|
|
Binary codes induced by the line graph of an n-dimensional cube
Janovský, Tomáš ; Žemlička, Jan (advisor) ; Šťovíček, Jan (referee)
This paper deals with the binary codes from the line graph of the n-cube, then the codes of the design, which is defined by the vertex graph of the n-cube, and finally the dual codes of these codes. The first chapter is devoted to an introduction to the topic and the necessary concepts from the theory of linear codes, graphs and designs are defined there. The second chapter is devoted to the construction of the aforementioned codes, the description of their basic parameters such as the dimension and the Hamming distance of the code, and finally the description of the generating matrices of these codes. Finally, the last chapter deals with the dual codes of the codes constructed in the second chapter especially again the description of the Hamming distance.
|
|
|
Graph neural networks and deep reinforcement learning in job scheduling
Bratko, Maroš ; Holeňa, Martin (advisor) ; Seidelmann, Thomas (referee)
The priority dispatching rule (PDR) is a greedy heuristic algorithm used to obtain approximate solutions to the NP-hard job scheduling problem (JSP). The manual design of PDRs requires domain knowledge to achieve good results. Recently, deep reinforcement learning has been used to auto- mate the process of designing PDRs, where PDRs are formulated as a Markov Decision Process exploiting graph representation of JSP, and a graph neural network (GNN) selects the operations to be dispatched. In this thesis, we present the overview of five models published in literature with their source code publicly available on GitHub and experimentally compare their per- formance on three different variants of JSP. Our experiments show that the choice of input features, e.g., the amount of remaining work, significantly affects the model's performance regardless of the GNN architecture. This suggests that the feature selection is essential for learning high-quality PDRs.
|
|
|
Tactics and strategies of the opposition - obstructions as a political weapon
Pchálková, Tereza ; Brunclík, Miloš (advisor) ; Jágr, David (referee)
The thesis focuses on the phenomenon of obstruction in the Chamber of Deputies of the Parliament of the Czech Republic in the eighth and ninth electoral terms, which is specific to the exchange of political parties in the executive and opposition roles. Theoretically, the thesis is set in the themes of opposition and obstruction. For this, the literature is primarily based on Miroslav Novák's article Political Party Systems and Basic Models of Opposition, Michal Kubát's publication Political Opposition in Theory and Central European Practice, Jan Wintr's book Czech Parliamentary Culture, and the Rules of Procedure of the Lower House of Parliament, resp. its annotated version by Jindřiška Syllová and colleagues. During theoretical research, the author came across an insufficient interpretation of the term and therefore came up with her own definition. The author of the thesis set the research question: How does the obstruction strategy of political parties of the opposition differ? and attempts to answer it through qualitative research. The subject of analysis in the thesis are case studies of two state budget bills, namely for 2021 and 2024, for which all three legislative readings were examined. The incidence of filibusters is then tracked through four codes: length of debate, factual remarks,...
|
|
| |
|
|
Detection and Correction of Silent Errors in Pipelined Krylov Subspace Methods
Hercík, Jakub ; Carson, Erin Claire (advisor) ; Tůma, Miroslav (referee)
This thesis focuses on the problem of silent error detection in the pipelined predict- and-recompute conjugate gradient (Pipe-PR-CG) algorithm, a pipelined Krylov subspace method for solving linear systems with a symmetric positive definite matrix. The theory of silent errors and conjugate gradient variants is introduced, and the structure of Pipe- PR-CG is subsequently utilized in rounding error analysis to derive criteria for silent error detection based on bounds of several quantities computed in finite precision arithmetic. The efficacy of the criteria is then tested in a robust numerical experiment, and a fault- tolerant version of the algorithm is introduced. Additionally, the sensitivity of Pipe-PR- CG to silent errors is examined. Codes in the Python programming language which were used for the main experiments and figures presented in this thesis are also provided. 1
|
|
|
Evaluation of the recycling potential of industrial waste in the Czech Republic
Kuba, Josef ; Pilnáček, Vojtěch (advisor) ; Innemanová, Petra (referee)
This bachelor's thesis deals with the determination of the amount and type of unused industrial waste in the Czech Republic, the evaluation of the waste recycling potential and subsequent methods of its recycling. Part of the work was the analysis of data from the ISOH working database (PDISOH) from 2019. The analysis focused on the amount of industrial waste production in the Czech Republic and then examined the amount of industrial waste that was disposed of by landfill or incineration (without energy use). The analyzed wastes were divided in the chapters according to the groups of their recycling possibilities. Each chapter contains a description of the methods of recycling waste groups and further refers to tables created by the author describing the names of the wastes, their waste codes, how much of the given waste is removed annually by landfilling and incineration (without energy use), the recycling potential, the available recycling potential that separates the specific wastes, which can be recycled using the described methods. The chapters also contain temperature maps of the Czech Republic, drawn up by the author, with the locations of the described industrial wastes shown in the tables. The results showed that up to 45,37 %, accordingly 3 432 866,8 tons of the total production of 7 564 897,35...
