-
Linear stability of lattice Boltzmann models with non-ideal equation of state
Authors:
S. A. Hosseini,
I. V. Karlin
Abstract:
Detailed study of spectral properties and of linear stability is presented for a class of lattice Boltzmann models with a non-ideal equation of state. Examples include the van der Waals and the shallow water models. Both analytical and numerical approaches demonstrate that linear stability requires boundedness of propagation speeds of normal eigen-modes. The study provides a basis for the construc…
▽ More
Detailed study of spectral properties and of linear stability is presented for a class of lattice Boltzmann models with a non-ideal equation of state. Examples include the van der Waals and the shallow water models. Both analytical and numerical approaches demonstrate that linear stability requires boundedness of propagation speeds of normal eigen-modes. The study provides a basis for the construction of unconditionally stable lattice Boltzmann models.
△ Less
Submitted 6 March, 2025;
originally announced March 2025.
-
AdaFisher: Adaptive Second Order Optimization via Fisher Information
Authors:
Damien Martins Gomes,
Yanlei Zhang,
Eugene Belilovsky,
Guy Wolf,
Mahdi S. Hosseini
Abstract:
First-order optimization methods are currently the mainstream in training deep neural networks (DNNs). Optimizers like Adam incorporate limited curvature information by employing the diagonal matrix preconditioning of the stochastic gradient during the training. Despite their widespread, second-order optimization algorithms exhibit superior convergence properties compared to their first-order coun…
▽ More
First-order optimization methods are currently the mainstream in training deep neural networks (DNNs). Optimizers like Adam incorporate limited curvature information by employing the diagonal matrix preconditioning of the stochastic gradient during the training. Despite their widespread, second-order optimization algorithms exhibit superior convergence properties compared to their first-order counterparts e.g. Adam and SGD. However, their practicality in training DNNs is still limited due to increased per-iteration computations compared to the first-order methods. We present \emph{AdaFisher}--an adaptive second-order optimizer that leverages a \emph{diagonal block-Kronecker} approximation of the Fisher information matrix for adaptive gradient preconditioning. AdaFisher aims to bridge the gap between enhanced \emph{convergence/generalization} capabilities and computational efficiency in second-order optimization framework for training DNNs. Despite the slow pace of second-order optimizers, we showcase that AdaFisher can be reliably adopted for image classification, language modeling and stands out for its stability and robustness in hyper-parameter tuning. We demonstrate that AdaFisher \textbf{outperforms the SOTA optimizers} in terms of both accuracy and convergence speed. Code is available from https://github.com/AtlasAnalyticsLab/AdaFisher.
△ Less
Submitted 10 March, 2025; v1 submitted 25 May, 2024;
originally announced May 2024.
-
Etale spaces of residuated lattices
Authors:
Saeed Rasouli,
Seyed Naser Hosseini,
Amin Dehghani
Abstract:
This paper explores the concept of étalé spaces associated with residuated lattices. Notions of bundles and étalés of residuated lattices over a given topological space are introduced and investigated. For a topological space $\mathscr{B}$, we establish that the category of étalés of residuated lattices over $\mathscr{B}$ with morphisms of étalés of residuated lattices is coreflective in the categ…
▽ More
This paper explores the concept of étalé spaces associated with residuated lattices. Notions of bundles and étalés of residuated lattices over a given topological space are introduced and investigated. For a topological space $\mathscr{B}$, we establish that the category of étalés of residuated lattices over $\mathscr{B}$ with morphisms of étalés of residuated lattices is coreflective in the category of bundles of residuated lattices over $\mathscr{B}$ along with morphisms of bundles of residuated lattices. We provide a method for transferring an étalé of residuated lattices over a topological space to another, utilizing a continuous map. Finally, we define a contravariant functor, called the section functor, from the category of étalés of residuated lattices with inverse morphisms to the category of residuated lattices.
△ Less
Submitted 14 February, 2024;
originally announced February 2024.
-
Quotients of span categories that are allegories and the representation of regular categories
Authors:
S. Naser Hosseini,
Amir R. Shir Ali Nasab,
Walter Tholen,
Leila Yeganeh
Abstract:
We consider the ordinary category Span(C) of (isomorphism classes of) spans of morphisms in a category C with finite limits as needed, composed horizontally via pullback, and give a general criterion for a quotient of Span(C) to be an allegory. In particular, when C carries a pullback-stable, but not necessarily proper, (E, M)-factorization system, we establish a quotient category Span_E(C) that i…
▽ More
We consider the ordinary category Span(C) of (isomorphism classes of) spans of morphisms in a category C with finite limits as needed, composed horizontally via pullback, and give a general criterion for a quotient of Span(C) to be an allegory. In particular, when C carries a pullback-stable, but not necessarily proper, (E, M)-factorization system, we establish a quotient category Span_E(C) that is isomorphic to the category Rel_M(C) of M-relations in C, and show that it is a (unitary and tabular) allegory precisely when M is a class of monomorphisms in C. Without this restriction, one can still find a least pullback-stable and composition-closed class E. containing E such that Span_E.(C) is a unitary and tabular allegory. In this way one obtains a left adjoint to the 2-functor that assigns to every unitary and tabular allegory the regular category of its Lawverian maps. With the Freyd-Scedrov Representation Theorem for regular categories, we conclude that every finitely complete category with a stable factorization system has a reflection into the huge 2-category of all regular categories.
△ Less
Submitted 8 December, 2021;
originally announced December 2021.
-
The exterior product and homology of Hom-Lie algebras
Authors:
Negur Shahni Karamzadeh,
Seyedeh Narges Hosseini,
Ali Reza Salemkar
Abstract:
In this article, we use the theory of (non-abelian) exterior product of Hom-Lie algebras to prove the Hopf formula for these algebras. As an application, we construct an eight-term sequence in the homology of Hom-Lie algebras. We also investigate the capability property of Hom-Lie algebras via the exterior product.
In this article, we use the theory of (non-abelian) exterior product of Hom-Lie algebras to prove the Hopf formula for these algebras. As an application, we construct an eight-term sequence in the homology of Hom-Lie algebras. We also investigate the capability property of Hom-Lie algebras via the exterior product.
△ Less
Submitted 24 April, 2021;
originally announced April 2021.
-
Some results on the second relative homology of Leibniz algebras
Authors:
Seyedeh Narges Hosseini,
Behrouz Edalatzadeh,
Ali Reza Salemkar
Abstract:
In this paper, the structure of the second relative homology and the relative stem cover of the direct sum of two pairs of Leibniz algebras are determined by means of the non-abelian tensor product of Leibniz algebras. We also characterize all pairs of finite dimensional nilpotent Leibniz algebras such that...
