Ofertes de projectes

Virtual reality is becoming a tool for helping patients with several illnesses (strokes, frozen shoulder...) to engage in the rehabilitation exercises.

The purpose of this project is the creation of a serious game that helps patients to engage in rehabilitation exercises for the frozen shoulder.

If you are interested, do not hesitate to ask for more information. But please, before doing so, check my "Prospective students" tab on my webpage.

A common trait of current computing systems is that their internal data communication has become a fundamental bottleneck. Traditional interconnects are insufficient and threatening to halt progress unless fast and versatile communication alternatives are developed.

In the NaNoNetworking Center in Catalunya (N3Cat, www.n3cat.upc.edu), we are investigating the use of ultra-short range wireless communications an alternative to current interconnects. In our EU project EWiC, we aim to emulate chip-to-chip wireless communications within a computer. For that, we will emulate the behavior of a multi-chip CPU with FPGAs, and connect them wirelessly using Software-Defined Radios (SDR). 

N3Cat is looking for students wanting to acquire hands-on experience in the programming and interconnection of FPGAs, and to participate in a cutting-edge project to interconnect these FPGAs wirelessly. In more detail, the objectives of the thesis are:

- To help establishing the emulator and be able to simulate its behavior.
- To work in the implementation of adequate communication protocols for the integration of wireless interconnects in the emulated multi-FPGA system.
- To assess the performance of the system via measurement campaigns.

This Bachelor Thesis will take place at the Departament d'Arquitectura de Computadors (DAC) in collaboration with the STAR group at the Barcelona Supercomputing Center (BSC).

The System Tools and Advanced Runtimes (STAR) group at BSC focuses on research crossing multiple software layers from OS, runtimes, and low-level APIs, to programming models, tools, and applications. We run traditional HPC, AI, and Data Analytics applications on massively parallel HPC and Cloud platforms.

Arithmetic operations derived from stencils calculations often benefit from compiler optimisations and automatic vectorisation in the cases where the compiler has enough information and the data structures are adequately aligned and traversed.  While classical applications rely on double precision arithmetic operations with 64-bit floating-point precision, CPU manufacturers have recently begun introducing native hardware support down to 16-bit precision. 

The student will be tasked to adapt single-precision computational kernels of a variety of applications including, for example, the N-body problem, the heat equation, and some computational stencil used in computational fluid dynamics problems to their mixed-precision counterparts. Next, he or she will vectorise the kernels with the aid of OpenMP pragma directives and identify other optimisation opportunities like loop unrolling or cache blocking.  Finally, the student will asses the effects of mixed-precision and, in particular, any potential accuracy losses. 

The student will have the support from researchers within the STAR group, which will provide them guidance throughout the duration of the thesis.  The GTP models will be evaluated using the computational resources of our group.

This Bachelor Thesis will take place at the Departament d'Arquitectura de Computadors (DAC) in collaboration with the STAR group at the Barcelona Supercomputing Center (BSC).

The System Tools and Advanced Runtimes (STAR) group at BSC focuses on research crossing multiple software layers from OS, runtimes, and low-level APIs, to programming models, tools, and applications. We run traditional HPC, AI, and Data Analytics applications on massively parallel HPC and Cloud platforms.

Some popular iterative methods employed in the numerical resolution of sparse linear systems of equations are the Jacobi and Gauss¿Seidel (GS) algorithms. These techniques are often combined with more sophisticated Krylov subspace methods (KVMs) such as the conjugate gradient (CG) and the biconjugate gradient stabilized (BiCGStab) algorithms. These KVMs rely primarily on sparse matrix-vector (SpMV) calculations and can achieve better convergence rates than Jacobi and Gauss¿Seidel.

The student will be tasked to adapt several sparse matrix encodings for the SpMV operation. These formats include: coordinate format (COO), compressed row storage (CRS), lower-diagonal-upper (LDU), modified sparse row (MSR), lower-diagonal-upper (LDU), and lower-diagonal-upper CSR (LDUCSR). Next, he or she will evaluate various iterative methods on both CPU and GPU architectures. Finally, the student will asses the advantages and disadvantages of the encodings for each particular architecture.

The student will have the support from researchers within the STAR group, which will provide them guidance throughout the duration of the thesis. The iterative methods will be evaluated using the computational resources of our group.

Recent advancements in nanotechnology have enabled the development of means for sensing and wireless communications with unprecedented miniaturization and capabilities, to the point that they can be introduced into the gastrointestinal tract inside a pill or into the bloodstream in the form of passively flowing nanomachines. 

