about spectral geometry.
Before working in Aarhus Polyxeni was a postdoctoral researcher in the group of Anton Deitmar at the University of Tübingen. She received her PhD from the University of Bonn, under the supervision of Werner Mueller after earning her Master's at the National and Technical University of Athens (Faculty of Applied Mathematics and Physics).
As postdoc she was also guest at the MPI for Mathematics in Bonn, the Institut des Hautes Etudes Scientifiques in Paris and the Oberwolfach Research Institute for Mathematics.
In her research she works on questions like: How can one obtain information about the geometry of a manifold, such as the volume, the curvature, or the length of the closed geodesics, provided that we can study the spectrum of certain differential operators? Harmonic analysis on locally symmetric spaces provides a powerful machinery in studying various invariants, such as the analytic torsion, as well as the dynamical zeta functions of Ruelle and Selberg.



References and further information
P. Spiliotti: Ruelle and Selberg zeta functions on compact hyperbolic odd dimensional manifolds PhD thesis, Bonn, 2015.
Greek Women in Mathematics Website
Celebration of Greek Women in mathematics, May 12 2022
Greek women in mathematics - First podcast episode
Eberhard Zeidler on Wikipedia



Podcasts
A. Pohl: Quantenchaos, Gespräch mit G. Thäter im Modellansatz Podcast, Folge 79, Fakultät für Mathematik, Karlsruher Institut für Technologie (KIT), 2016.

One of the reasons we started this podcast in 2013 was to provide a more realistic picture of mathematics and of the way mathematicians work. On Nov. 19 2021 Gudrun talked to Stephanie Anne Salomone who is Professor and Chair in Mathematics at the University of Portland. She is also Director of the STEM Education and Outreach Center and Faculty Athletic Representative at UP. She is an Associate Director of Project NExT, a program of the Mathematical Association of America that provides networking and professional development opportunities to mathematics faculty who are new to our profession. She is a wife and mother of three boys, Milo (13), Jude (10), and Theodore (8).
This conversation started on Twitter in the summer of 2021. There Stephanie (under the twitter handle @SitDownPee) and @stanyoshinobu Dr. Stan Yoshinobu invited their fellow mathematicians to the following workshop: Come help us build gender equity in mathematics! Picture a Mathematician workshop led by @stanyoshinobu Dr. Stan Yoshinobu and me, designed for men in math, but all genders welcome. Gudrun was curious to learn more and followed the provided link:

Workshop Abstract
Gender equity in the mathematical sciences and in the academy broadly is not yet a reality. Women (and people of color, and other historically excluded groups) are confronted with systemic biases, daily experiences, feelings of not being welcome or included, that in the aggregate push them out of the mathematical sciences. This workshop is designed primarily for men in math (although all genders are welcome to participate) to inform and inspire them to better see some of the key issues with empathy, and then to take action in creating a level-playing field in the academy.
Workshop activities include viewing “Picture a Scientist” before the workshop, a 2-hour synchronous workshop via zoom, and follow-up discussions via email and Discord server. *All genders welcome AND this workshop is designed for men to be allies.

This idea resonated strongly with Gudrun's experiences: Of course women and other groups which are minorities in research have to speak out to fight for their place but things move forward only if people with power join the cause. At the moment people with power in mathematical research mostly means white men. That is true for the US where Stephanie is working as well as in Germany. Allyship is a concept which was introduced by people of colour to name white people fighting for racial justice at their side. Of course, it is a concept which helps in all situations where a group is less powerful than another. Men working for the advancement of non-male mathematicians is strictly necessary in order for equality of chances and a diversity of people in mathematics to be achieved in the next generation. And to be clear: this has nothing to do with counting heads but it is about not ruining the future of mathematics as a discipline by creating obstacles for mathematicians with minoritized identities.
The important question is: How is it possible to educate men and especially powerful white men to become allies?
The idea of this first workshop designed by Stephanie and Stan was to invite men already interested in learning more and to build a basis with the documentary Picture a scientist (2020).