In this paper, the structure of the second relative homology and the relative stem cover of the direct sum of two pairs of Leibniz algebras are determined by means of the non-abelian tensor product of Leibniz algebras. We also characterize all pairs of finite dimensional nilpotent Leibniz algebras such that...
△ Less
Submitted 24 April, 2021;
originally announced April 2021.
-
On Reversing Operator Choi-Davis-Jensen inequality
Authors:
Seyyed Saeid Hashemi Karouei,
Mohammad Sadegh Asgari,
Mohsen Shah Hosseini
Abstract:
In this paper, we first provide a better estimate of the second inequality in Hermite-Hadamard inequality. Next, we study the reverse of the celebrated Davis-Choi-Jensen's inequality. Our results are employed to establish a new bound for the operator Kantorovich inequality.
In this paper, we first provide a better estimate of the second inequality in Hermite-Hadamard inequality. Next, we study the reverse of the celebrated Davis-Choi-Jensen's inequality. Our results are employed to establish a new bound for the operator Kantorovich inequality.
△ Less
Submitted 6 April, 2021;
originally announced April 2021.
-
A Proposed Lean Distribution System for Solar Power Plants Using Mathematical Modeling and Simulation Technique
Authors:
Mohsen Momenitabar,
Zhila Dehdari Ebrahimi,
Seyed Hassan Hosseini,
Mohammad Arani
Abstract:
Today, power waste is one of the most crucial problems which power stations across the world are facing. One of the recent trends of the energy system is the lean management technique. When lean management is indicated by the system, customer value is increased and the waste process in industry or in a power station is reduced. In this paper, first of all, we propose mathematical modeling to reduc…
▽ More
Today, power waste is one of the most crucial problems which power stations across the world are facing. One of the recent trends of the energy system is the lean management technique. When lean management is indicated by the system, customer value is increased and the waste process in industry or in a power station is reduced. In this paper, first of all, we propose mathematical modeling to reduce the cost of production, and then the simulation technique applies to electricity transmission distribution systems. Furthermore, we consider two criteria for comparison including the different costs of the system and the rate of energy waste during the transmission. The primary approach is to use both of the models in order to draw a comparison between simulation results and mathematical models. Finally, the analysis of the test results done by the CPLEX toolbox of MATLAB Software 2019 leads to a remarkable decrease in the costs of energy demand in the electricity transmitting network distribution system.
△ Less
Submitted 15 January, 2021;
originally announced January 2021.
-
Minimum Weighted Szeged Index Trees
Authors:
Pavol Hell,
César Hernández-Cruz,
Seyyed Aliasghar Hosseini
Abstract:
Weighted Szeged index is a recently introduced extension of the well-known Szeged index. In this paper, we present a new tool to analyze and characterize minimum weighted Szeged index trees. We exhibit the best trees with up to 81 vertices and use this information, together with our results, to propose various conjectures on the structure of minimum weighted Szeged index trees.
Weighted Szeged index is a recently introduced extension of the well-known Szeged index. In this paper, we present a new tool to analyze and characterize minimum weighted Szeged index trees. We exhibit the best trees with up to 81 vertices and use this information, together with our results, to propose various conjectures on the structure of minimum weighted Szeged index trees.
△ Less
Submitted 20 July, 2020;
originally announced July 2020.
-
AdaS: Adaptive Scheduling of Stochastic Gradients
Authors:
Mahdi S. Hosseini,
Konstantinos N. Plataniotis
Abstract:
The choice of step-size used in Stochastic Gradient Descent (SGD) optimization is empirically selected in most training procedures. Moreover, the use of scheduled learning techniques such as Step-Decaying, Cyclical-Learning, and Warmup to tune the step-size requires extensive practical experience--offering limited insight into how the parameters update--and is not consistent across applications. T…
▽ More
The choice of step-size used in Stochastic Gradient Descent (SGD) optimization is empirically selected in most training procedures. Moreover, the use of scheduled learning techniques such as Step-Decaying, Cyclical-Learning, and Warmup to tune the step-size requires extensive practical experience--offering limited insight into how the parameters update--and is not consistent across applications. This work attempts to answer a question of interest to both researchers and practitioners, namely \textit{"how much knowledge is gained in iterative training of deep neural networks?"} Answering this question introduces two useful metrics derived from the singular values of the low-rank factorization of convolution layers in deep neural networks. We introduce the notions of \textit{"knowledge gain"} and \textit{"mapping condition"} and propose a new algorithm called Adaptive Scheduling (AdaS) that utilizes these derived metrics to adapt the SGD learning rate proportionally to the rate of change in knowledge gain over successive iterations. Experimentation reveals that, using the derived metrics, AdaS exhibits: (a) faster convergence and superior generalization over existing adaptive learning methods; and (b) lack of dependence on a validation set to determine when to stop training. Code is available at \url{https://github.com/mahdihosseini/AdaS}.
△ Less
Submitted 11 June, 2020;
originally announced June 2020.
-
On the cop number of graphs of high girth
Authors:
Peter Bradshaw,
Seyyed Aliasghar Hosseini,
Bojan Mohar,
Ladislav Stacho
Abstract:
We establish a lower bound for the cop number of graphs of high girth in terms of the minimum degree, and more generally, in terms of a certain growth condition. We show, in particular, that the cop number of any graph with girth $g$ and minimum degree $δ$ is at least $\tfrac{1}{g}(δ- 1)^{\lfloor \frac{g-1}{4}\rfloor}$. We establish similar results for directed graphs. While exposing several reaso…
▽ More
We establish a lower bound for the cop number of graphs of high girth in terms of the minimum degree, and more generally, in terms of a certain growth condition. We show, in particular, that the cop number of any graph with girth $g$ and minimum degree $δ$ is at least $\tfrac{1}{g}(δ- 1)^{\lfloor \frac{g-1}{4}\rfloor}$. We establish similar results for directed graphs. While exposing several reasons for conjecturing that the exponent $\tfrac{1}{4}g$ in this lower bound cannot be improved to $(\tfrac{1}{4}+\varepsilon)g$, we are also able to prove that it cannot be increased beyond $\frac{3}{8}g$. This is established by considering a certain family of Ramanujan graphs. In our proof of this bound, we also show that the "weak" Meyniel's conjecture holds for expander graph families of bounded degree.
△ Less
Submitted 21 May, 2020;
originally announced May 2020.
