一本道

Måndag 27 april 2026, 14.15-15.00, Aditya Ghosh (IFM)

Titel: Data-Driven Surrogates for Complex Flows: Reduced-Order Models, Transformers, and CVD Applications

Sammanfattning: My research lies at the intersection of machine learning and computational fluid dynamics. A nonlinear reduced-order modelling framework, developed as part of a master’s thesis in Aeronautical Engineering, is developed. A beta-variational autoencoder extracts nonlinear modes from turbulent flow data analogous to POD modes but capable of capturing nonlinear structures. These latent representations are fed into a transformer architecture for temporal dynamics prediction, demonstrating the potential of combining deep generative models with attention based sequence prediction for flow forecasting.

Building on this data-driven foundation, my ongoing PhD project focuses on CFD modelling of a chemical vapour deposition (CVD) reactor using a 2nd order solver. The reactor operates in a low-Reynolds number regime with strong thermal coupling: an inlet jet, initially turbulent, undergoes rapid relaminarisation driven by intense heating within the reactor chamber. Understanding this transition and its impact on the internal flow field is a key focus of the ongoing work. Future directions include moving toward DNS using Neko and investigating how the resulting flow structures influence chemical precursor decomposition and deposition uniformity.

These experiences motivate the need for mathematically reliable and uncertainty-aware surrogate models that can advance predictive capabilities for complex, multi-physics CFD applications.

Onsdag 10 december 2025, 13.15-14.15, Mårten Wadenbäck, ISY

Titel: Radially Distorted Homographies, Revisited

Sammanfattning: Homographies are an important class of transformations in projective geometry and computer vision, but real images often suffer from lens distortions, making joint estimation of homography and distortion essential. I will start with a quick look at homogeneous coordinates and homographies, and then move on to present a unified approach for homography estimation that handles three configurations of radial distortion: in one image, identical in both, or independent in both. The approach leads to fast, stable, and accurate minimal solvers for all three cases, outperforming existing state-of-the-art in speed while maintaining accuracy.

Anordnas i samarbete med Matematiska kollokviet

Måndag 6 oktober 2025, 13.15-15.00, Minisymposium on Imaging

Talare: Jan Rolfes, MAI

Titel: Utilizing Material Properties in the 3D Reconstruction of Functional Materials via Nanotomography

Sammanfattning: Nanotomography techniques such as electron tomography (ET) and nano X-ray computed tomography (nano-CT) allow high-resolution 3D imaging of functional materials (e.g. catalyst supports, battery electrodes or photonic crystals) and the determination of important material properties such as particle and pore size, position and interconnectivity, or chemical composition.

However, tomography always comes along the challenge of the proper acquisition of projection data where the measured intensity is monotonically related to a certain material property, e.g. mass-thickness. For certain applications, e.g. phase contrast nano-CT, the correct consideration of the imaging physics in 3D reconstruction algorithms remains a challenge. If not properly described, this can lead to strong reconstruction artefacts. Moreover, beam sensitivity or specific sample geometry may prevent the acquisition of full 180° tilt series with adequate angular sampling which would be required for artefact-free 3D reconstruction. A possible solution to reduce the amount of projection data required while maintaining reconstruction quality is to use compressed sensing approaches by taking into account certain prior knowledge about the sample characteristics, such as that the sample contains only a small number of piecewise constant material phases. In a first approach, we utilized this information by deriving valid constraints on the allowed reconstructions, whereas later, we first approximated the nonconvex Sobel operator to capture “sharp edges”, i.e., a rapid transition between material and void, and then nudge the compressed sensing algorithm towards reconstructing sharper edges.

Talare: Jan Glaubitz, MAI

Titel: Better together: Image reconstruction and Bayesian scientific computing

Sammanfattning: How can we recover clear, reliable images or signals when our observations are indirect, incomplete, and corrupted by noise? This challenge arises in many fields—from medical scans and satellite imaging to data analysis in science and engineering. Mathematically, such problems are known as inverse problems, and they are often “ill-posed,” meaning that small errors in the data can lead to large errors in the result. Traditional methods, such as compressive sensing, have shown impressive results in producing high-resolution reconstructions. However, these approaches can be sensitive to parameter choices and typically provide only a single “best guess” solution—without telling us how confident we should be in that result.

