TSSP ExSci 2022 projects

The main component of the summer program is an active participation in the selected Science research project offered by the staff members of the Nicolaus Copernicus University, please see the topics and their descriptions below. Interested students are welcome to contact possible advisors for more details concerning the foreseen projects and discuss the dates that the project could be undertaken.

Stellar mass / Infrared Luminosity relation out to redshift ~7

The goal of the project is to use a sample of ~300k star- forming galaxies in order to find the average values of the IR luminosity as a function of the stellar mass. The student will bin the data in redshift and stellar mass and stack in the far-IR Herschel and JCMT maps. The resulting relations will be used in order to derive the average values of the IR luminosity for the vast majority of the star-forming galaxies population, for which it cannot be detected directly.

Supervisor:	Maciej Koprowski (mkoprowski[at]fizyka.umk.pl)
Time:	1 – 29 June 2022

^ ^ ^ ^ ^

PC-based control of Kinova Gen3 redundant robot arm using Linux

The mission is to explore low-level control features of Kinova Gen3 robot arm using Kortex and ROS Kortex Api in Linux. Project requires C++ experience. The plan of the work is as follows:

1. Software installation and configuration:
1. Setting up Ubuntu operating system.
2. Installation of required packages making it possible to use Kortex API.
2. Testing communication with a robot.
3. Exploring Github examples.
4. Implementation of grabbing objects depending on geometric size and color using color and depth streams.
5. Implementation of low-level position control and acquisition of robot arm data using 1 kHz loop.
6. Technical report containing information from all stages of project should be created.

Supervisor:	Tomasz Tarczewski (ttarczewski[at]umk.pl)
Co-supervisors:	Mateusz Tejer
Time:	1 July – 16 September 2022 (to be agreed)

^ ^ ^ ^ ^

Mining software repositories: Learn to rank approaches for bug localization problem

Software bugs occur in the development cycle of nearly all of software projects and can cause severe problems. Information about software bugs is often delivered by users, who submit bug reports containing details about encountered defects. The goal of developers is then to triage, reproduce and fix reported bugs, based on given information. To fix the bug the developer has to find relevant files that need to be changed; such process is called bug localization. Bug localization process can be improved by automatically suggesting which files require fixing. This can significantly decrease the time to fix and reduce the costs of software construction and maintenance. Usually, it is a specific form of Information Retrieval or Learn to Rank process, where we treat a bug report as a query, and a software repository as a collection of documents.

During this project Mining Software Repositories (MSR) will be used to answer some of these questions. MSR concentrates on analysing rich data sources used in software development, like version control repositories, bug tracking tools, mailing lists and other, with the goal to uncover interesting and usable information that can help to improve software development. Good programming skills are required, but prior experience in MSR is not a must. We will provide tutorials on relevant topics at the beginning of the project.

Tasks:

1. Learn basics of MSR including used techniques.
2. Learn basics of cluster computing.
3. Help with the creation of new scalable feature extraction code and experiments.

Supervisor:	Piotr Przymus (eror[at]mat.umk.pl)
Time:	15 June – 26 July 2022 (to be agreed)

^ ^ ^ ^ ^

Mining software repositories: Learn to rank approaches for code reviewer recommendations problem

Software quality is one of the main concerns of software engineering. There are numerous approaches to ensure this including peer code reviews. This practice helps to obtain better code quality, test coverage and lower number of bugs. Peer code reviews are time-consuming activities and require good cooperation of both the reviewer and the author of the code. Waiting a long time for a review could be discouraging. Thus, quick and accurate reviewer selection is on of the key success factors of peer code reviewing.

One solution is the automation of the code reviewer recommendation. This problem has been intensively studied. Current approaches to the problem are based on the similarity of various entities that quantify developers and the changed code and ranking using Information Retrival or Learn to Rank.

During this project Mining Software Repositories (MSR) will be used to answer some of these questions. MSR concentrates on analysing rich data sources used in software development, like version control repositories, bug tracking tools, mailing lists and other, with the goal to uncover interesting and usable information that can help to improve software development. Good programming skills are required, but prior experience in MSR is not a must. We will provide tutorials on relevant topics at the beginning of the project.