-
Cops and Robbers on graphs of bounded diameter
Authors:
Seyyed Aliasghar Hosseini,
Fiachra Knox,
Bojan Mohar
Abstract:
The game of Cops and Robbers is a well known game played on graphs. In this paper we consider the class of graphs of bounded diameter. We improve the strategy of cops and previously used probabilistic method which results in an improved upper bound for the cop number of graphs of bounded diameter. In particular, for graphs of diameter four, we improve the upper bound from $n^{\frac{2}{3}+o(1)}$ to…
▽ More
The game of Cops and Robbers is a well known game played on graphs. In this paper we consider the class of graphs of bounded diameter. We improve the strategy of cops and previously used probabilistic method which results in an improved upper bound for the cop number of graphs of bounded diameter. In particular, for graphs of diameter four, we improve the upper bound from $n^{\frac{2}{3}+o(1)}$ to $n^{\frac{3}{5}+o(1)}$ and for diameter three from $n^{\frac{2}{3}+o(1)}$ to $n^{\frac{4}{7}+o(1)}$.
△ Less
Submitted 16 December, 2019;
originally announced December 2019.
-
Meyniel's conjecture on graphs of bounded degree
Authors:
Seyyed Aliasghar Hosseini,
Bojan Mohar,
Sebastian Gonzalez Hermosillo de la Maza
Abstract:
The game of Cops and Robbers is a well known pursuit-evasion game played on graphs. It has been proved \cite{bounded_degree} that cubic graphs can have arbitrarily large cop number $c(G)$, but the known constructions show only that the set $\{c(G) \mid G \text{ cubic}\}$ is unbounded. In this paper we prove that there are arbitrarily large subcubic graphs $G$ whose cop number is at least…
▽ More
The game of Cops and Robbers is a well known pursuit-evasion game played on graphs. It has been proved \cite{bounded_degree} that cubic graphs can have arbitrarily large cop number $c(G)$, but the known constructions show only that the set $\{c(G) \mid G \text{ cubic}\}$ is unbounded. In this paper we prove that there are arbitrarily large subcubic graphs $G$ whose cop number is at least $n^{1/2-o(1)}$ where $n=|V(G)|$. We also show that proving Meyniel's conjecture for graphs of bounded degree implies a weak Meyniel's conjecture for all graphs.
△ Less
Submitted 20 November, 2020; v1 submitted 14 December, 2019;
originally announced December 2019.
-
Lomax distribution and asymptotical ML estimations based on record values for probability density function and cumulative distribution function
Authors:
Saman Hosseini,
Dler Hussein Kadir,
Kostas Triantafyllopoulos
Abstract:
Here in this paper, it is tried to obtain and compare the ML estimations based on upper record values and a random sample. In continue, some theorems have been proven about the behavior of these estimations asymptotically.
Here in this paper, it is tried to obtain and compare the ML estimations based on upper record values and a random sample. In continue, some theorems have been proven about the behavior of these estimations asymptotically.
△ Less
Submitted 17 May, 2021; v1 submitted 16 October, 2019;
originally announced October 2019.
-
Surrounding cops and robbers on graphs of bounded genus
Authors:
Peter Bradshaw,
Seyyed Aliasghar Hosseini
Abstract:
We consider a surrounding variant of cops and robbers on graphs of bounded genus. We obtain bounds on the number of cops required to surround a robber on planar graphs, toroidal graphs, and outerplanar graphs. We also obtain improved bounds for bipartite planar and toroidal graphs. We briefly consider general graphs of bounded genus and graphs with a forbidden minor.
We consider a surrounding variant of cops and robbers on graphs of bounded genus. We obtain bounds on the number of cops required to surround a robber on planar graphs, toroidal graphs, and outerplanar graphs. We also obtain improved bounds for bipartite planar and toroidal graphs. We briefly consider general graphs of bounded genus and graphs with a forbidden minor.
△ Less
Submitted 3 November, 2019; v1 submitted 21 September, 2019;
originally announced September 2019.
-
On the Closed Form Expression of Elementary Symmetric Polynomials and the Inverse of Vandermonde Matrix
Authors:
Mahdi S. Hosseini,
Alfred Chen,
Konstantinos N. Plataniotis
Abstract:
Inverse Vandermonde matrix calculation is a long-standing problem to solve nonsingular linear system $Vc=b$ where the rows of a square matrix $V$ are constructed by progression of the power polynomials. It has many applications in scientific computing including interpolation, super-resolution, and construction of special matrices applied in cryptography. Despite its numerous applications, the matr…
▽ More
Inverse Vandermonde matrix calculation is a long-standing problem to solve nonsingular linear system $Vc=b$ where the rows of a square matrix $V$ are constructed by progression of the power polynomials. It has many applications in scientific computing including interpolation, super-resolution, and construction of special matrices applied in cryptography. Despite its numerous applications, the matrix is highly ill-conditioned where specialized treatments are considered for approximation such as conversion to Cauchy matrix, spectral decomposition, and algorithmic tailoring of the numerical solutions. In this paper, we propose a generalized algorithm that takes arbitrary pairwise (non-repetitive) sample nodes for solving inverse Vandermonde matrix. This is done in two steps: first, a highly balanced recursive algorithm is introduced with $\mathcal{O}(N)$ complexity to solve the combinatorics summation of the elementary symmetric polynomials; and second, a closed-form solution is tailored for inverse Vandermonde where the matrix' elements utilize this recursive summation for the inverse calculations. The numerical stability and accuracy of the proposed inverse method is analyzed through the spectral decomposition of the Frobenius companion matrix that associates with the corresponding Vandermonde matrix. The results show significant improvement over the state-of-the-art solutions using specific nodes such as $N$th roots of unity defined on the complex plane. A basic application in one dimensional interpolation problem is considered to demonstrate the utility of the proposed method for super-resolved signals.
△ Less
Submitted 17 September, 2019;
originally announced September 2019.
-
Cops and robbers on directed and undirected abelian Cayley graphs
Authors:
Peter Bradshaw,
Seyyed Aliasghar Hosseini,
Jérémie Turcotte
Abstract:
We show that the cop number of directed and undirected Cayley graphs on abelian groups has an upper bound of the form of $O(\sqrt{n})$, where $n$ is the number of vertices, by introducing a refined inductive method. With our method, we improve the previous upper bound on cop number for undirected Cayley graphs on abelian groups, and we establish an upper bound on the cop number of directed Cayley…
▽ More
We show that the cop number of directed and undirected Cayley graphs on abelian groups has an upper bound of the form of $O(\sqrt{n})$, where $n$ is the number of vertices, by introducing a refined inductive method. With our method, we improve the previous upper bound on cop number for undirected Cayley graphs on abelian groups, and we establish an upper bound on the cop number of directed Cayley graphs on abelian groups. We also use Cayley graphs on abelian groups to construct new \emph{Meyniel extremal families}, which contain graphs of every order $n$ with cop number $Θ(\sqrt{n})$.