This opens the door to the idea of intra-body communication networks, this is, a swarm of nanosensors inside the human body that use communications to coordinate their actions to sense and localize specific events (lack of oxygen, biomarkers, etc). This can lead to the development of applications such as continuous monitoring of diabetes, detection and localization of cancer micro-tumors, or early detection of blood clots. These possibilities are currently investigated by our team at the N3Cat (www.n3cat.upc.edu).

In this context, We are looking for excellent and self-motivated individuals who are eager to develop ideas that will have a positive societal or socio-economic impact. In particular, the students will work on ONE of the following areas:

• Developing means of zero-power communications between nanosensors inside the human body.
• Designing protocols for communications from nanomachines inside the human body and implants or antennas in the skin surface.
• Studying mobility inside the gastrointestinal tract and human bloodstream towards creating mobility models for accurate simulation of intra-body communications.
• Dimensioning the system (inferring the required nanomachine density) to achieve a particular sensing goal/application.
• Developing AI schemes for the localization of events inside of the human body.

Seeing that not all neural networks fit to all problems, and that relational data is present in a wide variety of aspects of our daily life, the main focus of this thesis in N3Cat (www.n3cat.upc.edu) and BNN-UPC (www.bnn.upc.edu) is to explore the possibilities of the Graph Neural Networks (GNNs), whose aim is to learn and model graph-structured relational data. We are looking for students willing to study the uses, architectures, and algorithms of GNNs. To this end, the candidate will work on ONE of the following areas:

• Uses: Applying GNNs in emerging application areas, including but not limited to (1) electroencephalogram (EEG) analysis for Alzheimer's disease detection, epilepsy classification, motor imagery; (2) acceleration of the compilation of quantum computing algorithms
• Architectures: Investigating ways to accelerate the processing of GNNs in multiple computing platforms (CPU, GPU, accelerators).
• Algorithms: Developing meta-learning data-driven models to estimate the accuracy of a GNN for a given graph, without training, through synthetic graph generation and correlation analysis.

Gradient descend based RBM learning [1] is a computationally expensive procedure that requires the computation of a probability normalization constant (called the Partition function) that involves the computation of an exponentially large number of terms. In order to bypass that, different approximations exist, the most celebrated one being the Annealed Importance Sampling algorithm [2], that can be parallelized in one of its dimensions.

The goal of this project ir to implement the Annealed Importance Sampling for Restricted Boltzmann Machine (RBM) using python/CUDA, and compare the performance against a standard CPU implementation.

[1] "Training restricted Boltzmann machines: An introduction",
Asja Fischer and Christian Igel, Pattern Recognition 47, 25-39 (2014).

[2] "Annealed Importance Sampling", Radford M. Neal, Statistics and Computing 11, 125-139 (2001).

Public datasets on the usage of flagship high-performance computing (HPC) platforms are fundamental to test new scheduling algorithms. But, because these platforms are critical national infrastructure, this data cannot be published. In this project, we will apply the differential privacy anonymization technique to HPC logs. This is a cutting-edge technique of growing interest within the research AI community and currently used by the US Census and the Swiss
Census. Differential privacy adds noise to the dataset in a controlled manner in order to ensure that the output statistics are contained within bounds. In this project, the student will learn and apply differential privacy to public HPC datasets and test how the results are altered by the added noise. The outcome of this research will provide a solution for HPC admins to publish their data and the methodology for the community to treat them.

Computing systems are ubiquitous in our daily life and have transformed the way we learn, work, or communicate with each other, to the point that progress is intimately tied to the improvements brought by new generations of the processors that lie at the heart of these systems. 

A common trait of current computing systems is that their internal data communication has become a fundamental bottleneck. Traditional interconnects are insufficient and threatening to halt progress unless fast and versatile communication alternatives are developed.

In the NaNoNetworking Center in Catalunya (N3Cat, www.n3cat.upc.edu), we envisage a revolution in computer architecture enabled by the use of wireless communications (and other unconventional interconnect technologies) inside computer chips, both classical and quantum. Of course, this vision comes with its own set of challenges since we need to figure out the best way to architect these systems with the new capabilities of the interconnects. In this direction, the candidate will work on one of the following areas:

• Analysis of the communications workload in nowadays computer chips and development of network-on-chip architectures to serve such traffic optimally.
• Development of chiplet-based architectures in the classical domain in cache-coherent based systems, delving into the cache coherence protocol and the interface with the wireless interconnect.
• Development of a supercomputer-in-a-package, a chiplet-based architecture based on message passing, delving into how to speed up the MPI primitives with wireless communications.
• Study of scalable quantum computing systems enabled by wireless classical communication across the chips of the quantum computer and inside a cryogenic refrigerator.

