Papers by Maria Mitradjieva
A Conjugate Direction Frank-Wolfe Method with Applications to the Traffic Assignment Problem
Operations Research Proceedings 2002, 2003
ABSTRACT We present a version of the Frank-Wolfe method for linearly constrained convex programs,... more ABSTRACT We present a version of the Frank-Wolfe method for linearly constrained convex programs, in which consecutive search direction are made conjugate to each other. We also present preliminary computational studies in a MATLAB environment. In these we apply the pure Frank-Wolfe, the Conjugate Direction Frank-Wolfe (CDFW) and the “partanized” Frank-Wolfe to some classical traffic assignment problems. CDFW compares favorably to the other methods in this study.
Operations Research Proceedings 2002, 2003
Professor da ULBRA/RS PALAVRAS-CHAVE -Comunicação (Comunication) -Política Partidária (Party poli... more Professor da ULBRA/RS PALAVRAS-CHAVE -Comunicação (Comunication) -Política Partidária (Party politics) -Filosofia Polítca (Pollitical philosophy) 1 Introdução O Brasil é um país que apresenta um contraste bastante aguçado e original, talvez único no mundo, entre o altíssimo padrão de qualidade de sua televisão -tecnicamente uma das melhores que se produz no mundo (sem entrar no mérito da qualidade dos conteúdos) -e a baixíssima qualidade de sua cidadania, resultante de um longo processo histórico marcado pela desigualdade social, a falta de acesso da população de baixa renda ao conhecimento proporcionado pela educação e o autoritarismo das elites nacionais em relação à sociedade. O resultado dessa perversa combinação de fatores elevou a televisão brasileira à condição de um vigoroso e influente meio de comunicação social e, ao mesmo tempo, de fabricação e reprodução do poder, historicamente já bastante concentrado em poucas mãos. A relação entre os meios de comunicação de massas e o poder é objeto de estudo tanto da Ciência Política como da Comunicação Social contemporâneas, mas no Brasil a relação entre mídia e poder parece ter alcançado contornos ainda mais evidentes e exacerbados do que em outros países. Nesse contexto, a campanha eleitoral de Fernando Collor de Mello à presidência da República, em 1989, parece ter, por um lado, contribuído para evidenciar ainda mais essa constatação, e por outro lado, parece também ter inaugurado uma "nova era" no marketing político tal como até então era praticado no país. O efeito combinado desses dois fatores, isto é, do contraste entre a alta qualidade da nossa televisão e a baixa qualidade da nossa cidadania, por um lado, e por outro, do sucesso da candidatura de Collor de Mello na eleição de 89 -um candidato até então desconhecido da maioria da nação e que ascendeu meteoricamente à presidência da República através de uma bem concebida estratégia de comunicação e marketing eleitoral -, parece ter criado no mercado político brasileiro a falsa ilusão
Transportation Science, 2013
W e present versions of the Frank-Wolfe method for linearly constrained convex programs, in which... more W e present versions of the Frank-Wolfe method for linearly constrained convex programs, in which consec- utive search directions are made conjugate. Preliminary computational studies in a MATLAB environment applying pure Frank-Wolfe, conjugate direction Frank-Wolfe (CFW), bi-conjugate Frank-Wolfe (BFW), and "partanized" Frank-Wolfe methods to some classical Traffic Assignment Problems show that CFW and BFW compare favorably to the other methods. This spurred a more detailed study, comparing our methods to A3 an origin-based algorithm. This study indicates that our methods are competitive for accuracy requirements. We also show that CFW is globally convergent. We further point at independent studies by other researchers that show that our methods compare favorably with recent bush-based and gradient projection algorithms on computers with several cores.
Computational Optimization and Applications, 2008
The feasible direction method of Frank and Wolfe has been claimed to be efficient for solving the... more The feasible direction method of Frank and Wolfe has been claimed to be efficient for solving the stochastic transportation problem. While this is true for very moderate accuracy requirements, substantially more efficient algorithms are otherwise diagonalized Newton and conjugate Frank-Wolfe algorithms, which we describe and evaluate. Like the Frank-Wolfe algorithm, these two algorithms take advantage of the structure of the stochastic transportation problem. We also introduce a Frank-Wolfe type algorithm with multi-dimensional search; this search procedure exploits the Cartesian product structure of the problem. Numerical results for two classic test problem sets are given. The three new methods that are considered are shown to be superior to the Frank-Wolfe method, and also to an earlier suggested heuristic acceleration of the Frank-Wolfe method.
A sequential linear programming algorithm with multi-dimensional search: derivation and convergence
We present a sequential linear programming, SLP, algorithm in which the traditional line-search s... more We present a sequential linear programming, SLP, algorithm in which the traditional line-search step is replaced by a multi-dimensional search. The algorithm is based on inner approximations of bot ...
Journal of Geometry, 1999
We develop a technique for improving the universal linear programming bounds on the cardinality a... more We develop a technique for improving the universal linear programming bounds on the cardinality and the minimum distance of codes in proiective spaces I~P ~-1. We firstly investigate test functions Pj(m, n, s) having the property that Pj(m, n, s) < 0 for some j if and only if the corresponding universal linear programming bound can be further improved by linear programming. Then we describe a method for improving the universal bounds. We also investigate the possibilities for attaining the first universal bounds.