Tasks:

1. Learn basics of MSR including used techniques.
2. Learn basics of cluster computing.
3. Help with the creation of new scalable feature extraction code and experiments.

Supervisor:	Piotr Przymus (eror[at]mat.umk.pl)
Time:	15 June – 26 July 2022 (to be agreed)

^ ^ ^ ^ ^

Fixed points and convergence analysis of deep autoencoders

Deep learning based autoencoder has been proposed recently as a promising, and potentially disruptive approach to compresses angular power spectra in massive MIMO systems and later reconstruct the input spectra from the compressed signals. In this context, the nonnegative neural networks are of special interest, especially considering the recent rise of the state-of-the-art analog neural networks, which use conductance (nonnegative by the laws of physics) of programmable resistors as network's weights. Neural networks with nonnegative inputs and nonnegative parameters can be recognized as monotonic and (weakly) scalable functions within the framework of nonlinear Perron- Frobenius theory.

The aim of this project is to derive an algorithm solving the neural network learning problem using fixed point theory for networks applicable in wireless communications. In particular, we will be interested in weakening the assumptions about the monotonicity or scalability of the interference mapping, while maintaining convergence to the fixed point of such mapping. The main goal is to study the convergence and characterization of fixed points of this network. Good programming skills are required. We will provide tutorials on relevant topics at the beginning of the project.

The research tasks are as follows, prepare experiment code for:

1. simulations of angular power spectrum functions,
2. selected state-of-the-art deep learning based autoencoders (using existing methods in library),
3. create own implementation of autoencoder for power spectra reconstruction and study its convergence properties

Supervisor:	Krzysztof Rykaczewski (mozgun[at]mat.umk.pl)
Time:	June – July 2022 (to be agreed)

^ ^ ^ ^ ^

Partial differential equations and probability theory - how are they connected?

Our goal is to exhibit the relation between two seemingly distant fields of mathematics: partial differential equations (PDEs) and probability theory. We focus on probabilistic representations of solutions to concrete PDEs (mostly diffusion equations with various boundary conditions). This will lead us, among others, to celebrated Feynman-Kac formula. Our task is to provide such representations, to understand their meanings and indicates some applications.

Supervisor:	Tomasz Klimsiak (tomas[at]mat.umk.pl)
Co-supervisors:	Maurycy Rzymowski
Time:	June 2022 (to be agreed)

^ ^ ^ ^ ^

How to describe the movement of a particle - Brownian motions and stochastic equations

Our goal/task is to provide a rigorous construction of a Wiener process which is a mathematical description of Brownian motions. We will also refer to the Einstein and Smoluchowski models. The second goal/task is to introduce Itô's integral and Itô's formula and present their applications. Our goal will also be to learn to use this formula freely. Finally, we introduce stochastic differential equations and provide basic existence and uniqueness results. This will lead us to the Fokker- Planck equation but on the level of particles (diffusion process). Our task/goal will be providing methods which from the empirical data allows us to describe coefficients of underlying diffusion.

Supervisor:	Tomasz Klimsiak (tomas[at]mat.umk.pl)
Co-supervisors:	Maurycy Rzymowski
Time:	June 2022 (to be agreed)

^ ^ ^ ^ ^

Channelling silver nanowires – how to create a biosensor

The objective of this project is to observe real-time conjugation of photoactive proteins with silver nanowires (AgNWs) injected in a microfluidic channel via a syringe pump. Proper functionalization of a substrate will facilitate attachment and immobilization of AgNWs. The student will be involved in synthesis of metallic nanoparticles, injection of AgNWs into a microchannel, as well as demonstration of fluorescence detection using advanced microscopy setups. In this experiment, the students will learn details of fundamental interactions in the nanoworld, such as metal-enhanced fluorescence, fluorescence quenching, conjugation or surface modifications of metallic nanoparticles.

In this project we will test various protocols of surface modification of glass coverslips to show the influence on conjugation between AgNWs and photoactive proteins. On the other hand, we will vary the protein concentration to demonstrate detection of single proteins in real-time. The long-term aim of this experiment is not only to better understand fundamental interactions, but also to test designed architectures for biosensing. We expect that using AgNWs will improve the speed and the limit of detection of photoactive proteins.