SYNOPSIS
PICTURE A SCIENTIST chronicles the groundswell of researchers who are writing a new chapter for women scientists. Biologist Nancy Hopkins, chemist Raychelle Burks, and geologist Jane Willenbring lead viewers on a journey deep into their own experiences in the sciences, ranging from brutal harassment to years of subtle slights. Along the way, from cramped laboratories to spectacular field stations, we encounter scientific luminaries - including social scientists, neuroscientists, and psychologists - who provide new perspectives on how to make science itself more diverse, equitable, and open to all. (from the webpage)
In this film there are no mathematicians, but the situations in sciences and mathematics are very similar and for that it lends itself to show the situation.
In the podcast conversation Gudrun and Stephanie talk about why and in what way the documentary spoke to them. The huge and small obstacles in their own work as women mathematicians which do not make them feel welcome in a field they feel passionate about. The film shows what happens to women in Science. It shows also men in different roles. Obviously there are the bullies. Then there are the bystanders. There are universities which allow women to be hired and give them the smallest space available. But there are also men who consider themselves friends of their female collaegues who cannot believe that they did not notice how the behaviour of other men (and their own behavior in not taking a side). Seeing this play out over the course of the film is not a comfortable watch, and perhaps because of this discomfort, we hope to build empathy.
On the other hand, there is a story of women scientists who noticed that they were not treated as well as their male colleagues and who found each other to fight for office space and the recognition of their work. They succeded a generation ago.
The general idea of the workshop was to start with the documentary and to talk about different people and their role in the film in order to take them as prototypical for roles which we happen to observe in our life and which we might happen to play. This discussion in groups was moderated and guided in order to make this a safe space for everyone.
Stephanie spoke about how we have to let men grow into their responsibility to speak out against a hostile atmosphere at university created mostly by men. In the workshop it was possible to first develop and then train for possible responses in situations which ask for men stepping in as an ally.
The next iteration of the workshop Picture a Mathematician will be on May 11.

Biography: Stephanie Salomone earned her Ph.D. in Mathematics from UCLA in 2005 and joined the faculty at the University of Portland that year. She serves as Professor and Chair of Mathematics and Director of the STEM Education and Outreach Center at UP, as well as the Faculty Athletic Representative. She is an Associate Director of Project NExT, a national professional development program for new higher-education mathematics faculty. She was the PI on the NSF REFLECT program, advancing the use of evidence-based practices in STEM teaching at UP and the use of peer-observation for formative assessment of teaching, and has managed a combined $1.6 million as the PI on a subaward of the Western Regional Noyce Alliance grant and as PI of the NSF Noyce Program at UP. She is on the Board of Directors for Saturday Academy, a local 501c3 whose mission is to engage children in hands-on STEM learning. Dr. Salomone is the recipient of UP’s 2009 Outstanding Teaching Award and the recipient of the 2019 Oregon Academy of Sciences Outstanding Educator in STEM Higher Education Award.


Literature and further information
Allyship: What It Means to Be an Ally, Tulane university, School of social work
Guide to allyship
Ernest, Reinholz, and Shah: Hidden Competence: women’s mathematical participation in public and private classroom spaces, Educ Stud Math 102, 153–172 (2019). https://doi.org/10.1007/s10649-019-09910-w
J.R. Cimpian, T.H. Kimand, Z.T. McDermott: Understanding persistent gender gaps in STEM,
Science 368, Issue 6497, 1317-1319 (2020). https://doi.org/10.1126/science.aba7377
S.J. Ceci and W.M. Williams: Understanding current causes of women’s underrepresentation in science PNAS 108 3157–3162 (2011). https://doi.org/10.1073/pnas.1014871108
Inquirybased learning site
Equatiy and teaching math Blog post by Stan Yoshinobu