△ Less
Submitted 16 September, 2020; v1 submitted 11 September, 2019;
originally announced September 2019.
-
Solution of minimum spanning forest problems with reliability constraints
Authors:
Ida Kalateh Ahani,
Majid Salari,
Seyed Mahmoud Hosseini,
Manuel Iori
Abstract:
We propose the reliability constrained k-rooted minimum spanning forest, a relevant optimization problem whose aim is to find a k-rooted minimum cost forest that connects given customers to a number of supply vertices, in such a way that a minimum required reliability on each path between a customer and a supply vertex is satisfied and the cost is a minimum. The reliability of an edge is the proba…
▽ More
We propose the reliability constrained k-rooted minimum spanning forest, a relevant optimization problem whose aim is to find a k-rooted minimum cost forest that connects given customers to a number of supply vertices, in such a way that a minimum required reliability on each path between a customer and a supply vertex is satisfied and the cost is a minimum. The reliability of an edge is the probability that no failure occurs on that edge, whereas the reliability of a path is the product of the reliabilities of the edges in such path. The problem has relevant applications in the design of networks, in fields such as telecommunications, electricity and transports. For its solution, we propose a mixed integer linear programming model, and an adaptive large neighborhood search metaheuristic which invokes several shaking and local search operators. Extensive computational tests prove that the metaheuristic can provide good quality solutions in very short computing times.
△ Less
Submitted 7 September, 2019;
originally announced September 2019.
-
Vertex arboricity of cographs
Authors:
Sebastián González Hermosillo de la Maza,
Pavol Hell,
César Hernández Cruz,
Seyyed Aliasghar Hosseini,
Payam Valadkhan
Abstract:
Arboricity is a graph parameter akin to chromatic number, in that it seeks to partition the vertices into the smallest number of sparse subgraphs. Where for the chromatic number we are partitioning the vertices into independent sets, for the arboricity we want to partition the vertices into cycle-free subsets (i.e., forests). Arboricity is NP-hard in general, and our focus is on the arboricity of…
▽ More
Arboricity is a graph parameter akin to chromatic number, in that it seeks to partition the vertices into the smallest number of sparse subgraphs. Where for the chromatic number we are partitioning the vertices into independent sets, for the arboricity we want to partition the vertices into cycle-free subsets (i.e., forests). Arboricity is NP-hard in general, and our focus is on the arboricity of cographs. For arboricity two, we obtain the complete list of minimal cograph obstructions. These minimal obstructions do generalize to higher arboricities; however, we no longer have a complete list, and in fact, the number of minimal cograph obstructions grows exponentially with arboricity. We obtain bounds on their size and the height of their cotrees.
More generally, we consider the following common generalization of colouring and partition into forests: given non-negative integers $p$ and $q$, we ask if a given cograph $G$ admits a vertex partition into $p$ forests and $q$ independent sets. We give a polynomial-time dynamic programming algorithm for this problem. In fact, the algorithm solves a more general problem which also includes several other problems such as finding a maximum $q$-colourable subgraph, maximum subgraph of arboricity-$p$, minimum vertex feedback set and minimum $q$ of a $q$-colourable vertex feedback set.
△ Less
Submitted 16 July, 2019;
originally announced July 2019.
-
On the Operator Jensen Inequality for Convex Functions
Authors:
M. Shah Hosseini,
H. R. Moradi,
B. Moosavi
Abstract:
This paper is mainly devoted to studying operator Jensen inequality. More precisely, a new generalization of Jensen inequality and its reverse version for convex (not necessary operator convex) functions have been proved. Several special cases are discussed as well.
This paper is mainly devoted to studying operator Jensen inequality. More precisely, a new generalization of Jensen inequality and its reverse version for convex (not necessary operator convex) functions have been proved. Several special cases are discussed as well.
△ Less
Submitted 7 June, 2019; v1 submitted 30 May, 2019;
originally announced May 2019.
-
Cops and robbers on oriented toroidal grids
Authors:
Sebastian Gonzalez Hermosillo de la Maza,
Seyyed Aliasghar Hosseini,
Fiachra Knox,
Bojan Mohar,
Bruce Reed
Abstract:
The game of cops and robbers is a well-known game played on graphs. In this paper we consider the straight-ahead orientations of 4-regular quadrangulations of the torus and the Klein bottle and we prove that their cop number is bounded by a constant. We also show that the cop number of every k-regularly oriented toroidal grid is at most 13.
The game of cops and robbers is a well-known game played on graphs. In this paper we consider the straight-ahead orientations of 4-regular quadrangulations of the torus and the Klein bottle and we prove that their cop number is bounded by a constant. We also show that the cop number of every k-regularly oriented toroidal grid is at most 13.
△ Less
Submitted 11 May, 2020; v1 submitted 22 April, 2019;
originally announced April 2019.
-
Fraction, Restriction and Range Categories from Non-Monic Classes of Morphisms
Authors:
S. N. Hosseini,
A. R. Shir Ali Nasab,
W. Tholen
Abstract:
For a composition-closed and pullback-stable class S of morphisms in a category C containing all isomorphisms, we form the category Span(C,S) of S-spans (s,f) in C with first "leg" s lying in S, and give an alternative construction of its quotient category C[S^{-1}] of S-fractions. Instead of trying to turn S-morphisms "directly" into isomorphisms, we turn them separately into retractions and into…
▽ More
For a composition-closed and pullback-stable class S of morphisms in a category C containing all isomorphisms, we form the category Span(C,S) of S-spans (s,f) in C with first "leg" s lying in S, and give an alternative construction of its quotient category C[S^{-1}] of S-fractions. Instead of trying to turn S-morphisms "directly" into isomorphisms, we turn them separately into retractions and into sections in a universal manner, thus obtaining the quotient categories Retr( C,S) and Sect(C,S). The fraction category C[S^{-1}] is their largest joint quotient category.
Without confining S to be a class of monomorphisms of C, we show that Sect(C,S) admits a quotient category, Par(C,S), whose name is justified by two facts. On one hand, for S a class of monomorphisms in C, it returns the category of S-spans in C, also called S-partial maps in this case; on the other hand, we prove that Par(C,S) is a split restriction category (in the sense of Cockett and Lack). A further quotient construction produces even a range category (in the sense of Cockett, Guo and Hofstra), RaPar(C,S), which is still large enough to admit C[S^{-1}] as its quotient.