In this talk, I will describe how we can address these limitations by adopting a Bayesian perspective. Instead of producing just one answer, Bayesian methods treat image reconstruction as a problem of statistical inference: they combine information from the data (what we observed) with prior knowledge (what we expect the image to look like) to produce a full probability distribution of possible solutions. This distribution not only yields reconstructions that are robust but also provides a natural way to quantify uncertainty—for example, highlighting which parts of an image are reliable and which are not. I will discuss how we design priors that preserve essential features, such as edges, and how we efficiently compute with these models using ideas from optimization and measure transport. Additionally, I will explore how these methods perform in practice for applications in medical imaging, remote sensing, multi-measurement problems, and time-dependent imaging tasks.

Talare: Magnus Herberthson, MAI

Titel: Some optimization problems in diffusion MRI (magnetic resonance imaging)

Sammanfattningt: By performing a sufficient set of MRI measurement, it is possible to estimate various properties of tissue, e.g., the brain. Of interest for us is a second order tensor/matrix called the diffusion tensor and also a fourth order tensor (the covariance tensor). Finding these from the measurements is a fitting problem with can be solved by least squares. Under that approach, the estimates may be unphysical, and we discuss how one can use semidefinite programming to impose positivity conditions which are known to hold.

Måndag 17 mars 2025, 15.15-16.30, Robert Forchheimer, ISY

Titel: ChatGPT - How it works (and partly why)

Sammanfattning: The talk describes the core algorithm behind ChatGPT and similar large language models. The main components—autoregression, word vectorization, attention, and the use of neural networks—are explained and exemplified, showing how they come together to form the well-known Transformer architecture.

M氓ndag 27 maj 2024, kl. 13:15-14:00, Carl Johan Casselgren, Matematiska institutionen, Link枚pings universitet

Titel: Length-constrained Cycle Partition with an Application to UAV Routing

Sammanfattning: The following problem arises in routing of unmanned aerial vehicles. Let G be an undirected graph with nonnegative edge weights and assign to each vertex a critical weight value. We consider the problem of finding a covering of the vertices by disjoint cycles so that the weight of every cycle does not exceed the critical weight value of each vertex contained in it (where the weight of a cycle is the sum of the weights of the edges contained in it). The goal is to find such a cycle cover with a minimum number of cycles. We shall discuss complexity as well as some simple heuristics for this problem.

L盲nk till artikeln

Anordnas i samarbete med Seminarier i optimeringsl盲ra.

M氓ndag 6 maj 2024, kl. 13:15-14:00, Farnaz Adib Yaghmaie, Institutionen f枚r systemteknik, Link枚pings universitet

Titel: An Online Efficient Learning Algorithm: For control problems and from the control theory lens

Sammanfattning: Successful deployment of Machine Learning (ML) algorithms on physical systems requires the development of new learning mechanisms that can commit to making appropriate decisions on the fly and using only limited available data collected for an unseen task. Despite its vital importance, existing ML algorithms for control systems typically impose restrictive data requirements on the system in terms of the number of samples required to learn. Additionally, it is difficult to provide performance guarantees. On the other hand, classical approach for control systems are offline and ensures theoretical guarantees but lacks adaptability. The purpose of this talk is to propose a new learning framework for control systems that 1) is online and adaptive, 2) provides performance guarantees, and 3) is sample efficient.

Anordnas i samarbete med Seminarier i optimeringsl盲ra.

Tisdag 2 april 2024, kl 15:15, Magnus Herberthson, Matematiska institutionen, Link枚pings universitet

Titel: The central limit theorem for random walks on bounded domains

Sammanfattning: The standard central limit theorem typically assumes independent identically distributed (i.i.d.) random variabels X_i (with E[X_i]=0 and finite variance V[X_i]). A scaled average of N terms will then converge to a normal distribution as N -> oo, and the variance of this normal distribution is directly related to V[X_i].

We consider a random walk in a bounded domain Omega (for simplicity in the in the plane) which will imply that the random walk variables X_i are not any longer
independent. A central limit theorem still ensures a limiting distribution for the rescaled averages as the number of terms increases, but now the covariance matrix of the limiting distribution is not so obvious. We will discuss how this covariance matrix is given by the geometry of the bounding domain Omega.