Particular aspects to be studied within the TSSP project:

• Synthesis of silver nanowires
• Surface modification of AgNWs
• Conjugation between AgNWs and photoactive proteins
• Work with microfluidic channels
• Experiments using wide-field fluorescence imaging microscopy
• Fluorescence detection of photoactive proteins in real-time down to the single molecule level

Supervisor:	Sebastian Maćkowski (mackowski[at]fizyka.umk.pl)
Time:	July – August 2022 (to be agreed)

^ ^ ^ ^ ^

Thermal transport properties of doped InAlAs/InAs superlattice for thermoelectrics applications

The aim of this project is introduced student to well-known methods: photothermal infrared radiometry and thermoreflectance. Student will do measurement and analysis of the results.

Supervisor:	Michał Pawlak (mpawlak[at]umk.pl)
Co-supervisors:	Ameneh Mikaeeli
Time:	1 July – 29 July 2022

^ ^ ^ ^ ^

Nonlocal optical processes

In recent years, optical communication through metallic nanostructure has attracted significant attention. Silver or gold nanostructures allow the light to be guided beyond the diffraction limit, which opens new possibilities for further miniaturization and applications of optoelectronic devices. The project's main objective is the application of surface plasmon polaritons (propagating in a single silver nanowire) to control the optical properties of distant nanoparticles.

The following tasks are to be achieved:

• introduction to fluorescence microscopy
• confocal microscopy of single nanostructures
• remote optical processes controlled by surface plasmon polaritons

Supervisor:	Dawid Piątkowski (dapi[at]fizyka.umk.pl)
Time:	August / September 2022 (to be agreed)

^ ^ ^ ^ ^

Dimensionality Reduction in Molecular Dynamics

Data from molecular dynamics simulations (representing, for instance, DNA, proteins, soft matter) is known to be high- dimensional and difficult to analyse. In this project, we want to construct a low-dimensional representation of simulation trajectories in a way facilitating analysis and our understanding of the studied physical process. The aim of this project is to study low-dimensional representations of data from molecular dynamics simulations using a chosen unsupervised learning algorithm (e.g., PCA, autoencoders, t- SNE).

The student should perform the following tasks:

1. Choose a dimensionality reduction algorithm or propose an original method.
2. Test the method on several molecular dynamics data sets that are available free of charge.
3. Analyse if the obtained low-dimensional representations correctly capture the studied physical process related to the chosen data set (e.g., conformational changes, ligand dissociation).
4. Write a report or a publication summarising the results. The project will be performed under close supervision and guidance of the principal investigator.

Supervisor:	Jakub Rydzewski (jr[at]fizyka.umk.pl)
Time:	1 July – 1 August 2022

^ ^ ^ ^ ^

Superluminal light propagation? Theoretical investigation of how different measuring schemes of group velocity in atomic media can influence the result and why

The project analyses the group velocity of an electromagnetic pulse propagating in a one-dimensional optically dressed medium. The medium consists of three- level quantum systems in the ladder configuration, while the pulse is a classical electromagnetic wave. The dynamics are described with the set of Bloch-Maxwell equations. The project focuses on the regime in which superluminal propagation can in principle be achieved.

Until now, investigations reveal that the pulse's group velocity may depend on how it is measured: directly inside the ensemble (hardly accessible in experiments) or between its ends (usually measured in experiments). This discrepancy can be related to the nonzero lifetime of excited energy levels delaying the energy transfer. In consequence, we suspect in some cases a superluminal pulse may not be classified as such due to limitations of measurement techniques.

The goal of the project is to develop theoretical techniques to identify the key physical mechanisms responsible for the discrepancy between group velocities measured in different ways and to understand the conditions in which it appears.

For this purpose, we provide step-by-step guidance and a set of tools based on a numerical solver implemented in the Python3 programming language. The solver simulates the pulse dynamics and includes several methods for group velocity calculations.

Tasks for the participant are:
Week 1: Familiarising with the theory and the numerical solver
Learning about the theoretical aspects of group velocity calculations in the particular case of three-level quantum systems. Familiarising with the numerical solver of Bloch-Maxwell equations.
Week 2: Simulations of pulse propagation in predefined ranges of parameters
Simulations of propagation of the pulses through the medium for various conditions and selection the most promising results for the further investigation.
Week 3: Results analysis 1
Identification of parameter regimes in which the discrepancy occurs and identification of key parameters responsible for it.
Week 4: Results analysis 2
Quantitative characteristics of mentioned differences between group velocities calculated in different ways.