|
|
|
Modernization of the manipulator control unit for the placement of magnetic points
Adámek, Adam ; Appel, Martin (referee) ; Formánek, Martin (advisor)
This work is being taken up by upgrading the manipulator control unit. It is a so-called pick and place minupulator with CNC machine elements that places magnetic points on a film. On the manipulator, Arduino MEGA control unit with RAMPS v1.3 shield and control computer will be supplemented with Raspberry Pi microcomputer and a camera that will recognize QR codes on individual foils and accordingly select the appropriate program for placing the aformentioned magnetic points. The initial state of the manipulator will be assessed, its control unit and the pneumatic the system responsible for the transmission and placement of the magnetic points. The next section will describe the design of the new control, the use of the open-source G-code interpreter PyCNC, the setup of this software and also its incorporation with the already existing elements on the manipulator such as limit switches and solenoid valves. The last part of the thesis deals with the comparison of the achieved result, whether it is the functionality of the manipulator and its improvement, or the aesthetic aspect of installing the Raspberry Pi and the camera module on the manipulator. Also, the turn-on time and the duration of one cycle of placing the magnetic points on the film of the manipulator will be compared.
|
|
|
Zpětná extrakce Stateflow diagramu z kódu v jazyce C
Gavenda, Daniel ; Smrčka, Aleš (referee) ; Fiedor, Jan (advisor)
Simulink diagrams are widely used in industry, where they are used to specify control systems. The specific block used in them is the Stateflow diagram. The specified systems are often critical from the point of view of security, therefore the question of the correctness of the implementation of these systems is important. In this work, we will focus on the design, description and creation of a tool for the reverse reconstruction of the Stateflow diagram from the optimized code in the C language in order to prove the functional equivalence between the model and its corresponding code. Next, a created set of models representing models with Stateflow diagrams and codes generated from them is created, using several optimizations. The created tool will be tested on this set, it can also be used to evaluate other Stateflow analyzes and/or C code.
|
|
|
Design of rationalization of distribution block production
Kobza, Tomáš ; Zemčík, Oskar (referee) ; Sliwková, Petra (advisor)
This thesis focuses on the proposal of rationalization of the production process of the specified part from the company WT-PRO Czech s.r.o. The introductory part is devoted to the description of the specified part and the analysis of the existing production process, focusing mainly on machining operations. Theoretical part is focused on CAD/CAM systems, machining technologies - milling, drilling and threading, which are essential for the production of the component. It also focuses on tooling materials and chucks. In the practical part, the rationalization of the production process is proposed, which concerns two machining operations. A machining centre is specified for these operations, for which new clamping, new tooling, operating instructions and NC codes have been designed in GibbsCAM. The resulting design is described in the new manufacturing process. Finally, a techno-economic evaluation was carried out.
|
|
|
Accelerometer data classification within the patient ECG record
Kindl, Zdeněk ; Ředina, Richard (referee) ; Bulková, Veronika (advisor)
The subject of the bachelor's thesis is the classification of patient accelerometric data. The aim is to improve the clarification of pathologies in the ECG signal. The classification is performed on data measured by the Bittium Faros 180L device. A custom database of movements was created. Patient data is processed using a recurrent neural network, which classifies the movements into three basic groups: resting activity, moderate activity, and high activity. The output is a file with movement annotations. The thesis includes a description of neural networks, data, data processing, and the creation of the neural network with codes.
|
guest ::