Podcasts
Mathematically uncensored Podcast

In March 2018 Gudrun had a day available in London when travelling back from the FENICS workshop in Oxford. She contacted a few people working in mathematics at the University College London (ULC) and asked for their time in order to talk about their research. In the end she brought back three episodes for the podcast. This is the second of these conversations.
Gudrun talks to Marta Betcke. Marta is associate professor at the UCL Department of Computer Science, member of Centre for Inverse Problems and Centre for Medical Image Computing. She has been in London since 2009. Before that she was a postdoc in the Department of Mathematics at the University of Manchester working on novel X-ray CT scanners for airport baggage screening.
This was her entrance into Photoacoustic tomography (PAT), the topic Gudrun and Marta talk about at length in the episode. PAT is a way to see inside objects without destroying them. It makes images of body interiors. There the contrast is due to optical absorption, while the information is carried to the surface of the tissue by ultrasound. This is like measuring the sound of thunder after lightning. Measurements together with mathematics provide ideas about the inside. The technique combines the best of light and sound since good contrast from optical part - though with low resolution - while ultrasound has good resolution but poor contrast (since not enough absorption is going on).
In PAT, the measurements are recorded at the surface of the tissue by an array of ultrasound sensors. Each of that only detects the field over a small volume of space, and the measurement continues only for a finite time. In order to form a PAT image, it is necessary to solve an inverse initial value problem by inferring an initial acoustic pressure distribution from measured acoustic time series. In many practical imaging scenarios it is not possible to obtain the full data, or the data may be sub-sampled for faster data acquisition. Then numerical models of wave propagation can be used within the variational image reconstruction framework to find a regularized least-squares solution of an optimization problem.
Assuming homogeneous acoustic properties and the absence of acoustic absorption the measured time series can be related to the initial pressure distribution via the spherical mean Radon transform. Integral geometry can be used to derive direct, explicit inversion formulae for certain sensor geometries, such as e.g. spherical arrays.
At the moment PAT is predominantly used in preclinical setting, to image tomours and vasculature in small animals. Breast imaging, endoscopic fetus imaging as well as monitoring of perfusion and drug metabolism are subject of intensive ongoing research.
The forward problem is related to the absorption of the light and modeled by the wave equation assuming instanteneous absorption and the resulting thearmal expansion. In our case, an optical ultrasound sensor records acoustic waves over time, i.e. providing time series with desired spacial and temporal resolution. Given complete data, then one can mathematically reverse the time direction and find out the original object.
Often it is not possible to collect a complete data due to e.g. single sided access to the object as in breast imaging or underlying dynamics happening on a faster rate than one can collect data. In such situations one can formulate the problem in variational framework using regularisation to compensate for the missing data.
In particular in subsampling scenario, one would like to use raytracing methods as they scale linearly with the number of sensors. Marta's group is developing flexible acoustic solvers based on ray tracing discretisation of the Green's formulas. They cannot handle reflections but it is approximately correct to assume this to be true as the soundspeed variation is soft tissue is subtle. These solvers can be deployed alongside with stochastic iterative solvers for efficient solution of the variational formulation.
Marta went to school in Poland. She finished her education there in a very selected school and loved math due to a great math teacher (which was also her aunt). She decidede to study Computer Sciences, since there she saw more chances on the job market. When moving to Germany her degree was not accepted, so she had to enrol again. This time for Computer Sciences and Engineering at the Hamburg University of Technology. After that she worked on her PhD in the small group of Heinrich Voss there. She had good computing skills and fit in very well. When she finished there she was married and had to solve a two body problem, which brought the couple to Manchester, where a double position was offered.
Now both have a permanent position in London.



References
M. Betcke e.a.: Model-Based Learning for Accelerated, Limited-View 3-D Photoacoustic Tomography IEEE Transactions on Medical Imaging 37, 1382 - 1393, 2018.
F. Rullan & M. Betcke: Hamilton-Green solver for the forward and adjoint problems in photoacoustic tomography archive, 2018.
M. Betcke e.a.: On the adjoint operator in photoacoustic tomography Inverse Problems 32, 115012, 2016. doi
C. Lutzweiler and D. Razansky: Optoacoustic imaging and tomography - reconstruction approaches and outstanding challenges in image performance and quantification, Sensors 13 7345, 2013. doi: 10.3390/s130607345