A Conjugate Direction Frank-Wolfe Method with Applications to the Traffic Assignment Problem
Operations Research Proceedings 2002, 2003
We present a version of the Frank-Wolfe method for linearly constrained convex programs, in which... more We present a version of the Frank-Wolfe method for linearly constrained convex programs, in which consecutive search direction are made conjugate to each other. We also present preliminary computational studies in a MATLAB environment. In these we apply the pure Frank-Wolfe, the Conjugate Direction Frank-Wolfe (CDFW) and the “partanized” Frank-Wolfe to some classical traffic assignment problems. CDFW compares favorably to the other methods in this study.
A Conjugate Direction Frank-Wolfe Method for Nonconvex Problems
Department of Mathematics Technical Report, 2003
In this paper we propose an algorithm for solving problems with nonconvex objective function and ... more In this paper we propose an algorithm for solving problems with nonconvex objective function and linear constraints. We extend the previously suggested Conjugate direction Frank–Wolfe algorithm to ...
Improved Frank-Wolfe Directions with Applications to Traffic Problems
The main contribution of this thesis is the development of some new efficient algorithms for solv... more The main contribution of this thesis is the development of some new efficient algorithms for solving structured linearly constrained optimization problems. The conventional Frank-Wolfe method is on ...
A sequential linear programming algorithm with multi-dimensional search: derivation and convergence
We present a sequential linear programming, SLP, algorithm in which the traditional line-search s... more We present a sequential linear programming, SLP, algorithm in which the traditional line-search step is replaced by a multi-dimensional search. The algorithm is based on inner approximations of bot ...
Feasible Direction Methods for Constrained Nonlinear Optimization: Suggestions for Improvements
Multi-Class User Equilibria under Social Marginal Cost Pricing
Operations Research 2002, p.174-179, 2003
In the congested cities of today, congestion pricing is a tempting alternative. With a single use... more In the congested cities of today, congestion pricing is a tempting alternative. With a single user class, already Beckmann et al. showed that ``system optimal'' traffic flows can be achieved by social marginal cost (SMC) pricing where users have to pay for the delays the incur on others. However different user classes can have widly differing time values. Hence, when introducing tolls, one should consider multi-class user equilibria, where the classes have different time values. In the single class case, the equilibrium conditions can be viewn as optimality conditions of an equivalent optimization problem. In the multi-class case, however, netter claims that this is not possible. We show that, depending on the formulation, the multi-class SMC-pricing equilibrium problem (with different time values) can be stated either as an asymmetric or as a symmetric equilibrium problem. In the latter case, the corresponding optimization problems is in general non-convex. For this non-convex problem, we devise descent methods of Frank-Wolfe type. We apply the methods and study a synthetic case based on Sioux Falls.
A Comparison of Feasible Direction Methods for the Stochastic Transportation Problem
Computational optimization and applications, Volume 46, Issue 3, p. 451-466, Jul 2010
The feasible direction method of Frank and Wolfe has been claimed to be efficient for solving the... more The feasible direction method of Frank and Wolfe has been claimed to be efficient for solving the stochastic transportation problem. While this is true for very moderate accuracy requirements, substantially more efficient algorithms are otherwise diagonalized Newton and conjugate Frank–Wolfe algorithms, which we describe and evaluate. Like the Frank–Wolfe algorithm, these two algorithms take advantage of the structure of the stochastic transportation problem. We also introduce a Frank–Wolfe type algorithm with multi-dimensional search; this search procedure exploits the Cartesian product structure of the problem. Numerical results for two classic test problem sets are given. The three new methods that are considered are shown to be superior to the Frank–Wolfe method, and also to an earlier suggested heuristic acceleration of the Frank–Wolfe method.
The Stiff Is Moving--Conjugate Direction Frank-Wolfe Methods with Applications to Traffic Assignment.
Transportation Science. May2013, Vol. 47 Issue 2, p280-293. 14p. , May 2013
We present versions of the Frank-Wolfe method for linearly constrained convex programs, in which ... more We present versions of the Frank-Wolfe method for linearly constrained convex programs, in which consecutive search directions are made conjugate. Preliminary computational studies in a MATLAB environment applying pure Frank-Wolfe, conjugate direction Frank-Wolfe (CFW), bi-conjugate Frank-Wolfe (BFW), and "partanized" Frank-Wolfe methods to some classical Traffic Assignment Problems show that CFW and BFW compare favorably to the other methods. This spurred a more detailed study, comparing our methods to an origin-based algorithm. This study indicates that our methods are competitive for accuracy requirements ensuring link flow stability. We also show that CFW is globally convergent. We further point at independent studies by other researchers that show that our methods compare favorably with recent bush-based and gradient projection algorithms on computers with several cores.
Journal of Geometry, 1997
We develop a technique for improving the universal linear programming bounds on the cardinality a... more We develop a technique for improving the universal linear programming bounds on the cardinality and the minimum distance of codes in projective spaces I FP n−1 . We firstly investigate test functions P j (m, n, s) having the property that P j (m, n, s) < 0 for some j if and only if the corresponding universal linear programming bound can be further improved by linear programming. Then we describe a method for improving the universal bounds. We also investigate the possibilities for attaining the first universal bounds.
Uploads
Papers by Maria Mitradjieva