Supervisor:	Karolina Słowik (karolina[at]fizyka.umk.pl)
Co-supervisors:	Piotr Gładysz
Time:	1 month within the period mid July – mid September 2022

^ ^ ^ ^ ^

Atomic defects on graphene flakes: modelling optical response

The project analyses optical properties of graphene flakes modified by atomic defects: additional atoms attached to selected carbon sites.

Until now, we have shown that even single defect atoms can significantly influence the optical response of the system. The underlying mechanism is related to the geometry: a defect attached to a spacially symmetric flake breaks the symmetry and modifies the symmetries of the system's quantum mechanical eigenstates. In consequence, this impacts the optical selection rules, particularly inducing new previously forbidden transitions.

This project will investigate the roles of the defect parameters, such as its transition dipole moments, orbital energies, or spatial location and orientation with respect to small, triangular, armchair-edged graphene flakes. The goal is to characterise the defect's impact in terms of the system's modified energy landscape and absorption spectrum.

For this purpose, we provide step-by-step guidance and a set of tools based on a numerical solver implemented in the Python3 programming language. The solver has implemented the tight-binding model for the flake augmented with the light-matter interaction Hamiltonian. It allows solving the eigenproblem of the flake (with defects) and characterising its optical response for a said flake geometry/defect parameters.

Tasks for the participant involves:
Week 1: Familiarising with elements of the tight-binding theory and the numerical solver.
Week 2: Calculations of energy spectra and eigenstate characteristics in predefined ranges of the system parameters.
Week 3: Simulations of absorption cross-sections in predefined ranges of the system parameters.
Week 4: Results summary: identifying the conditions in which the impact of the defect is strongest. Possibly, characterising different regimes or different types of flake influence.

Supervisor:	Karolina Słowik (karolina[at]fizyka.umk.pl)
Time:	1 month within the period mid July – mid September 2022

^ ^ ^ ^ ^

Generation of non-diffracting beams via computer-generated holograms and evaluating its properties through scattering media

Light diffraction is a result of its deviation from a rectilinear propagation. A class of light beam that preserves its shape and beam width during propagation over a considerable distance (larger than Rayleigh distance defined for Gaussian beam) is termed as non-diffracting beams. A strong scattering and attenuation happen when the light propagates through biological tissues. Therefore, the non-diffracting beams provide a possible solution to overcome this scattering and attenuation due to their shape-preserving and self-healing properties during propagation. Non-diffracting beams can be generated by tailored fibers, axicon, metasurface, and spatial light modulator. Spatial light modulators are usually used to create non-diffracting beams because of their programmable property. In order to generate a non-diffracting beam, a spatially designed computer-generated hologram has to be displayed by SLM, which changes the beam profile of the irradiance by the encoded information of the hologram. The main objective of this project is to generate non-diffracting beams and test their properties through scattering media.
The main research goals are:

1. Simulation of computer-generated hologram for SLM
2. Building an optical set-up to test the generated beams
3. To test the properties of generated beams after passing through scattering media

Supervisor:	Spozmai Panezai (nmpanezai[at]yahoo.com)
Time:	June – July 2022 (to be agreed)

^ ^ ^ ^ ^

Strongly interaction regime in the context of density functional theory

Kohn-Sham (KS) density functional theory (DFT) is the most used electronic structure computational approach for molecular and solid-state systems. Its accuracy depends on the choice of the approximation for the exchange-correlation (XC) functional which, at the highest-rung of the Jacob’s ladder, involves all the occupied and virtual KS orbitals as well as the eigenvalues i.e. double-hybrid functionals. This type of functional was successfully applied in systems e.g. to model weakly interacting molecular systems, or to calculate properties such as bond-length alternations or charge-transfer excitations. Despite their success, they still are unable to properly describe systems with strong interaction characteristics. Within this project, we will try to solve this issue by proper inclusion of renormalized second-order correlation energy expression in the double hybrid functional form.

