### Matematiikan ja systeemianalyysin laitos

- Tutkimusryhmät
- Opiskelu
- Henkilökunta
- Ajankohtaista
- Yhteystiedot
- Sisäsivut

* Seuraavan viikon tapahtumat merkitty tähdellä

Wontae Kim**Some extensions on the higher integrability of the parabolic p-Laplace system***** ** Wednesday 26 January 2022, 12:15, M3 (M234)

Seminar on analysis and geometry

Matteo Allaix**Introduction to Quantum Error Correction (Part 2)***** ** Wednesday 26 January 2022, 15:15, M3 (M234)

In this seminar, we will first show the Quantum Teleportation algorithm, one of the most important known quantum algorithms. After a short description of quantum channels and quantum noise, we will finally show an example of a 3-qubit quantum error correction algorithm.

ANTA Seminar

Kim Myyryläinen**A weak Gurov-Reshetnyak class. Part 1**

Wednesday 02 February 2022, 12:15, M3 (M234)

Seminar on analysis and geometry

Dr. Ragnar Freij-Hollanti, Dr. Tapani Matala-aho, Niklas Miller, and Rahinatou Njah**Mini Math Days (talks from the Finnish math days)**

Wednesday 02 February 2022, 16:15, M3 (M234)

ANTA Seminar

Kim Myyryläinen**A weak Gurov-Reshetnyak class. Part 2**

Wednesday 09 February 2022, 12:15, M3 (M234)

Seminar on analysis and geometry

**Systeemitieteiden kandidaattiseminaari / Bachelor seminar in systems analysis**

Monday 21 February 2022, 09:30, Zoom

Further information

The link to the zoom-meeting is available from Juho Roponen

**Systeemitieteiden kandidaattiseminaari / Bachelor seminar in systems analysis**

Thursday 17 March 2022, 09:30, Zoom

Further information

The link to the zoom-meeting is available from Juho Roponen

**Systeemitieteiden kandidaattiseminaari / Bachelor seminar in systems analysis**

Wednesday 11 May 2022, 09:30, Zoom

Further information

The link to the zoom-meeting is available from Juho Roponen

**Systeemitieteiden kandidaattiseminaari / Bachelor seminar in systems analysis**

Wednesday 15 June 2022, 09:30, Zoom

Further information

The link to the zoom-meeting is available from Juho Roponen

**Systeemitieteiden kandidaattiseminaari / Bachelor seminar in systems analysis**

Friday 21 January 2022, 09:30, Zoom

Further information

The link to the zoom-meeting is available from Juho Roponen

Dr. Tapani Matala-aho**Hermite-Thue equation: Padé approximations and Siegel's lemma, Part 2**

Wednesday 19 January 2022, 16:15, M3 and Zoom

Further information

Padé approximations and Siegel's lemma are widely used tools in Diophantine approximation theory.
The homogeneous matrix equation representing both methods has an M x (L+1) coefficient matrix, where M is at most L. Due to the Bombieri-Vaaler version of Siegel's lemma, the upper bound of the minimal non-zero solution of the matrix equation can be improved by finding a big common factor of all the M x M minors of the coefficient matrix. Further, in the case M = L, the existence of this common factor is a step towards understanding the nature of the 'twin-type' Hermite-Padé approximations to the exponential function.
In this second lecture we reproduce the classical type II Hermite-Padé approximations of the exponential series by computing the homogeneous vector of the L x L minors (Cramer's rule).
These minors are Vandermonde-type block determinants which are challenging to unwrap.
For that we introduce appropriate determinant calculus tools which have interest of their own sake.
Joint work with Louna Seppälä.

ANTA Seminar

Timo Takala**An interesting example of a JNp function**

Wednesday 19 January 2022, 12:15, M3 (M234)

Seminar on analysis and geometry

Olle Hallqvist Elias (Alfréd Rényi Institute of Mathematics)**Interlacements and the GFF**

Friday 14 January 2022, 14:00, Zoom

Further information

mathematical physics, Kytölä & Peltola

Ommolbanin Behzad (University of Isfahan)**Exterior powers, Polynomial rings, and Representation of Lie Algebras**

Friday 14 January 2022, 11:00, Zoom

Further information

I will report on some recent work of myself, A.Contiero and D. Martins about representing lie algebras of vector space endomorphisms on exterior algebras, seeing it as the finite type case of the celebrated DJKM bosonic vertex operator representation of gl_∞(Q).

mathematical physics, Kytölä & Peltola

Olga Chekeres (University of Connecticut)**Quantum Wilson surfaces and topological interactions**

Tuesday 04 January 2022, 17:00, Zoom

Further information

mathematical physics, Kytölä & Peltola

Julien Roussillon (KTH)**Confluence of correlation functions in Liouville theory**

Monday 03 January 2022, 11:00, Zoom

Further information

mathematical physics, Kytölä & Peltola

Okko Makkonen**New schemes for secure distributed matrix multiplication (MSc thesis presentation)**

Friday 17 December 2021, 13:00, M3 (M234) and Zoom

Further information

ANTA Seminar

MSc Joona Karjalainen (Aalto, candidate) & Prof. Remco van der Hofstad (TU Eindhoven, opponent)**Structure and estimation of network models with overlapping communities (DSc defence)**

Friday 17 December 2021, 12:00, M1 (M232)

Further information

Live video stream: https://aalto.zoom.us/j/65777606406
Abstract: Many types of data in different fields of science can be naturally represented as networks. Social relationships in groups of people, the structure of the internet, and traffic networks can all be understood as collections of nodes and connections between them. Real-world networks often show signs of community structure, i.e., some groups of nodes are more densely connected to each other than to the rest of the nodes. Since communities may emerge through many different mechanisms, it is natural to describe these networks with statistical models where the communities are allowed to overlap. Even in the absence of obvious communities, various other types of structure are commonly observed in data. For example, the degrees of adjacent nodes tend to be correlated, and node pairs have an increased probability of being adjacent if they have common neighbors.
This dissertation is concerned with the structure of large and sparse statistical network models with overlapping communities. This structure is described using statistical quantities and distributions and their limits as the number of nodes tends to infinity. The focus is on the asymptotic behavior of subgraph frequencies, joint degree distributions of adjacent nodes, and various summary statistics. New results are proved on their convergence, and exact formulas are provided for their limits. These results lead to new estimators of the model parameters based on counting the frequencies of small subgraphs. The consistency of these estimators is proved under complete or partly incomplete data.
The results show that the models have structural similarities with many real-world networks, such as non-trivial clustering, degree correlations, and power laws. This illustrates how some empirical observations on network data can be explained with an underlying overlapping community structure.

