-
Optimization on manifolds: A symplectic approach
Authors:
Guilherme França,
Alessandro Barp,
Mark Girolami,
Michael I. Jordan
Abstract:
Optimization tasks are crucial in statistical machine learning. Recently, there has been great interest in leveraging tools from dynamical systems to derive accelerated and robust optimization methods via suitable discretizations of continuous-time systems. However, these ideas have mostly been limited to Euclidean spaces and unconstrained settings, or to Riemannian gradient flows. In this work, w…
▽ More
Optimization tasks are crucial in statistical machine learning. Recently, there has been great interest in leveraging tools from dynamical systems to derive accelerated and robust optimization methods via suitable discretizations of continuous-time systems. However, these ideas have mostly been limited to Euclidean spaces and unconstrained settings, or to Riemannian gradient flows. In this work, we propose a dissipative extension of Dirac's theory of constrained Hamiltonian systems as a general framework for solving optimization problems over smooth manifolds, including problems with nonlinear constraints. We develop geometric/symplectic numerical integrators on manifolds that are "rate-matching," i.e., preserve the continuous-time rates of convergence. In particular, we introduce a dissipative RATTLE integrator able to achieve optimal convergence rate locally. Our class of (accelerated) algorithms are not only simple and efficient but also applicable to a broad range of contexts.
△ Less
Submitted 4 July, 2023; v1 submitted 23 July, 2021;
originally announced July 2021.
-
On dissipative symplectic integration with applications to gradient-based optimization
Authors:
Guilherme França,
Michael I. Jordan,
René Vidal
Abstract:
Recently, continuous-time dynamical systems have proved useful in providing conceptual and quantitative insights into gradient-based optimization, widely used in modern machine learning and statistics. An important question that arises in this line of work is how to discretize the system in such a way that its stability and rates of convergence are preserved. In this paper we propose a geometric f…
▽ More
Recently, continuous-time dynamical systems have proved useful in providing conceptual and quantitative insights into gradient-based optimization, widely used in modern machine learning and statistics. An important question that arises in this line of work is how to discretize the system in such a way that its stability and rates of convergence are preserved. In this paper we propose a geometric framework in which such discretizations can be realized systematically, enabling the derivation of "rate-matching" algorithms without the need for a discrete convergence analysis. More specifically, we show that a generalization of symplectic integrators to nonconservative and in particular dissipative Hamiltonian systems is able to preserve rates of convergence up to a controlled error. Moreover, such methods preserve a shadow Hamiltonian despite the absence of a conservation law, extending key results of symplectic integrators to nonconservative cases. Our arguments rely on a combination of backward error analysis with fundamental results from symplectic geometry. We stress that although the original motivation for this work was the application to optimization, where dissipative systems play a natural role, they are fully general and not only provide a differential geometric framework for dissipative Hamiltonian systems but also substantially extend the theory of structure-preserving integration.
△ Less
Submitted 28 April, 2021; v1 submitted 14 April, 2020;
originally announced April 2020.
-
Fractal and Multifractal Properties of Electrographic Recordings of Human Brain Activity: Toward Its Use as a Signal Feature for Machine Learning in Clinical Applications
Authors:
Lucas G. S. França,
José G. V. Miranda,
Marco Leite,
Niraj K. Sharma,
Matthew C. Walker,
Louis Lemieux,
Yujiang Wang
Abstract:
The brain is a system operating on multiple time scales, and characterisation of dynamics across time scales remains a challenge. One framework to study such dynamics is that of fractal geometry. However, currently there exists no established method for the study of brain dynamics using fractal geometry, due to the many challenges in the conceptual and technical understanding of the methods. We ai…
▽ More
The brain is a system operating on multiple time scales, and characterisation of dynamics across time scales remains a challenge. One framework to study such dynamics is that of fractal geometry. However, currently there exists no established method for the study of brain dynamics using fractal geometry, due to the many challenges in the conceptual and technical understanding of the methods. We aim to highlight some of the practical challenges of applying fractal geometry to brain dynamics and propose solutions to enable its wider use in neuroscience. Using intracranially recorded EEG and simulated data, we compared monofractal and multifractal methods with regards to their sensitivity to signal variance. We found that both correlate closely with signal variance, thus not offering new information about the signal. However, after applying an epoch-wise standardisation procedure to the signal, we found that multifractal measures could offer non-redundant information compared to signal variance, power and other established EEG signal measures. We also compared different multifractal estimation methods and found that the Chhabra-Jensen algorithm performed best. Finally, we investigated the impact of sampling frequency and epoch length on multifractal properties. Using epileptic seizures as an example event in the EEG, we show that there may be an optimal time scale for detecting temporal changes in multifractal properties around seizures. The practical issues we highlighted and our suggested solutions should help in developing a robust method for the application of fractal geometry in EEG signals. Our analyses and observations also aid the theoretical understanding of the multifractal properties of the brain and might provide grounds for new discoveries in the study of brain signals. These could be crucial for understanding of neurological function and for the developments of new treatments.
△ Less
Submitted 11 December, 2018; v1 submitted 11 June, 2018;
originally announced June 2018.
