2024 Edition

09h30 - 09h45 Welcoming & Opening Session

09h45 - 10h45 Rita Guerra (Former student) - Fourier-type operators, properties and applications

10h45 - 11h15        Coffee-Break

11h15 - 11h45 Paulo Monteiro  - Computational experiments on Lehman matrices, mni and their cores  

11h45 - 12h15 Manisha Jain - Adding concurrency to Quantum dynamic logic

12h15 -12h45         Zita Abreu - Pseudo-MDP convolutional codes

12h45 - 14h30 Lunch-Break

14h30 - 15h30 Daniel Figueiredo (Former student) -  Relation-changing structures (and applications to Health decision-making)

15h30 - 16h00        Marta Maltez - New developments in analysing multivariate compositional data 

16h00 - 16h30 Rui Martins - A C++ library for sensitivity analysis of ODEs 

16h30- 17h00 Chen Liang  - Axisymmetric Bosonic Stars: bifurcations with spherical bosonic stars 

Chairs: Juliana Cunha and Miguel Almeida

Abstract: See here!

Abstract: In this talk, I will introduce a new construction of convolutional codes and examine their erasure correction capabilities. MDP convolutional codes are known for their optimal correction performance when decoding is performed sequentially. However, few constructions of MDP convolutional codes exist, and those that do typically require a large finite field. To address this, I will propose a new method by taking an encoder from an MDP convolutional code and repeating one of its (matrix) coefficients. The resulting code, which I refer to as Pseudo-MDP, is defined over the same finite field as the original code. While it may not be an MDP code, it demonstrates improved erasure correction performance compared to the original.

Abstract: Principal component analysis aims to summarize the multivariate data structure and is known for its dimension reduction, which achieves maximum variability. Compositional data are constrained positive data that reflect an overall composition, such as histograms of categorical variables or percentages of parts within a whole. An observation defined by a composition of p D-compositional variables (i.e. p variables each with D-part compositional components) is a compositional data. This type of multivariate observation is called a (p-dimensional) compositional data vector. Since each vector can be thought of as two-dimensional, a dataset formed by these compositional data vectors can be seen as a three-way array.

The methodology under development divides the data into two parts: the total data and the partial data, corresponding to R different geographical regions. Each region is a three-dimensional structure (n*p*D).  Since the total can be considered with the same structure and corresponds to the sum of the R regions, a dimensionality reduction is performed on the total, by PCA for compositional data vectors. The eigenvectors of the previous procedure are used in each region, corresponding to the partial. The scores of the main principal component are obtained. Finally, longitudinal plots are constructed using the clr coefficients and pivot coordinates, where each part of the composition is given by an age group, explaining how each group behaves over time. 

Abstract: A physical system may remain stable under certain perturbations or evolve in a new direction. In other words, the solutions to the equations describing the system may develop new branches under specific perturbations. Such bifurcation phenomena can occur in systems ranging from small droplets to compact astrophysical objects like black holes, on a cosmic scale. We find that exotic compact objects, composed of Einstein gravity minimally coupled to spin-0 and spin-1 massive bosonic fields, exhibit similar bifurcation phenomena. Under specific axisymmetric perturbations, spherically symmetric excited-state bosonic stars branch into two new solutions: chain-like and ring-like axisymmetric bosonic stars. Given that most studies of perturbations on exotic compact objects have only considered spherical perturbations, this work provides new insights into the behavior of such objects under axisymmetric perturbations.