Perttu Saarela**On coding theory and private information retrieval: A new robust scheme for Reed-Muller codes (MSc thesis presentation)**

Friday 17 December 2021, 12:00, M3 (M234) and Zoom

Further information

ANTA Seminar

Gabriele Rembado (Bonn)**Singular modules for affine Lie algebras and applications to irregular WZNW conformal blocks**

Thursday 16 December 2021, 15:15, M2 (M233)

Eveliina Peltola

Marc Härkönen (Georgia Institute of Technology)**Solving PDE with nonlinear algebra**

Thursday 16 December 2021, 11:00, M3 (M234) and https://aalto.zoom.us/j/66578928227

In an undergraduate differential equations course we learn to solve a linear ordinary differential equation by factoring the characteristic polynomial. This works also for in more generality for linear PDE with constant coefficients, where primary decomposition of ideals and modules plays the role of factorization. The celebrated Fundamental Theorem by Ehrenpreis and Palamodov equates the primary components to families of solutions for the corresponding PDE. This yields an alternative characterization of an ideal or module as a set of solutions to a PDE, which can be exploited in computations. In this talk I will present some historical notes along with some recent algorithmic advances in this direction.

Kubjas

Ifrah Sheikh**Kierretyn nauhan mallinnuspalkkina Kirchhoffin yhtälön avulla**

Wednesday 15 December 2021, 14:15, Zoom

Further information

Kandidaatintyö

Harri.Hakula@aalto.fi

Niklas Miller**On tame lattices**

Wednesday 15 December 2021, 09:30, Zoom

Further information

Tame lattices were introduced in 2020 by Mantilla-Soler and Damir, to capture the key properties of lattices arising from tame abelian number fields of either prime degree or conductor, via the Minkowski embedding. Families of well-rounded sublattices of tame lattices were constructed to generalize the observations of Costa et al., that certain submodules of the ring of integers of tame number fields of odd prime degree produce well-rounded lattices.
Later the packing density of well-rounded tame sublattices was characterized and it was also noted that they are either generic well-rounded or similar to the root lattice A_n. Tame well-rounded sublattices closely resemble nearly orthogonal lattices, which have a basis of almost orthogonal vectors. In a 2020 paper by Fukshansky et al., nearly orthogonal well-rounded lattices were studied in more detail, and it was shown that they are, among other things, not local maxima to the sphere packing density function.

ANTA Seminar

Marianne Honkasaari**On Optimization of the Logistics Related to Recycling of Nutrients in Wastewater Sludges (Master's thesis presentation)**

Tuesday 14 December 2021, 15:00, Zoom

Further information

Roosa Ilvonen**Using simulated TMS-EEG data in source localization analysis**

Tuesday 14 December 2021, 14:15, Zoom

Further information

Kandidaatintyö: https://aalto.zoom.us/j/67770051751

Harri.Hakula@aalto.fi

Mikko Seesto**Machine learning with qubits: Experimental realisation of a quantum kernel method**

Tuesday 14 December 2021, 10:00, M3 (M234)

Prof. Joachim Schöberl**NGSolve - A finite element package for teaching and research**

Thursday 09 December 2021, 13:15, Zoom

Further information

Matteo Allaix**Introduction to Quantum Error Correction**

Wednesday 08 December 2021, 14:15, M3 (M234)

In this seminar, we will introduce some definitions of quantum information theory in order to describe some quantum error correction algorithms. We will first define density operators and mixed state to introduce the Von Neumann entropy and some distance measures for quantum systems. After a brief review of classical error correction, we will show an example in the quantum setting.

ANTA Seminar

Lauri Särkiö (masters thesis talk)**Local higher integrability of the parabolic p-Laplace equation**

Wednesday 08 December 2021, 12:15, M3 (M234)

Seminar on analysis and geometry

Petteri Kaski**Algebraic fingerprinting and the shortest even cycle problem**

Thursday 02 December 2021, 10:00, Zoom

Further information

This talk gives an introduction to the algebraic fingerprinting technique in algorithm design and looks at a recent application of the technique to the shortest even cycle problem in directed graphs.
(Joint work with Andreas Björklund and Thore Husfeldt --- https://arxiv.org/abs/2111.02992 .)

StAGe

**Systeemitieteiden kandidaattiseminaari / Bachelor seminar in systems analysis**

Thursday 02 December 2021, 09:30, Zoom

Further information

Emma-Karoliina Kurki**Characterizations of weak reverse Hölder inequalities on metric spaces**

Wednesday 01 December 2021, 12:15, M3 (M234)

Seminar on analysis and geometry

Augustin Lafay (ENS)**Web models as generalizations of statistical loop models**

Friday 26 November 2021, 10:15, Zoom

Further information

Two dimensional gases of non intersecting loops have been a subject of study in mathematical physics for more than thirty years because of their numerous connections to integrability, two dimensional conformal field theory, random geometry and combinatorics. In this talk, I will present a natural generalization of loop models to gases of graphs possessing branchings. These graphs are called webs and first appeared in the mathematical community as diagrammatic presentations of categories of representations of quantum groups. The web models posses properties similar to the loop models. For instance, it will be shown that they describe, for some tuning of the parameters, interfaces of spin clusters in Zn spin models. Focusing on the numerically more accesible case of Uq(sl3) webs (or Kuperberg webs), it is possible to identify critical phases that are analogous to the dense and dilute phases of the loop models. These phases are then described by a Coulomb Gas with a two component bosonic field.