M氓ndag 5 februari 2024, Bj枚rn Mor茅n, Matematiska institutionen, Link枚pings universitet

Titel: Experiences From Being a Postdoc at a Medical Physics Unit

Sammanfattning: This presentation will be about my postdoc period for half a year at McGill University, Montreal, Canada. I was at a Medical Physics Unit with much research focus on AI and brachytherapy, which is a type of radiation therapy where the radiation is given from within the body. In the first part of the presentation, I will describe how the cooperation was formed, my motivation for going there, and how it is to be in an environment that is very different from what you are used to.

In the last part I will discuss one of my projects during the visit in more details. One of their specialties is developing new types of delivery techniques for brachytherapy as different techniques are used for each treatment site. When I came there, they worked on a new prototype for brachytherapy treatment of rectal cancer, and they were planning the next steps to do a clinical trial. Our research project was aimed at investigating positional uncertainties in the treatment and the robustness of their delivery prototype.

Måndag 24 april 2023, kl 13.15 David Abramian, Institutionen för medicinsk teknik, Linköpings universitet

Titel: Introduction to convolutional neural networks with applications in medical imaging

Sammanfattning: Convolutional neural networks (CNNs) have driven the recent explosion in interest in artificial neural networks, advancing the state-of-the-art for a variety of computer vision tasks, as well as bringing a number of consumer applications to the public. Although somewhat displaced in recent years by the development of more advanced models, they are still the de facto standard for image applications in deep learning. This seminar will provide a gentle introduction to CNNs, discussing their principles

Plats: Kompakta rummet, B-huset, ingång 23, plan 2, Campus Valla

Onsdag 1 juni 2022, Ezio Venturino, Università di Torino

Titel: Fighting alien species invasions through mathematical modeling

Sammanfattning: Recently, we have investigated the phenomenon of native populations replacement by exotic ones, focusing on accidentally or intentionally introduced species in Italy and UK. Two major cases are of interest. American grey squirrels are gradually outcompeting the indigenous red ones both in Italy and in UK. In the latter case, the situation is worsened by the presence of a virus, carried by the aliens, for which it is harmless, but which ultimately is found lethal for the native species. In Italy the European hare is slowly forcing the extinction of the native mountain hare. The main difference in this case is that the contacts among the two occur just on the boundary region of the territories occupied by the two species. This allows us to formulate a non-classical interaction model, based on the concept of herd behavior. The latter is extended to other possible situations, showing the onset of some new features in the phase space, revealing quite different results from the classical ones.
OBS! Lokal: C3

Anordnas i samarbete med Matematiska kollokviet

Onsdag 11 maj 2022, Magnus Herberthson, Matematiska institutionen, Linköpings universitet

Titel: Characterising the second order moments of diffusion tensor distributions

Sammanfattning: In magnetic resonance imaging (MRI), it is possible to probe tissue on a sub-resolution level. The signal obtain by MRI is then formed by a family of diffusion tensors, with may have certain statistical properties. In this talk we are interested in characterising the second order moment of this family, which is a forth order tensor with certain symmetries.
Plats: i  som ligger i B-huset, ingång 23, plan 2, Campus Valla i Linköping. 
Seminariet är ett samarrangemang med Matematiska kollokviet.

Onsdag 21 oktober 2020, Åsa Holm, Furlands Revisionsbyrå

Forskarskolan fortsätter att uppmärksamma sitt 20-årsjubileum med ännu ett distansseminarium av en forskarskolealumn.

Åsa Holm över ämnet: Tillämpad optimering möter den riktiga verkligheten - tema strålbehandling (och lite ekonomi).

Onsdag 14 oktober 2020, Andreas Rietz, Scania

Onsdagen 14 oktober ger Andreas Rietz (med bakgrund i forskarskolan i tvärvetenskaplig matematik) ett distansseminarium med start 13.15.

Ämnet som avhandlas är: Solid mechanics of trucks, and a remark on the functional spaces (Lastbilars hållfasthet, och lite om funktionsrummen.) 

Onsdag 22 januari 2020, Elina Rönnberg, Matematiska institutionen, Linköpings universitet

Titel: Efficient use of hardware resources in avionic systems

Sammanfattning: A key ingredient when designing an avionic system – i. e. the electronic system of an aircraft – is to make sure that it always can be trusted. In modern integrated modular avionic systems, different aircraft functions share hardware resources on a common avionic platform. For such architectures, it is necessary to create a spatial and temporal partitioning of the system to prevent faults from propagating between different functions. One way to establish a temporal partitioning is through pre-runtime scheduling.