-
Epidemic SIR model on a face-to-face interaction network: new mobility induced phase transitions
Authors:
Paulo Freitas Gomes,
Andrey Gonçalves França,
Fábio Luiz Paranhos Costa,
Henrique Almeida Fernandes
Abstract:
In this work, we study the epidemic SIR model on a system which takes into consideration face-to-face interaction networks. This approach has been used as prototype to describe people interactions in different kinds of social organizations and, here, it is considered by means of three features of human interactions: the mobility, the duration of the interaction among people, and the dependence of…
▽ More
In this work, we study the epidemic SIR model on a system which takes into consideration face-to-face interaction networks. This approach has been used as prototype to describe people interactions in different kinds of social organizations and, here, it is considered by means of three features of human interactions: the mobility, the duration of the interaction among people, and the dependence of the number of interactions of each person on the time evolution of the system. For this purpose, the initial configuration of the system is set as a regular square lattice where the nodes are the individuals which, in turn, are able to move in a random walk along the network. So, the connectivity among the individuals evolve with time and is defined by the positions of the individuals at each iteration. In a time unit, each individual is able move up to a distance $v$ creating different networks along the time evolution of the system. In addition, the individuals are interacting with each other only if they are within the interaction distance $δ$ and, in this case, they are considered as neighbors. If a given individual is interacting with other ones, he performs the random walk with a diffusion probability $ω$. Otherwise, the diffusion occurs with probability 1. The study was carried out through non-equilibrium Monte Carlo Simulations and we take into account the asynchronous updating scheme. The results show that, for a given $v>0$, there exist a critical line in the $(c, δ)$ space, where $c$ is the immunization rate. We also obtain the dynamic critical exponent $θ$ for some points belonging to this line and show that this model does not belong to the directed percolation universality class.
△ Less
Submitted 24 May, 2018; v1 submitted 21 March, 2018;
originally announced March 2018.
-
Two-dimensional Bose and Fermi gases beyond weak coupling
Authors:
Guilherme Franca,
Andre LeClair,
Joshua Squires
Abstract:
Using a formalism based on the two-body S-matrix we study two-dimensional Bose and Fermi gases with both attractive and repulsive interactions. Approximate analytic expressions, valid at weak coupling and beyond, are developed and applied to the Berezinskii-Kosterlitz-Thouless (BKT) transition. We successfully recover the correct logarithmic functional form of the critical chemical potential and d…
▽ More
Using a formalism based on the two-body S-matrix we study two-dimensional Bose and Fermi gases with both attractive and repulsive interactions. Approximate analytic expressions, valid at weak coupling and beyond, are developed and applied to the Berezinskii-Kosterlitz-Thouless (BKT) transition. We successfully recover the correct logarithmic functional form of the critical chemical potential and density for the Bose gas. For fermions, the BKT critical temperature is calculated in BCS and BEC regimes through consideration of Tan's contact.
△ Less
Submitted 17 July, 2017; v1 submitted 29 March, 2017;
originally announced March 2017.
-
Nonextensivity in Geological Faults?
Authors:
G. S. França,
C. S. Vilar,
R. Silva,
J. S. Alcaniz
Abstract:
Geological fault systems, as the San Andreas fault (SAF) in USA, constitute typical examples of self-organizing systems in nature. In this paper, we have considered some geophysical properties of the SAF system to test the viability of the nonextensive models for earthquakes developed in [Phys. Rev. E {\bf 73}, 026102, 2006]. To this end, we have used 6188 earthquakes events ranging in the magni…
▽ More
Geological fault systems, as the San Andreas fault (SAF) in USA, constitute typical examples of self-organizing systems in nature. In this paper, we have considered some geophysical properties of the SAF system to test the viability of the nonextensive models for earthquakes developed in [Phys. Rev. E {\bf 73}, 026102, 2006]. To this end, we have used 6188 earthquakes events ranging in the magnitude interval $2 < m < 8$ that were taken from the Network Earthquake International Center catalogs (NEIC, 2004-2006) and the Bulletin of the International Seismological Centre (ISC, 1964-2003). For values of the Tsallis nonextensive parameter $q \simeq 1.68$, it is shown that the energy distribution function deduced in above reference provides an excellent fit to the NEIC and ISC SAF data.
△ Less
Submitted 10 April, 2006;
originally announced April 2006.
-
Nonextensive models for earthquakes
Authors:
R. Silva,
G. S. Franca,
C. S. Vilar,
J. S. Alcaniz
Abstract:
We have revisited the fragment-asperity interaction model recently introduced by Sotolongo-Costa and Posadas (Physical Review Letters 92, 048501, 2004) by considering a different definition for mean values in the context of Tsallis nonextensive statistics and introducing a new scale between the earthquake energy and the size of fragment $ε\propto r^3$. The energy distribution function (EDF) dedu…
▽ More
We have revisited the fragment-asperity interaction model recently introduced by Sotolongo-Costa and Posadas (Physical Review Letters 92, 048501, 2004) by considering a different definition for mean values in the context of Tsallis nonextensive statistics and introducing a new scale between the earthquake energy and the size of fragment $ε\propto r^3$. The energy distribution function (EDF) deduced in our approach is considerably different from the one obtained in the above reference. We have also tested the viability of this new EDF with data from two different catalogs (in three different areas), namely, NEIC and Bulletin Seismic of the Revista Brasileira de Geofísica. Although both approaches provide very similar values for the nonextensive parameter $q$, other physical quantities, e.g., the energy density differs considerably, by several orders of magnitude.
△ Less
Submitted 9 January, 2006; v1 submitted 14 November, 2005;
originally announced November 2005.