Kytölä & Peltola

Tuomas Kelomäki**A Geometric Proof of the Borsuk-Ulam Theorem**

Thursday 25 November 2021, 10:00, M3 (M234)

We will introduce a classical result in topology called the Borsuk-Ulam theorem and provide a somewhat elementary proof to it. The machinery used in the proof will not use any algebraic topology. Instead we will make use of a carefully constructed simplicial approximation. If the time allows, we will also show some applications of the theorem. Since I am a new PhD student this talk will be based on my master thesis and will not contain any new results.

StAGe Seminar

Cintia Pacchiano**Higher integrability and stability for (p,q)-quasiminimizers**

Wednesday 24 November 2021, 12:15, M3 (M234)

Seminar on analysis and geometry

Sonja Oksanen (Aalto)**Predicting residential property prices with decision tree models (MSc thesis presentation)**

Monday 22 November 2021, 14:15, M203

The price of a residential property is determined by diverse attributes, such as the size, condition, or location of a property. A number of studies have predicted property prices utilising these attributes and researched the most significant determinants in property price formation. Typically, property prices have been estimated with so called hedonic price models. Recently, however, the popularity of machine learning methods in property price estimation has increased. In this thesis, a machine learning framework for predicting residential property prices is developed. Random forests, gradient boosting machine, and XGBoost algorithms are implemented. Property prices are predicted utilising real-life data of apartment transactions with information of location-specific attributes and specific housing features. The results indicate that the machine learning models predict residential property prices accurately. XGBoost and gradient boosting machine outperform random forests in prediction accuracy, and XGBoost produces the best computational performance. Finally, the derived machine learning framework is tested on a research area in a city district of Espoo where the future average price level of the district is predicted. The developed machine learning framework can improve understanding of the formation of residential property value and thus be used as a tool for decision making by different operators, such as real estate investors, urban planners, home buyers, or politicians.

Lasse Leskelä

Oskar Henriksson**Geometric perspectives on the steady states of chemical reaction networks**

Thursday 18 November 2021, 10:00, M3 (M234)

This talk gives an introduction to the algebraic study of biochemical reaction networks, and some of the ways in which tools from computational algebraic geometry can help us understand their dynamics. In particular, we will discuss some recent results about generic dimension and monomial parametrizability of the set of steady states, based on a joint work in progress with Elisenda Feliu and Beatriz Pascual Escudero. No background in chemistry or algebraic geometry will be assumed.

Matteo Allaix**Introduction to Quantum Information Theory**

Wednesday 17 November 2021, 15:15, M3 (M234)

In this seminar, we will introduce the basic notations and definitions of Quantum Information Theory. We will first describe the three postulates of quantum mechanics and then we will introduce the notions of qubit, quantum state, quantum gate, entanglement and possibly distance measures.

ANTA Seminar

Julian Weigt**Covering techniques for the maximal operator**

Wednesday 17 November 2021, 12:15, M3 (M234)

Seminar on analysis and geometry

Yizheng Yuan**Refined regularity of SLE**

Monday 15 November 2021, 14:15, Y229a

SLE (Schramm-Loewner evolution) is a family of random planar curves that have some natural conformal invariance properties. They appear in a variety of planar models that exhibit conformal invariance in the scaling limit. Regarding their regularity, the optimal Hoelder and p-variation exponents are known from previous works. In this talk, I will present refinements of the regularity statements to the logarithmic scale. I will present a new argument for obtaining these results and discuss some applications.

Eveliina Peltola

Miikka Tiainen **Computational Entropy from Distributional hardness (master's thesis presentation)**

Monday 15 November 2021, 10:00, Zoom

Further information

A central problem in cryptography is the construction of basic primitives, or lowlevel algorithms, from computational complexity-based assumptions. One way of viewing the hardness of a problem from the view of a computationally bounded adversary is via the notion of entropy. Much like the toss of a normal coin is considered random due to limitations of human observers, the real entropy, or uncertainty, of a system can be much higher or lower than the entropy that is observable by an efficient adversary. In this thesis we establish results obtaining this type of computational entropy from distributionally hard primitives. This notion of distributional hardness captures that it is hard for an adversary to output a uniform pre-image of a randomly sampled image value. We use this computational entropy from distributional hardness to expand on existing results constructing pseudorandom generators from next-block pseudoentropy, and statistically hiding commitment schemes from accessible entropy. Although the existence of such constructions were implicit in existing literature, we establish much more efficient constructions with tighter bounds on the computational entropy than has previously been considered. Furthermore, the current known optimal construction of pseudorandom generators in terms of input (or seed) length appears to hold with equivalent parameters for the much more general notion of distributional hardness, establishing that the much more general notion of distributional hardness may in itself yield conceptually interesting constructions. We improve on existing results using known known information theoretic inequalities. Most centrally we use an inequality due to Bretagnolle and Huber relating the statistical distance and relative entropy of distributions in a much tighter way in the context of highly disjoint distributions than the famous Pinsker bound.
Join Zoom Meeting
https://aalto.zoom.us/j/68643838319

Dissertation

Valentina Candiani, opponent: Prof. Erkki Somersalo (Case Western Reserve University, Cleveland)**Computational approaches in electrical impedance tomography with applications to head imaging**

Friday 12 November 2021, 13:15, M1 (M232)