Both, Par and RaPar, are the left adjoints of global 2-adjunctions. When restricting these to their "fixed objects", one obtains precisely the 2-equivalences by which their name givers characterized restriction and range categories. Hence, both Par(C,S)$ and RaPar(C,S may be naturally presented as Par(D,T)$ and RaPa(D,T), respectively, where now T is a class of monomorphisms in D. In summary, while there is no {\em a priori} need for the exclusive consideration of classes of monomorphisms, one may resort to them naturally
△ Less
Submitted 20 October, 2019; v1 submitted 28 February, 2019;
originally announced March 2019.
-
Constructing Geometric Graphs of Cop Number Three
Authors:
Seyyed Aliasghar Hosseini,
Masood Masjoody,
Ladislav Stacho
Abstract:
The game of cops and robbers is a pursuit game on graphs where a set of agents, called the cops try to get to the same position of another agent, called the robber. Cops and robbers has been studies on several classes of graphs including geometrically represented graphs. For example, it has been shown that string graphs, including geometric graphs, have cop number at most 15. On the other hand, li…
▽ More
The game of cops and robbers is a pursuit game on graphs where a set of agents, called the cops try to get to the same position of another agent, called the robber. Cops and robbers has been studies on several classes of graphs including geometrically represented graphs. For example, it has been shown that string graphs, including geometric graphs, have cop number at most 15. On the other hand, little is known about geometric graphs of any cop number less than 15 and there is only one example of a geometric graph of cop number three that has as many as 1440 vertices. In this paper we present a construction for subdividing planar graphs of maximum degree $\le 5$ into geometric planar graphs of at least the same cop number. Indeed, our construction shows that there are infinitely many planar geometric graphs of cop number three. We also present another construction that consists in clique substitutions alongside subdividing the edges in a planar graph of maximum degree $\le 9$, resulting in geometric, but not necessarily planar, graphs of at least the same cop number as the starting graphs.
△ Less
Submitted 17 March, 2019; v1 submitted 10 November, 2018;
originally announced November 2018.
-
Viewing *-multiplication operators between Orlics spaces
Authors:
Jahangir Cheshmavar,
Seyed Kamel Hosseini
Abstract:
In this paper, Lambert multipliers acting between Orlicz spaces are characterized based on conditional expectation operators. Also, we give necessary and sufficient conditions to *-multiplication operators to be closed range. Finally, the necessary condition under which a *-multiplication operator to be Fredholm operator is investigated.
In this paper, Lambert multipliers acting between Orlicz spaces are characterized based on conditional expectation operators. Also, we give necessary and sufficient conditions to *-multiplication operators to be closed range. Finally, the necessary condition under which a *-multiplication operator to be Fredholm operator is investigated.
△ Less
Submitted 4 January, 2019; v1 submitted 15 October, 2018;
originally announced October 2018.
-
An Alternative Estimate for the Numerical Radius of Hilbert Space Operators
Authors:
M. Shah Hosseini,
B. Moosavi,
H. R. Moradi
Abstract:
We give an alternative lower bound for the numerical radii of Hilbert space operators. As a by-product, we find conditions such that
\begin{equation*} ω\left(\left[\begin{array}{cc} 0 & R \\ S & 0 \end{array}\right]\right)=\frac{\Vert R \Vert +\Vert S\Vert }{2} \end{equation*} where $R, S \in \mathbb{B}(\mathcal{H})$.
We give an alternative lower bound for the numerical radii of Hilbert space operators. As a by-product, we find conditions such that
\begin{equation*} ω\left(\left[\begin{array}{cc} 0 & R \\ S & 0 \end{array}\right]\right)=\frac{\Vert R \Vert +\Vert S\Vert }{2} \end{equation*} where $R, S \in \mathbb{B}(\mathcal{H})$.
△ Less
Submitted 27 March, 2019; v1 submitted 24 May, 2018;
originally announced May 2018.
-
The evolution of the structure of ABC-minimal trees
Authors:
Seyyed Aliasghar Hosseini,
Bojan Mohar,
Mohammad Bagher Ahmadi
Abstract:
The atom-bond connectivity (ABC) index is a degree-based molecular descriptor that found diverse chemical applications. Characterizing trees with minimum ABC-index remained an elusive open problem even after serious attempts and is considered by some as one of the most intriguing open problems in mathematical chemistry. In this paper, we describe the exact structure of the extremal trees with suff…
▽ More
The atom-bond connectivity (ABC) index is a degree-based molecular descriptor that found diverse chemical applications. Characterizing trees with minimum ABC-index remained an elusive open problem even after serious attempts and is considered by some as one of the most intriguing open problems in mathematical chemistry. In this paper, we describe the exact structure of the extremal trees with sufficiently many vertices and we show how their structure evolves when the number of vertices grows. An interesting fact is that their radius is at most~$5$ and that all vertices except for one have degree at most 54. In fact, all but at most $O(1)$ vertices have degree 1, 2, 4, or 53. Let $γ_n = \min\{\abc(T) : T \text{ is a tree of order } n\}$. It is shown that $γ_n = \tfrac{1}{365} \sqrt{\tfrac{1}{53}} \Bigl(1 + 26\sqrt{55} + 156\sqrt{106} \Bigr) n + O(1) \approx 0.67737178\, n + O(1)$.
△ Less
Submitted 4 November, 2019; v1 submitted 5 April, 2018;
originally announced April 2018.
-
Some Investigations about the Properties of Maximum Likelihood Estimations Based on Lower Record Values for a Sub-Family of the Exponential Family
Authors:
Saman Hosseini,
Parviz Nasiri,
Dler Hussein Kadir,
Sharad Damodar Gore
Abstract:
Here, in this paper it has been considered a sub family of exponential family. Maximum likelihood estimations (MLE) for the parameter of this family, probability density function, and cumulative density function based on a sample and based on lower record values have been obtained. It has been considered Mean Square Error (MSE) as a criterion for determining which is better in different situations…
▽ More
Here, in this paper it has been considered a sub family of exponential family. Maximum likelihood estimations (MLE) for the parameter of this family, probability density function, and cumulative density function based on a sample and based on lower record values have been obtained. It has been considered Mean Square Error (MSE) as a criterion for determining which is better in different situations. Additionally, it has been proved some theories about the relations between MLE based on lower record values and based on a random sample. Also, some interesting asymptotically properties for these estimations have been shown during some theories.
△ Less
Submitted 25 September, 2019; v1 submitted 5 November, 2017;
originally announced November 2017.