Quantum computers promise exponential improvements over conventional ones due to the extraordinary properties of qubits. However, a quantum computer faces many challenges relative to the movement of qubits which is completely different from the movement of classical data. This thesis delves into these challenges and proposes solutions to create scalable quantum computers enabled by quantum communications, following the current European projects at N3Cat (www.n3cat.upc.edu) on scalable quantum computing.

The interested candidate will work in a group of several PhD students and in collaboration with Universitat Politècnica de València, working in ONE of the following areas:

• Designing classical (bit) and quantum (qubit) communication protocols and network architectures for communications inside a quantum computer.
• Assisting in the modeling of quantum short-range communications within our simulation framework, which is based on NetSquid.
• Developing qubit mapping, partitioning, and routing for scalable quantum computers.
• Study the use of AI to model quantum computers or develop qubit mapping and routing schemes.
• Study the potential adaptation of the existing quantum computing architectures to specific quantum machine learning algorithms (QAOA, VQE, others)

To evaluate the impact of the AI assistant, we will conduct a user study that examines both its effectiveness in supporting participants and its overall usability within the VR simulation.

High-Performance Computing (HPC) plays a crucial role in scientific modeling applications that require solving large-scale numerical problems efficiently. PETGEM is an open-source 3D electromagnetic modeling tool developed to solve Maxwell's equations using a high-order vector finite element method. Originally implemented in Python, PETGEM has demonstrated significant flexibility and scalability on HPC systems. However, due to the increasing demand for performance and adaptability to emerging pre-exascale architectures, a complete migration of the codebase to the C programming language is underway.

This project focuses on the refactoring of PETGEM's magnetotelluric (MT) modeling kernel, rewriting it from Python to C. The objective is to improve computational performance, reduce memory footprint, and ensure compatibility with low-level parallelization frameworks and memory management strategies suitable for modern HPC environments. A strong emphasis will be placed on code verification, ensuring the numerical correctness of the C implementation through comparisons with the original Python version using realistic geophysical models.

The project addresses core areas of Informatics Engineering, including software refactoring, numerical computing, parallel programming, and scientific software engineering. The student will gain hands-on experience in HPC programming practices, C language optimization techniques, and validation workflows for scientific simulations.

We assume a distributed control architecture consisting in a multi-agent system, where agents exchange data and perform inteligent decision-making actions, thus emulating autonomous operation of a 6G network infrastructure. The project will study and compare different strategies to integrate part of the intelligence at the SIEM level, with the aim to improve efficacy and performance of secure autonomous operation methods.

L'objectiu principal d'aquest projecte és desenvolupar un model basat en autòmats cel·lulars capaç de simular la propagació de l'aigua i l'impacte de les inundacions en una àrea geogràfica determinada a causa d'episodis de DANA.

Els objectius específics inclouen:

• Recerca i Fundamentació: Investigar els principis de les DANA, els seus efectes hidrològics i les metodologies existents de modelització d'inundacions. Estudiar a fons la teoria dels autòmats cel·lulars aplicada a la hidrologia i la simulació de fluxos.

• Adquisició i Processament de Dades Geoespacials: Recopilar dades d'elevació del terreny (model digital d'elevacions - DEM), informació de la cobertura del sòl (usos del sòl, vegetació, impermeabilitat) i dades històriques de precipitació (DANA) per a l'àrea d'estudi seleccionada. Processar aquestes dades per adaptar-les al format de l'autòmat cel·lular.

• Disseny i Implementació del Model d'Autòmats Cel·lulars:

  • Definir la quadrícula cel·lular que representi la topografia de l'àrea.

  • Establir les regles de transició que governin el moviment de l'aigua entre cel·les, considerant l'elevació, la resistència del sòl (permeabilitat/impermeabilitat), la velocitat del flux i la interacció entre les cel·les veïnes (regles de Von Neumann o Moore).

  • Integrar un component de precipitació que afegeixi aigua a les cel·les segons les dades de pluja.

  • Incorporar mecanismes per simular la infiltració i l'escorrentia superficial.

• Validació i Anàlisi de Resultats: Comparar els resultats de la simulació amb dades d'inundacions reals (si estan disponibles per a l'àrea d'estudi i es poden adquirir) o amb models hidrològics convencionals. Analitzar els patrons d'inundació, les zones de risc identificades i la sensibilitat del model a diferents paràmetres.

• Visualització: Desenvolupar eines de visualització per representar de manera clara i intuïtiva la propagació de l'aigua, els nivells d'inundació i les àrees afectades al llarg del temps.

This project focuses on improving the performance and usability of an existing Python code used to simulate wave propagation in two dimensions (2D) using three numerical methods: the pseudospectral method, second-order finite differences, and third-order finite differences.