Prof. Daniela Calvetti (Case Western Reserve University, Cleveland)**Bayes meets Data Science to identify changes in brain activity during meditation from MEG measurements**

Thursday 11 November 2021, 13:15, M1 (M232)

Ardiyansyah Muhammad**Distinguishing Phylogenetic Networks Using Phylogenetic Invariants**

Thursday 11 November 2021, 10:00, M3 (M234)

Phylogenetics is a field in biology that studies the evolutionary relationship between organisms. Phylogenetic networks can represent evolutionary events that cannot be described by phylogenetic trees. In this talk, we introduce how to define a phylogenetic model on a particular class of phylogenetic networks to obtain a probability distribution on tuples of DNA bases observed from the extant species. Moreover, we will introduce the notion of distinguishability of phylogenetic networks. Using an algebraic approach, namely using discrete Fourier transformation, we will present some results on the distinguishability of some level-2 networks using phylogenetic invariants, which are polynomials associated with a phylogenetic network model.

Kristian Moring**Hölder regularity for obstacle problem to the porous medium equation**

Wednesday 10 November 2021, 12:15, M3 (M234)

Seminar on analysis and geometry

Ettore Teixeira Turatti (University of Florence)**Tensors determined by their eigenscheme**

Thursday 04 November 2021, 10:00, M3 (M234)

We will introduce the notion of eigentensor and eigenschemes for multisymmetric tensors. We then study the question: given a general tensor t, there exist other tensors that have the same eigentensors of t? We will show that if there is at least a component of degree odd, then just t has these eigentensors, otherwise there is a 1-dimensional space of tensors with the same eigentensors as t.

Niklas Miller**The Twisted Ring-LWE Problem**

Wednesday 03 November 2021, 15:15, M3 (M234)

In an interesting paper by Ortiz, Araujo, Aranha, Costa and Dahab, the authors consider a generalisation of the Ring-LWE problem. The usual RLWE uses the canonical embedding to map an underlying ring to a lattice in R^n. The twisted RLWE (RLWE^t) generalises this by considering a twisted embedding. The authors provide a security reduction from RLWE to RLWE^t, and show that the twisted embedding allows for more algebraic lattices to be used in lattice-based cryptosystems.

ANTA Seminar

Kim Myyryläinen**Dyadic maximal operator on the dyadic JohnNirenberg space**

Wednesday 03 November 2021, 12:15, M3 (M234)

Seminar on analysis and geometry

Lassi Ruoppa**Maximal number of subsets occurring as substrings (B.Sc. thesis presentation)**

Tuesday 02 November 2021, 10:15, Zoom

Further information

Let s be a string of length n over the alphabet [m]:={1,2,...,m}. We say that a set S occurs as a substring in s, if some substring of s contains precisely the elements of S, some possibly repeated. We write C(m,n) for the maximum number of subsets occurring as substrings across all strings of length n over [m]. We will present both an efficient algorithm for computing C(m,n) and exact analytic expressions for entries on the diagonal C(m,m) and first superdiagonal C(m,m+1).

ANTA Seminar

Lassi Ruoppa**Maximal number of subsets occurring as substrings / Osamerkkijonoina esiintyvien osajoukkojen maksimaalinen lukumäärä (Kandiseminaari)**

Tuesday 02 November 2021, 10:15, Zoom

https://aalto.zoom.us/j/68278283503

Alexander Engström (Aalto)**Betti polytopes**

Thursday 28 October 2021, 10:00, M3 (M234)

Milo Orlich and I recently proved that if certain Betti numbers of some ideals vanish, then almost all those ideals have the same CastelnuovoMumford regularity. The almost is crucial, otherwise it is false. Instead of Betti numbers vanishing, one might consider inequalities for them to get Betti polytopes, and the similar question if almost all of ideals on a facet have the same CastelnuovoMumford regularity. I will discuss some rather preliminary work in progress with Milo.

StAGe

**AGENT Forum 2021**

Wednesday 27 October 2021, 11:00, AS2

Further information

Kalle Kytölä**Formal proofs for (and by) amateurs, informally**

Tuesday 26 October 2021, 15:15, M2 (M233)

An informal discussion of formal proofs (in Lean).

**Bachelor Seminar in Systems Analysis**

Friday 22 October 2021, 09:30, Zoom

For Zoom link contact Juho Roponen

Milo Orlich(Aalto)**Parabolic Betti numbers and regularity of edge ideals of graphs**

Thursday 21 October 2021, 10:00, M3 (M234)

To a finite undirected graph G with no multiple edges and no loops, one associates its so-called edge ideal I(G), in a polynomial ring with coefficients in a field. The Betti numbers are numerical invariants defined in a purely algebraic way for any homogeneous ideal in such a polynomial ring, in particular for edge ideals I(G). The Betti numbers of an edge ideal I(G) have well-known combinatorial interpretations in terms of the graph G. However, a satisfactory explicit description of these numbers in terms of "easy" invariants of G (such as the number of edges of G, the number of triangles, etc.) is still out of reach in general. Another numerical invariant, much coarser than the Betti numbers, is the regularity of I(G). In spite of the great deal of research the regularity of I(G) has been the subject of, there are still no general "easy" formulas for it, either. In a recent joint work with Alexander Engström we introduce the concept of "parabolic Betti number" and employ methods from extremal graph theory to determine the exact value of the regularity of I(G), for almost all graphs G with a given parabolic Betti number equal to zero. Large part of the talk will be devoted to defining all the notions involved, in order to make it as accessible as possible. Next week's talk by Alex Engström is going to be closely related, and this talk can be seen as an introduction to that.