-
Maximum Likelihood Estimations Based on Upper Record Values for Probability Density Function and Cumulative Distribution Function in Exponential Family and Investigating Some of Their Properties
Authors:
S. D. Gore,
S. Hosseini,
P. Nasiri
Abstract:
In this paper a useful subfamily of the exponential family has been considered. The ML estimation based on upper record values has been calculated for the parameter, Cumulative Density Function, and Probability Density Function of the family. Also, the relations between MLE based on record values and a random sample has been discussed. Additionally, some properties of these estimators have been in…
▽ More
In this paper a useful subfamily of the exponential family has been considered. The ML estimation based on upper record values has been calculated for the parameter, Cumulative Density Function, and Probability Density Function of the family. Also, the relations between MLE based on record values and a random sample has been discussed. Additionally, some properties of these estimators have been investigated. Finally, it has been proven that these estimators have some useful properties for samples with large size
△ Less
Submitted 2 November, 2017; v1 submitted 29 October, 2017;
originally announced October 2017.
-
Finite Differences in Forward and Inverse Imaging Problems--MaxPol Design
Authors:
Mahdi S. Hosseini,
Konstantinos N. Plataniotis
Abstract:
A systematic and comprehensive framework for finite impulse response (FIR) lowpass/fullband derivative kernels is introduced in this paper. Closed form solutions of a number of derivative filters are obtained using the maximally flat technique to regulate the Fourier response of undetermined coefficients. The framework includes arbitrary parameter control methods that afford solutions for numerous…
▽ More
A systematic and comprehensive framework for finite impulse response (FIR) lowpass/fullband derivative kernels is introduced in this paper. Closed form solutions of a number of derivative filters are obtained using the maximally flat technique to regulate the Fourier response of undetermined coefficients. The framework includes arbitrary parameter control methods that afford solutions for numerous differential orders, variable polynomial accuracy, centralized/staggered schemes, and arbitrary side-shift nodes for boundary formulation. Using the proposed framework four different derivative matrix operators are introduced and their numerical stability is analyzed by studying their eigenvalues distribution in the complex plane. Their utility is studied by considering two important image processing problems, namely gradient surface reconstruction and image stitching. Experimentation indicates that the new derivative matrices not only outperform commonly used method but provide useful insights to the numerical issues in these two applications.
△ Less
Submitted 25 September, 2017;
originally announced September 2017.
-
More on the Annihilator-Ideal Graph of a Commutative Ring
Authors:
M. J. Nikmehr,
S. M. Hosseini
Abstract:
Let $R$ be a commutative ring with identity and $\Bbb A (R)$ be the set of ideals of $R$ with non-zero annihilator. The annihilator-ideal graph of $R$, denoted by $A_{I} (R) $, is a simple graph with the vertex set $\Bbb A(R)^{\ast} := \Bbb A (R) \setminus\lbrace (0) \rbrace $, and two distinct vertices $I$ and $J$ are adjacent if and only if…
▽ More
Let $R$ be a commutative ring with identity and $\Bbb A (R)$ be the set of ideals of $R$ with non-zero annihilator. The annihilator-ideal graph of $R$, denoted by $A_{I} (R) $, is a simple graph with the vertex set $\Bbb A(R)^{\ast} := \Bbb A (R) \setminus\lbrace (0) \rbrace $, and two distinct vertices $I$ and $J$ are adjacent if and only if $\mathrm{Ann} _{R} (IJ) \neq \mathrm{Ann} _{R} (I) \cup \mathrm{Ann} _{R} (J)$.
In this paper, we study the affinity between the annihilator-ideal graph and the annihilating-ideal graph $\Bbb A \Bbb G (R)$ (a well-known graph with the same vertices and two distinct vertices $I,J$ are adjacent if and only if $IJ=0$) associated with $R$. All rings whose $A_{I}(R) \neq \Bbb A \Bbb G (R)$ and $\mathrm{gr} (A_{I}(R)) =4$ are characterized. Among other results, we obtain necessary and sufficient conditions under which $A_{I} (R)$ is a star graph.
△ Less
Submitted 15 July, 2017;
originally announced July 2017.
-
Simulated Tornado Optimization
Authors:
S. Hossein Hosseini,
Tohid Nouri,
Afshin Ebrahimi,
S. Ali Hosseini
Abstract:
We propose a swarm-based optimization algorithm inspired by air currents of a tornado. Two main air currents - spiral and updraft - are mimicked. Spiral motion is designed for exploration of new search areas and updraft movements is deployed for exploitation of a promising candidate solution. Assignment of just one search direction to each particle at each iteration, leads to low computational com…
▽ More
We propose a swarm-based optimization algorithm inspired by air currents of a tornado. Two main air currents - spiral and updraft - are mimicked. Spiral motion is designed for exploration of new search areas and updraft movements is deployed for exploitation of a promising candidate solution. Assignment of just one search direction to each particle at each iteration, leads to low computational complexity of the proposed algorithm respect to the conventional algorithms. Regardless of the step size parameters, the only parameter of the proposed algorithm, called tornado diameter, can be efficiently adjusted by randomization. Numerical results over six different benchmark cost functions indicate comparable and, in some cases, better performance of the proposed algorithm respect to some other metaheuristics.
△ Less
Submitted 31 December, 2016;
originally announced January 2017.
-
Lambert Multipliers Between $L^P$-spaces as a Banach Algebra
Authors:
Jahangir Cheshmavar,
Seyed Kamel Hosseini
Abstract:
For 1 ⤠p < â, it is known that the set K^*_p contains of all Lambert multipliers acting between L^p-spaces is a Banach space. In this study, we introduce a new induced norm by conditional expectation operators to show that K^*_p is a commutative Banach algebra with respect to this norm. Furthermore, in main result, the Fredholm *-multiplication operators on L^p-spaces are characterized, and…
▽ More
For 1 ⤠p < â, it is known that the set K^*_p contains of all Lambert multipliers acting between L^p-spaces is a Banach space. In this study, we introduce a new induced norm by conditional expectation operators to show that K^*_p is a commutative Banach algebra with respect to this norm. Furthermore, in main result, the Fredholm *-multiplication operators on L^p-spaces are characterized, and some more results are obtained.
△ Less
Submitted 23 May, 2020; v1 submitted 20 September, 2016;
originally announced September 2016.