The code is part of the course "Bojos per la Ciència: Simulate the Earth", offered at the Barcelona Supercomputing Center (BSC). This course is included in the Bojos per la Ciència program, coordinated by Fundació Catalunya La Pedrera, which encourages scientific and technological vocations among high school students by providing theoretical and practical sessions led by researchers at leading Catalan research institutions.

Given that the target audience is composed of students around 18 years old, many of whom have limited prior knowledge in computer science, the project aims to enhance the code to facilitate its use during course sessions. This includes optimizing performance where possible, improving code readability and structure, and adding user-friendly features that aid comprehension and interaction during the teaching process.

The project also includes validating the numerical correctness of the improvements, studying key numerical properties such as stability and dispersion, and ensuring the code runs efficiently on high-performance computing platforms like MareNostrum.

This work will provide valuable experience in scientific computing, software optimization, and educational software development within HPC contexts.

Actualmente, una gran parte de los proyectos tecnológicos se desarrollan por diversas empresas y/o instituciones (partners del proyecto). Proyectos en los cuales el valor e impacto de éstos se mide en función de los elementos (IPs ¿ Intellectual Properties) que se desarrollan con el fin de aportar una solución a un problema planteado. La gestión de estos IPs es crítica, no sólo para las empresas, universidades y centros de investigación, sino también para los organismos que financian dichos proyectos. Entender las dependencias tecnológicas (hardware y software) de las soluciones son básicas para poder entender el nivel de madurez de la tecnología, trazar planes estratégicos, de transferencia tecnológica, gestión de conocimiento y, explotación.

En los últimos años, el crecimiento exponencial en todos los ámbitos de la tecnología, la gestión de IPs ha ganado gran relevancia, hasta ser un aspecto crítico de los proyectos. Hasta ahora el modo habitual de operar es completamente manual y totalmente dependiente de las personas. El flujo de trabajo se puede resumir en tres pasos: 1. recopilación de información de los IPs de todos y cada uno de los partners del proyecto, 2. Análisis de la información, y 3. Generación de los informes de seguimiento y estado de los IPs. Todo esto se desarrolla de manera manual y repetitiva, pues se debe iterar en diferentes etapas del proyecto.

En el mejor de los casos, el modo de recopilar esta información es utilizando una plantilla (template) preparada para recoger la información de los partners del proyecto en hojas de cálculo. Este template puede contener macros para seleccionar varias opciones en una misma casilla, que las coloque de forma ordenada, o que permita "input box" dependiendo la casilla seleccionada, entre otras cosas. Con cada nuevo proyecto, se crea un nuevo documento a base de adaptar la plantilla a las necesidades y características propias del proyecto.

La idea del template, tal y como está diseñado, tiene como objetivo simplificar el proceso de recopilación de información y ahorrar tiempo a los partners. Sin embargo, la estructura actual acarrea problemas:

• El formato online es incompatible con macros.

• Compartir la información en formato digital puede generar problemas de seguridad.

• Aún en el caso de enviar el template por mail, las macros pueden ser incompatibles con todos/alguno de los sistemas y/o políticas que utilizan los partners.

• El proceso es manual y repetitivo, lo cual es propenso a errores.

• El proceso no facilita la reutilización de la información de manera eficiente, lo cual dificulta la evolución y comprensión del estado de las IPs, de cara a la explotación de resultados.

Para solventar estos problemas y facilitar el proceso de introducción y, gestión y explotación de los datos, proponemos: Desarrollar una aplicación web que permita:

• tener una interfaz de trabajo genérica para la gestión de IPs (hardware y/o software)

• definir diferentes roles y usuarios, de manera que permita al gestor de proyecto: 1. adaptar la información a introducir, en base a las necesidades de un proyecto concreto -quién y de qué modo- 2. Crear y gestionar usuarios (partners) para que puedan introducir y modificar la información de manera sencilla y segura, 3. gestionar la información de los partners y, si fuera posible 4. Crear dashboards de acuerdo con las necesidades de proyecto y a la información de los partners

• Volcar la información en la base de datos del proyecto.

Esta aplicación (open-source) permitirá una gestión eficaz, eficiente y amigable de la tecnología (IPs) desarrollada en proyectos por los participantes del proyecto y, para los gestores de proyecto. Además, podrá permitir a los evaluadores o revisores del proyecto tener una visión en tiempo-real del estado de los IPs del proyecto. Será una herramienta que les facilite la configuración y adaptación de la información, la introducción de datos, y su posterior gestión y explotación. Todo esto en un entorno amigable y seguro, que garantice la confidencialidad requerida para cada partner y proyecto.

El proyecto será guiado por Marisa Gil, del departamento de AC en la parte técnica, y apoyado por profesionales con experiencia en el área (Vanessa Iglesias, Teresa Cervero, BSC).