StAGe

Aleksi Avela **On Handling Imbalanced Data in Text Classification (Master's thesis presentation)**

Wednesday 20 October 2021, 16:00, Zoom

Further information

Florian Kohl (Aalto)**Unconditional Reflexive Polytopes**

Thursday 14 October 2021, 10:00, M3 (M234)

A convex body is unconditional if it is symmetric with respect to reflections in all coordinate hyperplanes. In this talk, we investigate unconditional lattice polytopes with respect to geometric, combinatorial, and algebraic properties. In particular, we characterize unconditional reflexive polytopes in terms of perfect graphs. As a prime example, we study a type-B analogue of the Birkhoff polytope. No background knowledge of polytopes or graphs is needed. In particular, I will explain every word in the title. This talk is based on joint work with McCabe Olsen and Raman Sanyal.

StAGe --- Seminar on Statistics, Algebra, and Geometry

Dr. Tapani Matala-aho**Hermite-Thue Equation: Padé approximations and Siegels Lemma**

Wednesday 13 October 2021, 15:15, Zoom

Further information

Padé approximations and Siegels lemma are widely used tools in Diophantine approximation theory. The appropriate homogeneous matrix equation representing both methods has an M x (L+1) coefficient matrix, where M≤L. Due to the Bombieri-Vaaler version of Siegels lemma, the upper bound of the minimal non-zero solution of the matrix equation can be improved by finding a big common factor of all the M x M minors of the coefficient matrix. We will present some key ingredients on how to find such a big common factor in the case of the exponential function. Further, in the case M=L, the existence of this common factor is a step towards understanding the nature of the twin type Hermite-Padé approximations to the exponential function. Joint work with Louna Seppälä.

ANTA Seminar

Alexi Morin-Duchesne**On correlation functions of dense loop models (part 3/3)**

Thursday 07 October 2021, 11:15, M3 (M234)

Eveliina Peltola

Luca Sodomaco**Kalman Varieties of Matrices and Tensors**

Thursday 07 October 2021, 10:00, M3 (M234)

The name ''Kalman variety'' was first introduced by Ottaviani and Sturmfels to indicate the algebraic variety of square matrices possessing at least one eigenvector on a fixed linear subspace. In this talk, we recall its definition and describe its main properties.
Secondly, we introduce another type of Kalman variety: the variety of rectangular matrices possessing at least one singular vector pair with the first component on a fixed linear subspace. A tensor analog of this variety has been studied recently by Ottaviani and Shahidi, where singular vector tuples replace singular vector pairs. They computed codimensions and degrees of these varieties of tensors.
In the last part of the talk, we report on recent joint work with Shahidi and Ventura, where we extend the results of Ottaviani and Shahidi to partially symmetric tensors. Furthermore, we describe a generating function whose coefficients are the degrees of these Kalman varieties and analyze its asymptotics.
The talk is intended for a diverse audience, so no knowledge of projective algebraic geometry is required.

StAGe Seminar

Alexi Morin-Duchesne**On correlation functions of dense loop models (part 2/3)**

Wednesday 06 October 2021, 10:15, M3 (M234)

Eveliina Peltola

Alexi Morin-Duchesne**On correlation functions of dense loop models (part 1/3)**

Monday 04 October 2021, 14:15, M3 (M234)

Eveliina Peltola

Heikki Kettunen**Normaalijakaumien muunnokset pistepilvien sovituksessa (kandiseminaari)**

Friday 01 October 2021, 13:15, M2 (M233)

Leevi Kaukonen**Tree species classification from point cloud data using deep learning (Kandiseminaari)**

Friday 24 September 2021, 13:15, M2 (M233)

Onni Pohjavirta**(Master thesis talk) Optimization of Projection Geometries in X-ray Tomography**

Thursday 16 September 2021, 13:15, Zoom

Further information

In X-ray tomography, the inner structure of an object is reconstructed by taking X-ray projections from different prespectives. In general, the number of projections should be kept low in order to limit the X-ray dose. The reconstruction with a limited number of projections is an ill-posed problem and the reconstruction quality is affected by the choice of projection geometries, i.e., the experimental design. This thesis considers Bayesian inversion and Bayesian optimal experimental design in X-ray tomography.
The use of a Gaussian prior results in an offline optimization of the projection geometries without feedback from measurements, whereas a total variation (TV) prior leads to an online optimization that promotes blocky and sharp-edged reconstructions. This thesis presents the derivation of the so-called lagged diffusivity iteration which leads to a Gaussian approximation of the TV prior. The main contributions are (1) a convergence proof for the lagged diffusivity iteration in discrete X-ray imaging, (2) derivation of a gradient that can be used in the optimization of projection geometries, and (3) numerical experiments employing the proposed gradient and analogously derived Hessian matrix.
While the proposed gradient is mostly sufficiently accurate, excessive numerical errors can appear locally. Gradient descent suffers from convergence to local optima in the long run, which can be countered to some extent by using multiple random initializations. Statistically speaking, the errors in the gradient seem to have minor impact on the reconstruction errors while they make the efficient use of line search methods difficult. Excessive errors in the analogously derived Hessian presumably make the use of Newton's method infeasible.

Tero Hyytiäinen (Varian)**4D dose calculation in pencil beam scanning proton therapy (master's thesis presentation)**

Wednesday 15 September 2021, 14:15, Zoom

Further information

Jaakko Olkkonen**Hausdorffin mitta ja dimensio (Kandiseminaari)**

Monday 13 September 2021, 12:30, Zoom

https://aalto.zoom.us/j/61170340099

Anna Kosklin**Finite Element Modeling of Cell-Tissue Interactions (Kandiseminaari)**

Wednesday 08 September 2021, 12:30, Zoom

https://aalto.zoom.us/j/3090165908

Inari Puhto **Finding polynomial equations from samples (Kandiseminaari)**

Wednesday 01 September 2021, 10:00, Zoom

https://aalto.zoom.us/j/68512634218

Eveliina Peltola**On the conformal invariance of the uniform spanning tree model**