-
Almost All Regular Graphs are Normal
Authors:
Seyed Saeed Changiz Rezaei,
Seyyed Aliasghar Hosseini,
Bojan Mohar
Abstract:
In 1999, De Simone and Körner conjectured that every graph without induced $C_5,C_7,\overline{C}_7$ contains a clique cover $\mathcal C$ and a stable set cover $\mathcal I$ such that every clique in $\mathcal C$ and every stable set in $\mathcal I$ have a vertex in common. This conjecture has roots in information theory and became known as the Normal Graph Conjecture. Here we prove that all graphs…
▽ More
In 1999, De Simone and Körner conjectured that every graph without induced $C_5,C_7,\overline{C}_7$ contains a clique cover $\mathcal C$ and a stable set cover $\mathcal I$ such that every clique in $\mathcal C$ and every stable set in $\mathcal I$ have a vertex in common. This conjecture has roots in information theory and became known as the Normal Graph Conjecture. Here we prove that all graphs of bounded maximum degree and sufficiently large odd girth (linear in the maximum degree) are normal. This implies that for every fixed $d$, random $d$-regular graphs are a.a.s. normal.
△ Less
Submitted 24 November, 2015;
originally announced November 2015.
-
Online Distributed Optimization on Dynamic Networks
Authors:
Saghar Hosseini,
Airlie Chapman,
Mehran Mesbahi
Abstract:
This paper presents a distributed optimization scheme over a network of agents in the presence of cost uncertainties and over switching communication topologies. Inspired by recent advances in distributed convex optimization, we propose a distributed algorithm based on a dual sub-gradient averaging. The objective of this algorithm is to minimize a cost function cooperatively. Furthermore, the algo…
▽ More
This paper presents a distributed optimization scheme over a network of agents in the presence of cost uncertainties and over switching communication topologies. Inspired by recent advances in distributed convex optimization, we propose a distributed algorithm based on a dual sub-gradient averaging. The objective of this algorithm is to minimize a cost function cooperatively. Furthermore, the algorithm changes the weights on the communication links in the network to adapt to varying reliability of neighboring agents. A convergence rate analysis as a function of the underlying network topology is then presented, followed by simulation results for representative classes of sensor networks.
△ Less
Submitted 22 December, 2014;
originally announced December 2014.
-
Online Distributed ADMM on Networks
Authors:
Saghar Hosseini,
Airlie Chapman,
Mehran Mesbahi
Abstract:
This paper examines online distributed Alternating Direction Method of Multipliers (ADMM). The goal is to distributively optimize a global objective function over a network of decision makers under linear constraints. The global objective function is composed of convex cost functions associated with each agent. The local cost functions, on the other hand, are assumed to have been decomposed into t…
▽ More
This paper examines online distributed Alternating Direction Method of Multipliers (ADMM). The goal is to distributively optimize a global objective function over a network of decision makers under linear constraints. The global objective function is composed of convex cost functions associated with each agent. The local cost functions, on the other hand, are assumed to have been decomposed into two distinct convex functions, one of which is revealed to the decision makers over time and one known a priori. In addition, the agents must achieve consensus on the global variable that relates to the private local variables via linear constraints. In this work, we extend online ADMM to a distributed setting based on dual-averaging and distributed gradient descent. We then propose a performance metric for such online distributed algorithms and explore the performance of the sequence of decisions generated by the algorithm as compared with the best fixed decision in hindsight. This performance metric is called the social regret. A sub-linear upper bound on the social regret of the proposed algorithm is then obtained that underscores the role of the underlying network topology and certain condition measures associated with the linear constraints. The online distributed ADMM algorithm is then applied to a formation acquisition problem demonstrating the application of the proposed setup in distributed robotics.
△ Less
Submitted 2 October, 2015; v1 submitted 22 December, 2014;
originally announced December 2014.
-
On General Closure Operators and Quasi Factorization Structures
Authors:
S. Sh. Mousavi,
S. N. Hosseini,
A. Ilaghi-Hosseini
Abstract:
In this article the notions of (quasi weakly hereditary) general closure operator $\mb{C}$ on a category $\cx$ with respect to a class $\cm$ of morphisms, and quasi factorization structures in a category $\cx$ are introduced. It is shown that under certain conditions, if $(\ce, \cm)$ is a quasi factorization structure in $\cx$, then $\cx$ has quasi right $\cm$-factorization structure and quasi lef…
▽ More
In this article the notions of (quasi weakly hereditary) general closure operator $\mb{C}$ on a category $\cx$ with respect to a class $\cm$ of morphisms, and quasi factorization structures in a category $\cx$ are introduced. It is shown that under certain conditions, if $(\ce, \cm)$ is a quasi factorization structure in $\cx$, then $\cx$ has quasi right $\cm$-factorization structure and quasi left $\ce$-factorization structure. It is also shown that for a quasi weakly hereditary and quasi idempotent QCD-closure operator with respect to a certain class $\cm$, every quasi factorization structure $(\ce, \cm)$ yields a quasi factorization structure relative to the given closure operator; and that for a closure operator with respect to a certain class $\cm$, if the pair of classes of quasi dense and quasi closed morphisms forms a quasi factorization structure, then the closure operator is both quasi weakly hereditary and quasi idempotent. Several illustrative examples are furnished.
△ Less
Submitted 3 September, 2019; v1 submitted 22 December, 2014;
originally announced December 2014.
-
High-Accuracy Total Variation for Compressed Video Sensing
Authors:
Mahdi S. Hosseini,
Konstantinos N. Plataniotis
Abstract:
Numerous total variation (TV) regularizers, engaged in image restoration problem, encode the gradients by means of simple $[-1,1]$ FIR filter. Despite its low computational processing, this filter severely deviates signal's high frequency components pertinent to edge/discontinuous information and cause several deficiency issues known as texture and geometric loss. This paper addresses this problem…
▽ More
Numerous total variation (TV) regularizers, engaged in image restoration problem, encode the gradients by means of simple $[-1,1]$ FIR filter. Despite its low computational processing, this filter severely deviates signal's high frequency components pertinent to edge/discontinuous information and cause several deficiency issues known as texture and geometric loss. This paper addresses this problem by proposing an alternative model to the TV regularization problem via high order accuracy differential FIR filters to preserve rapid transitions in signal recovery. A numerical encoding scheme is designed to extend the TV model into multidimensional representation (tensorial decomposition). We adopt this design to regulate the spatial and temporal redundancy in compressed video sensing problem to jointly recover frames from under-sampled measurements. We then seek the solution via alternating direction methods of multipliers and find a unique solution to quadratic minimization step with capability of handling different boundary conditions. The resulting algorithm uses much lower sampling rate and highly outperforms alternative state-of-the-art methods. This is evaluated both in terms of restoration accuracy and visual quality of the recovered frames.
△ Less
Submitted 4 March, 2014; v1 submitted 1 September, 2013;
originally announced September 2013.