Research tasks:

• preparation of computational setup
• test of the accuracy of new expression for strongly-correlated problems
• analysis of the results

Supervisor:	Szymon Śmiga (szsmiga[at]fizyka.umk.pl)
Time:	1 – 31 July 2022

^ ^ ^ ^ ^

So similar and yet different - examining potassium channels using molecular dynamics methods

Inwardly-rectifying potassium (Kir) channels contribute to the regulation of electrical excitation in many cell types. Despite their great similarities, their structure varies slightly from one system to another. Therefore, such channels can be present comprehensively in the body yet have strictly specialised functions in different tissues.

This project aims to compare Kir family channels at the sequence, structure, and dynamics level. The participant in the summer research program will learn to build models of molecular systems based on existing structures and homology models. We will use bioinformatics tools to compare sequences and identify essential motifs and ligands binding sites. Then, we will use classical and enhanced molecular dynamics methods to determine hot spots in the systems and understand their structural properties.

Supervisor:	Katarzyna Walczewska-Szewc (kszewc[at]fizyka.umk.pl)
Time:	July 2022 (to be agreed)

^ ^ ^ ^ ^

External cavity diode laser

This project aims to assemble an external cavity diode laser and characterize its performance. The laser will be built from the components previously purchased and provided to the student. The student will select an appropriate diffraction grating and lens to collimate the laser beam and install the Peltier module for thermal stabilization. Then the laser diode light will be coupled with the external cavity in the Littrow configuration. The next task will be to measure the lasing threshold. In the final step, the student will characterize the laser beam parameters and measure the frequency stability of the laser. All activities related to the project will be performed by the student under the guidance of the project supervisor.

Supervisor:	Marcin Witkowski (marcin_w[at]umk.pl)
Time:	June 2022 (to be agreed)

^ ^ ^ ^ ^

Thermal stabilization of a fibre laser system

This project aims to design and assemble an active thermal stabilization platform for a system of two fibre- coupled lasers: a 1062 nm laser module and a 1W1062 nm optical amplifier. The purpose of thermal stabilization is twofold: firstly, as the laser frequency is temperature-dependent, stable temperature improves the frequency stability. Secondly, an actively stabilized thermal platform enables more efficient heat dissipation from the optical amplifier.

For heat dissipation, the Peltier modules will be used. The student will select the appropriate Peltier elements according to the required heat transfer. The next task will be to design an optimal geometry of the platform, which will then be manufactured by the CNC. In the final step, the student will write a program for active control on the Peltiers’ current and optimize the thermal parameters of the assembled system. All activities related to the project will be performed by the student under the guidance of the project supervisor.

Supervisor:	Marcin Witkowski (marcin_w[at]umk.pl)
Time:	June 2022 (to be agreed)

^ ^ ^ ^ ^

Solvent Effect in dye-sensitized solar cells utilizing red algae-based dye

Dye-sensitized solar cell (DSSC) is one of the low cost solar cell technologies. It was developed as a substitute for solar cell made of silicon, which uses a lot of energy and has a high production cost. DSSCs can produce photocurrent using fewer and less expensive materials. DSSCs are environmentally friendly and do not present any risks associated with the use of hazardous materials. In the assembly of DSSC, the dye plays an important role in harvesting solar energy and converting it to electrical energy with the aid of a TiO2 semiconducting photoanode. Therefore, the cell performance is mainly dependent on the type of the dye used. Natural dyes present several advantages as an easy extraction without polluting the environment and diversity in colors compared to synthetic dyes.

The objective of this project is to prepare the dye-sensitized solar cell by sensitizing TiO2 with the natural dye extracted from red algae, using different solvents. Studying the effects of different extracting solvents on DSSC is one way to improve the efficiency of the cell. For this purpose, we will perform the main optical properties (photoluminescence and UV–visible absorption) that the dye should confirm. The suitability of the solvent to extract the dye can exhibit an extended light absorption range in the UV–visible region. Then, it will go on investigating the efficiency of each DSSC from the (I-V) characteristics. As a result, we expect to increase the performance of DSSC based on red algae by at least 30% by using the appropriate solvent.

Supervisor:	Anna Zawadzka (azawa[at]fizyka.umk.pl)
Time:	July 2022

^ ^ ^ ^ ^