Tuesday 31 August 2021, 16:15, Zoom

Further information

Mathematical physics seminar day 31.8.2021

Kalle Kytölä, Eveliina Peltola

Kalle Kytölä**Boundary conformal field theory inspired diagonalization of the Ising transfer matrix**

Tuesday 31 August 2021, 15:30, Zoom

Further information

Mathematical physics seminar day 31.8.2021

Kalle Kytölä, Eveliina Peltola

Osama Abuzaid**On Loewner evolutions**

Tuesday 31 August 2021, 14:45, Zoom

Further information

Mathematical physics seminar day 31.8.2021

Kalle Kytölä, Eveliina Peltola

Tuomas Tuukkanen**The free fermion boundary conformal field theory**

Tuesday 31 August 2021, 14:00, Zoom

Further information

Mathematical physics seminar day 31.8.2021

Kalle Kytölä, Eveliina Peltola

Shinji Koshida**Tensor product in conformal field theory**

Tuesday 31 August 2021, 12:15, Zoom

Further information

Mathematical physics seminar day 31.8.2021

Kalle Kytölä, Eveliina Peltola

Aapo Pajala**TBA**

Tuesday 31 August 2021, 11:30, Zoom

Further information

Mathematical physics seminar day 31.8.2021

Kalle Kytölä, Eveliina Peltola

Rami Echriti**TBA**

Tuesday 31 August 2021, 10:45, Zoom

Further information

Mathematical physics seminar day 31.8.2021

Kalle Kytölä, Eveliina Peltola

David Adame-Carrillo**Ising model and discrete s-holomorphicity**

Tuesday 31 August 2021, 10:00, Zoom

Further information

Mathematical physics seminar day 31.8.2021

Kalle Kytölä, Eveliina Peltola

Antti Haavikko**Polar decomposition in 3D: Theory and implementation in C (Kandiseminaari)**

Monday 30 August 2021, 10:00, Zoom

https://aalto.zoom.us/j/67685327192

Venla Valve**Estimating the dimension of algebraic varieties from samples (Kandiseminaari)**

Friday 27 August 2021, 09:15, M1 (M232)

Aada Hakula**Optimal Experimental Design in X-ray Imaging (Kandiseminaari)**

Wednesday 25 August 2021, 14:45, M1 (M232)

Sonja Rantala**Variaatiolaskennan sovelluksia (Kandiseminaari)**

Wednesday 25 August 2021, 14:00, M1 (M232)

Atte Mäkelä**(Master thesis talk) Deep reinforcement learning as a tool for search engine campaign budget optimization**

Wednesday 28 July 2021, 10:00, Zoom

https://aalto.zoom.us/j/7736264770

Miikka Runolinna**Confidence interval estimation of extreme quantiles in semiconductor processes (Master's thesis presentation)**

Thursday 15 July 2021, 14:15, Zoom

Further information

Anton Vavilov**Planar Qasiconformal Mappings: Fundamental Properties and Characterizations**

Friday 09 July 2021, 12:00, Zoom

Further information

Master's thesis presentation

Axel Päivänsalo**Robust PCA using the L1-norm (Bachelor thesis talk)**

Wednesday 07 July 2021, 11:00, Zoom

https://aalto.zoom.us/j/64735530261

Henry Pietilä**Height Estimation Using Sensor Fusion (Bachelor thesis talk)**

Thursday 01 July 2021, 13:00, Zoom

https://aalto.zoom.us/j/7736264770

Dr. Negin Karimi**Machine learning talk series: Distributed machine learning**

Wednesday 23 June 2021, 15:15, Zoom

Further information

Nowadays, with the growing technological expansion of the world, there is a huge amount of data that is created every day. Currently, one of the best approaches for storing data is distributed systems. The earliest example of distributed storage systems was invented in 1970 when for the first time computers would be able to send messages to other systems with a local IP address. Telephone and cellular networks are other examples of distributed systems. These systems perform machine learning tasks in a distributed environment. In this talk, I'll present an introduction to distributed machine learning, which refers to multi-node machine learning algorithms and systems that are designed to improve performance, increase accuracy, and scale to larger input data sizes.

ANTA Seminar

Iiro Tynnilä**Kvasi- Monte Carlo -menetelmät**

Wednesday 23 June 2021, 10:45, Zoom

Further information

Kandiseminaari

Paul Närhi**Lämpöyhtälön vahvan maksimiperiaatteen todistus**

Wednesday 23 June 2021, 10:00, Zoom

Further information

Kandiseminaari

Dr. Taoufiq Damir**Lattice-based cryptography, part 2 (talk series for PhD students in number theory)**

Thursday 17 June 2021, 15:00, Zoom

Further information

ANTA Seminar

Dr. Taoufiq Damir**Lattice-based cryptography, part 1 (talk series for PhD students in number theory)**

Thursday 10 June 2021, 15:00, Zoom

Further information

ANTA Seminar

Prof. Guillermo Mantilla-Soler**New perspectives on Arithmetic equivalence**

Wednesday 09 June 2021, 15:15, Zoom

Further information

Two number fields are called arithmetically equivalent if their Dedekind zeta functions coincide. It is known that two arithmetically equivalent number fields share degree, discriminant, Galois closure and several other invariants. In this talk I'll recall briefly all these classical results and show how the a perspective of Galois representations and one from arithmetic geometry lead us to prove new results, and put all previous in a clear context, about A.E.