-
Bayesian Inference For Exponential Distribution Based On Upper Record Range
Authors:
P. Nasiri,
S. Hosseini,
M. Yarmohammadi,
F. Hatami
Abstract:
This paper deals with Bayesian estimations of scale parameter of the exponential distribution based on upper record range (Rn). This has been done in two steps; point and interval. In the first step the quadratic, squared error and absolute error, loss functions have been considered to obtain Bayesian-point estimations. Also in the next step the shortest Bayes interval (Hight Posterior Density int…
▽ More
This paper deals with Bayesian estimations of scale parameter of the exponential distribution based on upper record range (Rn). This has been done in two steps; point and interval. In the first step the quadratic, squared error and absolute error, loss functions have been considered to obtain Bayesian-point estimations. Also in the next step the shortest Bayes interval (Hight Posterior Density interval) and Bayes interval with equal tails based on upper record range have been found. Therefore, the Homotopy Perturbation Method(HPM) has been applied to obtain the limits of Hight Posterior Density intervals. Moreover, efforts have been made to meet the admissibility conditions for linear estimators based on upper record range of the form mRn+d by obtained Bayesian point estimations. So regarding the consideration of loss functions, the prior distribution between the conjunction family has been chosen to be able to produce the linear estimations from upper record range statistics. Finally, some numerical examples and simulations have been presented.
△ Less
Submitted 25 June, 2013;
originally announced June 2013.
-
Extended affinization of Invariant Affine Reflection Algebras
Authors:
Saeid Azam,
S. Reza Hosseini,
Malihe Yousofzadeh
Abstract:
Invariant affine reflection algebras are the last and the most general known extension of affine Kac-Moody Lie algebras, introduced in recent years. We develop a method known as "affinization" to the class of invariant affine reflection algebras, and show that starting with an algebra from this class together with a certain finite order automorphism, and applying the so called "loop construction",…
▽ More
Invariant affine reflection algebras are the last and the most general known extension of affine Kac-Moody Lie algebras, introduced in recent years. We develop a method known as "affinization" to the class of invariant affine reflection algebras, and show that starting with an algebra from this class together with a certain finite order automorphism, and applying the so called "loop construction", we obtain again an invariant affine reflection algebra. This can be considered as an important step towards the realization of invariant affine reflection algebras.
△ Less
Submitted 31 August, 2011; v1 submitted 9 May, 2011;
originally announced May 2011.
-
Voltage/Pitch Control for Maximization and Regulation of Active/Reactive Powers in Wind Turbines with Uncertainties
Authors:
Yi Guo,
S. Hossein Hosseini,
John N. Jiang,
Choon Yik Tang,
Rama G. Ramakumar
Abstract:
This paper addresses the problem of controlling a variable-speed wind turbine with a Doubly Fed Induction Generator (DFIG), modeled as an electromechanically-coupled nonlinear system with rotor voltages and blade pitch angle as its inputs, active and reactive powers as its outputs, and most of the aerodynamic and mechanical parameters as its uncertainties. Using a blend of linear and nonlinear con…
▽ More
This paper addresses the problem of controlling a variable-speed wind turbine with a Doubly Fed Induction Generator (DFIG), modeled as an electromechanically-coupled nonlinear system with rotor voltages and blade pitch angle as its inputs, active and reactive powers as its outputs, and most of the aerodynamic and mechanical parameters as its uncertainties. Using a blend of linear and nonlinear control strategies (including feedback linearization, pole placement, uncertainty estimation, and gradient-based potential function minimization) as well as time-scale separation in the dynamics, we develop a controller that is capable of maximizing the active power in the Maximum Power Tracking (MPT) mode, regulating the active power in the Power Regulation (PR) mode, seamlessly switching between the two modes, and simultaneously adjusting the reactive power to achieve a desired power factor. The controller consists of four cascaded components, uses realistic feedback signals, and operates without knowledge of the C_p-surface, air density, friction coefficient, and wind speed. Finally, we show the effectiveness of the controller via simulation with a realistic wind profile.
△ Less
Submitted 7 August, 2010; v1 submitted 28 June, 2010;
originally announced June 2010.
-
Correction to: "Blind maximum likelihood separation of a linear-quadratic mixture"
Authors:
Shahram Hosseini,
Yannick Deville
Abstract:
An error occurred in the computation of a gradient in our paper entitled "Blind maximum likelihood separation of a linear-quadratic mixture", presented in ICA'2004. The equations (20) in Appendix and (17) in the text were not correct. The current paper presents the correct version of these equations.
An error occurred in the computation of a gradient in our paper entitled "Blind maximum likelihood separation of a linear-quadratic mixture", presented in ICA'2004. The equations (20) in Appendix and (17) in the text were not correct. The current paper presents the correct version of these equations.
△ Less
Submitted 6 January, 2010;
originally announced January 2010.
-
Effect of indirect dependencies on "A mutual information minimization approach for a class of nonlinear recurrent separating systems"
Authors:
Yannick Deville,
Alain Deville,
Shahram Hosseini
Abstract:
In a recent paper [4], Duarte and Jutten investigated the Blind Source Separation (BSS) problem, for the nonlinear mixing model that they introduced in that paper. They proposed to solve this problem by using information-theoretic tools, more precisely by minimizing the mutual information (MI) of the outputs of the separating structure. When applying the MI approach to BSS problems, one usually…
▽ More
In a recent paper [4], Duarte and Jutten investigated the Blind Source Separation (BSS) problem, for the nonlinear mixing model that they introduced in that paper. They proposed to solve this problem by using information-theoretic tools, more precisely by minimizing the mutual information (MI) of the outputs of the separating structure. When applying the MI approach to BSS problems, one usually determines the analytical expressions of the derivatives of the MI with respect to the parameters of the considered separating model. In the literature, these calculations were mainly reported for linear mixtures up to now. They are more complex for nonlinear mixtures, due to dependencies between the considered quantities. Moreover, the notations commonly employed by the BSS community in such calculations may become misleading when using them for nonlinear mixtures, due to the above-mentioned dependencies. We claim that the calculations reported in [4] contain an error, because they did not take into account all these dependencies. In this document, we therefore explain this phenomenon, by showing the effect of indirect dependencies on the application of the MI approach to the mixing and separating models considered in [4]. We thus introduce a corrected expression of the gradient of the considered BSS criterion based on MI. This correct gradient may then e.g. be used to optimize the adaptive coefficients of the considered separating system by means of the well-known gradient descent algorithm. As explained hereafter, this investigation has some similarities with an analysis that we previously reported in another arXiv document [3]. However, these two investigations concern different problems (mixture and separating structure, mathematical tools: see paper).
△ Less
Submitted 25 October, 2009; v1 submitted 23 October, 2009;
originally announced October 2009.