While the avionic systems are growing more and more complex, so is the challenge of scheduling them. Scheduling of the system has an important role when a new avionic system is developed. Typically, functions are added to the system over a period of several years and a scheduling tool is used both to determine if the platform can host the new functionality and, in case this is possible, to create a new schedule.

In this talk, I will discuss a design case from Saab Aeronautics and present an optimisation-based scheduling tool that we have developed. From an optimisation point of view, the problem can be described as a rich multiprocessor scheduling problem that also includes a communication network to be scheduled. Results are presented for practically relevant large-scale instances with up to 60 000 tasks.

Torsdag 17 maj 2018, Daniel Petersson, Veoneer

Seminariet är ett samarrangemang med Seminarier i Optimeringslära.

Titel: Eyes for self-driving cars

Sammanfattning: The development of technical systems has in many ways been dramatically changed during the last years through the breakthrough of deep learning. Traditional machine learning techniques using hand-crafted features have been successful for many applications, such as object detection in vision systems. However, the deep learning framework often enables higher performance, shared computations between different applications, and the ability to solve more advanced problems such as autonomous driving.

In the seminar, I will share experiences from the Veoneer development of vision systems for cars, such as data driven development and how the new advancements in the field of deep learning affect our products/services, as well as our ways of working.

Onsdag 15 november 2017, Jan-Åke Larsson, Institutionen för systemteknik, Linköpings universitet

Seminariet är ett samarrangemang med Matematiska kollokviet.

Titel: Efficient simulation of some quantum computer algorithms

Sammanfattning: A long-standing aim of quantum information research is to understand what gives quantum computers their advantage. Such an understanding would be of great benefit when attempting to build a quantum computer. Here we present a framework that uses classical resources but still is able to efficiently run, for example Deutsch-Jozsa and Simon's algorithms, and also can run Shor's factoring algorithm with some systematic errors. We also perform an experiment factoring 15 using classical pass-transistor logic at room temperature, with smaller systematic errors than any former experimental implementation, and the same amount of resources in time and space as a scalable quantum computer. Our results give further insight into the resources needed for quantum computation, aiming for a true understanding of the subject.

Onsdag 3 maj 2017, David Rule, Matematiska institutionen, Linköpings universitet

Seminariet är ett samarrangemang med Matematiska kollokviet.

Titel: The global boundedness of Fourier integral operators on local Hardy spaces

Sammanfattning: The question of the local $L^p$-boundedness of Fourier integral operators when $p\neq2$ was answered in work of Seeger-Sogge-Stein in the early nineties. But only recently have Ruzhansky-Sugimoto found sufficient conditions to prove global $L^p$-boundedness. We build on their methods to prove the global boundedness of Fourier integral operators in the (mostly quasi-Banach) setting of local Hardy spaces $h^p$ in the range $n/(n+1) < p \leq 1$. This is joint work with Salvador Rodríguez-López and Wolfgang Staubach.

Onsdag 1 februari 2017, Clas Rydergren, Institutionen för teknik och naturvetenskap, Linköpings universitet

Seminariet är ett samarrangemang med Matematiska kollokviet.

Titel: New sources of input data for travel demand estimation models

Sammanfattning: Forecasts of how journeys are made, from where and to where, is usually done using simple mathematical models. The results from the models are used as estimates of the load on the transport system. Estimates are made for the load today, in the future, and to estimate the change in travel demand when changes in the infrastructure are made. The model result is critical input to traffic planners.

Different types of models are used depending on, among other things, if the analysis requires the results to include all modes of transport, or not, and the length of the forecast horizon. Models for long-term forecasts (several years) often contain components to describe the travelers' values and perceptions whereas models for the current situation or with a very short forecast horizon, often is based exclusively a network model and data from dedicated traffic measurements.

During this seminar, I will present an example of a traditional demand model, and exemplify how this model is used. The ongoing digitization have led to new sources of input data for this type of models. I will exemplify this by presenting inputs used in a couple of research projects ongoing at the Division of Communications and Transport System (KTS) at ITN.