ANTA Seminar

Dr. Taoufiq Damir**On lattice constructions (talk series for PhD students in number theory)**

Thursday 27 May 2021, 15:00, Zoom

Further information

ANTA Seminar

Max Cedercreutz**Wind comfort analysis in forested areas (Master's thesis presentation)**

Friday 21 May 2021, 11:00, Zoom

Further information

Pekka Huhtala**Loop O(n) models: a numerical transfer matrix study of long range order and s-holomorphicity**

Thursday 20 May 2021, 15:15, Zoom

Further information

Master's thesis presentation

Dr. Taoufiq Damir**Lattices and wiretap channels (talk series for PhD students in number theory)**

Thursday 20 May 2021, 15:00, Zoom

Further information

ANTA Seminar

Jere Karttunen**The effect of transducer curvature on the required energy in multi-locus transcranial magnetic stimulation (Bachelor thesis talk)**

Wednesday 19 May 2021, 13:00, Zoom

https://aalto.zoom.us/j/62150352321

Dr. Razane Tajeddine (HIIT)**Machine learning talk series: Privacy preserving data sharing on vertically partitioned data**

Wednesday 12 May 2021, 15:15, Zoom

Further information

Predictive machine learning uses big amounts of data to predict the likelihood of future outcomes. To build predictive models, confidential personal data needs to be collected and shared. Therefore, user data privacy is a major concern when training such models. In many cases, the data is partitioned between multiple servers or parties, where each party holds parts of the data which can not be revealed to the other parties due to privacy concerns. Combining data from different parties gives additional information, and thus the quality of the model is significantly improved. Applying machine learning algorithms to a partitioned dataset without violating the privacy of the users is a challenging task. This talk will be concerned with methods for building predictive models on partitioned data while preserving user privacy.

ANTA Seminar

Prof. Henrik Garde (Aarhus University)**Series reversion in Calderón's problem**

Wednesday 12 May 2021, 13:15, Zoom

Further information

I will talk about reconstruction in the inverse problem of
electrical impedance tomography (Calderón's problem), on finding a
conductivity distribution from boundary electrical measurements.
Here I introduce a series reversion of the forward map, that leads to
higher order numerical methods for reconstructing a perturbation to a
conductivity coefficient, without increasing the computational complexity compared to solving a linearised problem. The numerical methods can be derived for both a continuum model and for a (smoothened) complete electrode model.
Joint work with Nuutti Hyvönen

Dr. Taoufiq Damir**Introduction to lattices (talk series for PhD students in number theory)**

Thursday 06 May 2021, 15:00, Zoom

Further information

ANTA Seminar

Dr. Laura Jakobsson**Machine learning talk series: Machine learning meets commutative algebra: Table ideal identification**

Wednesday 05 May 2021, 15:15, Zoom

Further information

Table ideals are monomial ideals that come from tables, i.e. arrangements of integers that satisfy some conditions on the entries, and these ideals have many nice properties. Given a table, the ideal coming from it in a non-minimal form is easy to compute. However, the other direction is not as easy, that is, given a monomial ideal how do we know if it is a table ideal? In this talk, I will introduce the machine learning problem of distinguishing table ideals from non-table ideals, and the results we have obtained from using machine learning to this maths question.

ANTA Seminar

Camilla Hollanti**Wireless communications and number field lattice codes (talk series for PhD students in number theory)**

Thursday 29 April 2021, 15:00, Zoom

Further information

ANTA Seminar

Dr. Mika Malinen ( Aalto University)**A local reparametrization by approximate lines of curvature coordinates**

Thursday 29 April 2021, 13:15, Zoom

Further information

A surface parametrization by orthogonal curvilinear coordinates offers many benefits from the viewpoint of both analysis and practical implementation. However, ways to construct such representations are not immediate. The utility of lines of curvature as a strategy for obtaining such coordinates globally in a smooth surface is severely limited by the presence of umbilical points. Nevertheless, a global parametrization is not necessarily needed in the numerical solution of PDE problems on surfaces, which gives us a motivation for developing
methods to obtain a local reparametrization by consistently accurate lines of curvature coordinates. A way to avoid the possible instabilities of reparametrization in a vicinity of an isolated umbilical point is considered via a sensitivity analysis. Some illustrative examples are also shown.

Dr. Laia Amoros**Machine learning talk series: Practical introduction to machine learning for mathematicians**

Wednesday 28 April 2021, 15:15, Zoom

Further information

This talk is aimed at mathematicians that have heard about machine learning and are interested in getting more familiar with the field, but dont know where to start. Machine learning (ML) covers a huge amount of algorithms that aim at learning useful information from data. Whether the best learning option is supervised, unsupervised or any other kind of learning depends on the type of our data. It turns out that handling the data takes most of the effort, and defining a model to make predictions out of the data can be done with a few simple lines of Python. We will see some easy examples in a Python notebook environment, so the interested mathematician can reproduce the experiments with their own data. We will show how ML can in particular be used to learn information about mathematical objects, which in turn can be useful to better know your data, be it monomial ideals, number fields or Calabi-Yau hypersurfaces.
For the slides and related material, click "further info".

ANTA Seminar

Samuel Nurmi**(Bachelor Thesis Talk) Controlling coil overlap in multi-locus transcranial magnetic stimulation transducers **

Thursday 22 April 2021, 13:00, Zoom

Further information

Bachelor thesis talk

Dissertation

Stavros Evdoridis**On geometric properties of harmonic and polyharmonic mappings**

Wednesday 14 April 2021, 12:00, Zoom

Link to the event https://www.aalto.fi/en/events/defense-of-dissertation-in-the-field-of-mathematics-msc-stavros-evdoridis

Tuomas Leidén**Predictive Multi-target Learning for Industrial Plant Engineering (Master's thesis presentation)**

Thursday 08 April 2021, 13:00, Teams

Further information

Prof. Paul Van Dooren, Université catholique de Louvain**Detecting roles in large graphs**

Thursday 01 April 2021, 13:15, https://aalto.zoom.us/j/67710120716

Computing meaningful clusters of nodes is crucial to analyze very large networks, but since this is a combinatorial optimization problem, its complexity can be prohibitive. In this talk, we present a pairwise node similarity measure that allows to extract roles, i.e. group of nodes sharing similar flow patterns within a network. We then propose a low rank iterative scheme to approximate the similarity measure for very large networks and we show that our low rank similarity score successfully extracts the different roles in random graphs of Erdos-Renyi type. We then argue that when projecting our similarity measure onto a low rank manifold (say, of rank r), the role models can be recovered using classical clustering techniques in a r-dimensional space, which reduces the complexity significantly. We illustrate these ideas with a number of real-world examples.