Podcasts
G. Thäter, K. Page: Embryonic Patterns, Gespräch im Modellansatz Podcast, Folge 161, Fakultät für Mathematik, Karlsruher Institut für Technologie (KIT), 2018.
F. Cakoni, G. Thäter: Linear Sampling, Conversation im Modellansatz Podcast, Episode 226, Department of Mathematics, Karlsruhe Institute of Technology (KIT), 2019.
G. Thäter, R. Aceska: Dynamic Sampling, Gespräch im Modellansatz Podcast, Folge 173, Fakultät für Mathematik, Karlsruher Institut für Technologie (KIT), 2018.
S. Fliss, G. Thäter: Transparent Boundaries. Conversation in the Modellansatz Podcast episode 75, Department of Mathematics, Karlsruhe Institute of Technology (KIT), 2015.
S. Hollborn: Impedanztomographie. Gespräch mit G. Thäter im Modellansatz Podcast, Folge 68, Fakultät für Mathematik, Karlsruher Institut für Technologie (KIT), 2015.
M. Kray, G. Thäter: Splitting Waves. Conversation in the Modellansatz Podcast episode 62, Department of Mathematics, Karlsruhe Institute of Technology (KIT), 2015.
F. Sayas, G. Thäter: Acoustic scattering. Conversation in the Modellansatz Podcast episode 58, Department of Mathematics, Karlsruhe Institute of Technology (KIT), 2015.

This is the third of three conversation recorded during the Conference on mathematics of wave phenomena 23-27 July 2018 in Karlsruhe.

Gudrun is in conversation with Anne-Sophie Bonnet-BenDhia from ENSTA in Paris about transmission properties in perturbed waveguides.
The spectral theory is essential to study wave phenomena. For instance, everybody has experimented with resonating frequencies in a bathtube filled with water. These resonant eigenfrequencies are eigenvalues of some operator which models the flow behaviour of the water. Eigenvalue problems are better known for matrices. For wave problems, we have to study eigenvalue problems in infinite dimension. Like the eigenvalues for a finite dimensional matrix the Spectral theory gives access to intrinisic properties of the operator and the corresponding wave phenomena.
Anne-Sophie is interested in waveguides. For example, optical fibres can guide optical waves while wind instruments are guides for acoustic waves. Electromagnetic waveguides also have important applications.

A practical objective is to optimize the transmission in a waveguide, even if there are some perturbations inside. It is known that for certain frequencies, there is no reflection by the perturbations but it is not apriori clear how to find these frequencies.
Anne-Sophie uses complex analysis for that. The idea is to complexify the (originally real) coordinates by analytic extension. It is a classic idea for resonances that she adapts to the problem of transmission.

This mathematical method of complex scaling is linked to the method of perfectly matched layers in numerics. It is used to solve problems set in unbounded domains on a computer by finite elements. Thanks to the complex scaling, she can solve a problem in a bounded domain, which reproduces the same behaviour as in the infinite domain.
Finally, Anne-Sophie is able to get numerically a complex spectrum of frequencies, related to the quality of the transmission in a perturbed waveguide. The imaginary part of the complex quantity gives an indication of the quality of the transmission in the waveguide. The closer to the real axis the better the transmission.



References
A-S. Bonnet-BenDhia, L. Chesnel and V. Pagneux:Trapped modes and reflectionless modes as eigenfunctions of the same spectral problem Proceedings of the Royal Society A, 2018, doi 10.1098/rspa.2018.0050
A-S. Bonnet-BenDhia: Mathematical and numerical treatment of plasmonic waves at corners of metals and metamaterials Emerging Topics in Optics, IMA, Minneapolis, 2017
A-S. Bonnet-BenDhia, L. Chesnel and S. Nazarov: Perfect transmission invisibility for waveguides with sound hard walls Journal de Mathématiques Pures et Appliquées, 2017, doi 10.1016/j.matpur.2017.07.020
A.-S. Bonnet-BenDhia e.a.: A method to build non-scattering perturbations of two-dimensional acoustic waveguides Math. meth. appl. sci., vol. 40, pp. 335–349, 2015 doi 10.1002/mma.3447



Podcasts
S. Fliss, G. Thäter: Transparent Boundaries. Conversation in the Modellansatz Podcast episode 75, Department of Mathematics, Karlsruhe Institute of Technology (KIT), 2015.
M. Kray, G. Thäter: Splitting Waves. Conversation in the Modellansatz Podcast episode 62, Department of Mathematics, Karlsruhe Institute of Technology (KIT), 2015.
F. Sayas, G. Thäter: Acoustic scattering. Conversation in the Modellansatz Podcast episode 58, Department of Mathematics, Karlsruhe Institute of Technology (KIT), 2015.