Perttu Yli-Opas**Detumble algorithm for the AuroraSat-1 propulsive attitude control system**

Thursday 25 March 2021, 14:15, https://aalto.zoom.us/j/67618085167

Bachelor's thesis talk

Luca Ferranti (University of Vaasa)**Optimised polynomial solvers for computer vision and positioning**

Monday 22 March 2021, 10:15, https://aalto.zoom.us/j/68512634218

Systems of polynomial equations naturally arise in several engineering applications. General polynomial solving techniques may not be enough to solve these problems, especially at higher robustness and efficiency requirements. For this reason, specialised polynomial solvers, exploiting the specific structure of the problem, are desired. In the talk, I will describe how optimised polynomial solvers can be automatically generated using Gröbner bases. I will also give an overview of several success stories of these techniques in computer vision and positioning applications, where these automatically generated solvers could solve challenging problems in microseconds. Finally, I will briefly discuss what problems are still open and what are current research trends in this area.

Algebra and Discrete Mathematics Seminar

Lauri Nyman**Solving Systems of 3 Equations in 3 Variables via Resultants**

Thursday 18 March 2021, 13:15, https://aalto.zoom.us/j/67710120716

This talk is a presentation of my master's thesis.
I will discuss a root-finding algorithm that computes numerically the real solutions of a system of three equations in three variables by using the Cayley resultant method. The Cayley resultant method works by first approximating the system of functions by polynomials, and then transforming the resulting multidimensional polynomial root-finding problem into a polynomial eigenvalue problem. The resulting polynomial eigenvalue problem is solved through a linearization given by the colleague pencil. I will present the details of the implementation as well as numerical results.

Olli Winberg**Compressed Sensing of Fourier sparse quasiparticle-interference map**

Wednesday 10 March 2021, 11:00, Zoom: https://aalto.zoom.us/j/62501058048

Bachelor's thesis talk

Emilia Blåsten (Aalto University)**Inverse spectral problem on discrete graphs**

Thursday 04 March 2021, 13:15, https://aalto.zoom.us/j/67710120716

I will introduce the results of my latest publication, arXiv:2101.10026, and its background. The inverse spectral problem in partial differential equation asks for determining unknown coefficients of given partial differential equation, or the underlying manifold structure, from boundary spectral measurements. These are the pairs of all eigenvalues with the boundary normal derivatives of their corresponding Dirichlet eigenfunctions to a fixed partial differential operator. The eigenvalues themselves are not enough to determine the underlying geometry: one cannot hear the shape of a drum. The question has been studied extensively for quantum graphs, but with limited success except for trees. My latest publication takes this topic into the context of discrete graphs. Given a subset of vertices (call it the boundary), and boundary values of Neumann eigenvectors to a graph Laplacian, we show that the rest of the graph structure can be determined assuming that there is "enough" boundary vertices.

Jaakko Pere**On estimation of extreme probability contours under multivariate elliptical Fréchet distributions (Master's thesis presentation)**

Wednesday 03 March 2021, 09:00, https://aalto.zoom.us/j/9150514867

Maija Vahteristo**On the effects of cervical cancer screening (Master's thesis presentation)**

Friday 19 February 2021, 10:00, https://us04web.zoom.us/j/75952975071?pwd=YVJSL3lhS2U4ck1GaWlJMEF1Y0xJZz09

Passcode:0yF0Vh

María Quintana Ponce, Aalto University**Linearizations of Rational Matrices**

Thursday 18 February 2021, 13:15, https://aalto.zoom.us/j/67710120716

Zoom-link : https://aalto.zoom.us/j/67710120716
The numerical solution of rational eigenvalue problems (REPs) is getting a lot of attention. REPs appear directly from applications or they can be used for solving arbitrary nonlinear eigenvalue problems (NLEPs) by approximating the corresponding NLEP by a REP and, then, solving the REP. They also play a fundamental role in system and control theory. Nowadays, one of the most competitive methods for solving REPs is based on linearizations of the corresponding rational matrix. In this talk, we introduce a new notion of linearization for rational matrices, and we provide very simple criteria to determine when a linear polynomial matrix is one of these linearizations. As an application, we study the structure of linearizations constructed in the literature for rational approximations of NLEPs.

Leevi Lindgren**On Moment-Based Estimation of Extreme Probability Contours Under Ellipticity (Master's thesis presentation)**

Wednesday 10 February 2021, 10:00, in Teams

https://teams.microsoft.com/l/meetup-join/19%3ameeting_OGI4YTk3YjktOGVkOC00MzBiLThlZmUtMmQzY2U5NTU3NWUz%40thread.v2/0?context=%7b%22Tid%22%3a%22af2afc02-96d2-4cff-b1b5-d43897ff97ad%22%2c%22Oid%22%3a%22e36756e7-9e4d-48ef-bad1-d1a1e6d333ec%22%7d

Tuomas Tuukkanen**Haar measures on locally compact groups (harmonic analysis workshop)**

Friday 05 February 2021, 10:15, https://aalto.zoom.us/j/67515140063

Seminar on analysis and geometry

Timo Takala**The duality of Hardy functions and BMO functions (harmonic analysis workshop)**

Friday 29 January 2021, 10:15, https://aalto.zoom.us/j/61457600970

Seminar on analysis and geometry

Sivusta vastaa: webmaster-math [at] list [dot] aalto [dot] fi