-
Single Element Error Correction/ in a Euclidean Distance Matrix
Authors:
Abdo Alfakih,
Woosuk L. Jung,
Henry Wolkowicz,
Tina Xu
Abstract:
We consider the \emph{exact} error correction of a noisy Euclidean distance matrix, EDM, where the elements are the squared distances between $n$ points in $R^d$. For our problem we are given two facts: (i) the embedding dimension, $d$, (ii) \emph{exactly one} distance in the data is corrupted by \emph{nonzero noise}. But we do \underline{not} know the magnitude nor position of the noise. Thus the…
▽ More
We consider the \emph{exact} error correction of a noisy Euclidean distance matrix, EDM, where the elements are the squared distances between $n$ points in $R^d$. For our problem we are given two facts: (i) the embedding dimension, $d$, (ii) \emph{exactly one} distance in the data is corrupted by \emph{nonzero noise}. But we do \underline{not} know the magnitude nor position of the noise. Thus there is a combinatorial element to the problem. We present three solution techniques. These use three divide and conquer strategies in combination with three versions of facial reduction that use: exposing vectors, facial vectors, and Gale transforms. This sheds light on the connections between the various forms of facial reduction related to Gale transforms. Our highly successful empirics confirm the success of these approaches as we can solve huge problems of the order of $100,000$ nodes in approximately one minute to machine precision. \\Our algorithm depends on identifying whether a principal submatrix of the \EDM contains the corrupted element. We provide a theorem for doing this that is related to the existing results for identifying \emph{yielding} elements, i.e.,~we provide a characterization for guaranteeing the perturbed EDM remains an EDM with embedding dimension $d$. The characterization is particularly simple in the $d=2$ case. \\In addition, we characterize when the intuitive approach of the nearest EDM problem, solves our problem. In fact, we show that this happens if, and only if, the original distance element is $0$, degenerate, and the perturbation is negative.
△ Less
Submitted 22 June, 2024;
originally announced June 2024.
-
Exact Solutions for the NP-hard Wasserstein Barycenter Problem using a Doubly Nonnegative Relaxation and a Splitting Method
Authors:
Abdo Alfakih,
Jeffery Cheng,
Woosuk L. Jung,
Walaa M. Moursi,
Henry Wolkowicz
Abstract:
The simplified Wasserstein barycenter problem consists in selecting one point from $k$ given sets, each set consisting of $n$ points, with the aim of minimizing the sum of distances to the barycenter of the $k$ points chosen. This problem is known to be NP-hard. We compute the Wasserstein barycenter by exploiting the Euclidean distance matrix structure to obtain a facially reduced doubly nonnegati…
▽ More
The simplified Wasserstein barycenter problem consists in selecting one point from $k$ given sets, each set consisting of $n$ points, with the aim of minimizing the sum of distances to the barycenter of the $k$ points chosen. This problem is known to be NP-hard. We compute the Wasserstein barycenter by exploiting the Euclidean distance matrix structure to obtain a facially reduced doubly nonnegative relaxation (DNN). The facial reduction provides a natural splitting for applying the symmetric alternating directions method of multipliers (sADMM) to the DNN relaxation. The sADMM method exploits structure in the subproblems to find strong upper and lower bounds.
The purpose of this paper is twofold. First we want to illustrate the strength of this DNN relaxation with a splitting approach. Our numerical tests then illustrate the surprising success on random problems, as we generally, efficiently, find the provable exact solution of this NP-hard problem. Comparisons with current commercial software illustrate this surprising efficiency. However, we demonstrate and prove that there is a duality gap for problems with enough multiple optimal solutions, and that this arises from problems with highly symmetrized structure.
△ Less
Submitted 8 November, 2023;
originally announced November 2023.
-
A Structure-guided Effective and Temporal-lag Connectivity Network for Revealing Brain Disorder Mechanisms
Authors:
Zhengwang Xia,
Tao Zhou,
Saqib Mamoon,
Amani Alfakih,
Jianfeng Lu
Abstract:
Brain network provides important insights for the diagnosis of many brain disorders, and how to effectively model the brain structure has become one of the core issues in the domain of brain imaging analysis. Recently, various computational methods have been proposed to estimate the causal relationship (i.e., effective connectivity) between brain regions. Compared with traditional correlation-base…
▽ More
Brain network provides important insights for the diagnosis of many brain disorders, and how to effectively model the brain structure has become one of the core issues in the domain of brain imaging analysis. Recently, various computational methods have been proposed to estimate the causal relationship (i.e., effective connectivity) between brain regions. Compared with traditional correlation-based methods, effective connectivity can provide the direction of information flow, which may provide additional information for the diagnosis of brain diseases. However, existing methods either ignore the fact that there is a temporal-lag in the information transmission across brain regions, or simply set the temporal-lag value between all brain regions to a fixed value. To overcome these issues, we design an effective temporal-lag neural network (termed ETLN) to simultaneously infer the causal relationships and the temporal-lag values between brain regions, which can be trained in an end-to-end manner. In addition, we also introduce three mechanisms to better guide the modeling of brain networks. The evaluation results on the Alzheimer's Disease Neuroimaging Initiative (ADNI) database demonstrate the effectiveness of the proposed method.
△ Less
Submitted 1 December, 2022;
originally announced December 2022.
-
On the Colin de Verdiere graph number and penny graphs
Authors:
A. Y. Alfakih
Abstract:
The Colin de Verdiere number of graph G, denoted by μ(G), is a spectral invariant of G that is related to some of its topological properties. For example, μ(G) \leq 3 iff G is planar. A penny graph is the contact graph of equal-radii disks with disjoint interiors in the plane. In this note we prove lower bounds on μ(G) when the complement \bar{G} is a penny graph.
The Colin de Verdiere number of graph G, denoted by μ(G), is a spectral invariant of G that is related to some of its topological properties. For example, μ(G) \leq 3 iff G is planar. A penny graph is the contact graph of equal-radii disks with disjoint interiors in the plane. In this note we prove lower bounds on μ(G) when the complement \bar{G} is a penny graph.
△ Less
Submitted 9 June, 2020;
originally announced June 2020.
-
On Theorems of Sinajova, Rankin and Kuperberg Concerning Spherical Point Configurations
Authors:
A. Y. Alfakih
Abstract:
This note presents simple linear algebraic proofs of theorems due to Sinajova, Rankin and Kuperberg concerning spherical point configurations. The common ingredient in these proofs is the use of spherical Euclidean distance matrices and the Perron-Frobenius theorem.
This note presents simple linear algebraic proofs of theorems due to Sinajova, Rankin and Kuperberg concerning spherical point configurations. The common ingredient in these proofs is the use of spherical Euclidean distance matrices and the Perron-Frobenius theorem.
△ Less
Submitted 16 August, 2019; v1 submitted 2 August, 2019;
originally announced August 2019.
-
On Unit Spherical Euclidean Distance Matrices Which Differ in One Entry
Authors:
A. Y. Alfakih
Abstract:
A unit spherical Euclidean distance matrix (EDM) D is a matrix whose entries can be realized as the interpoint (squared) Euclidean distances of n points on a unit sphere. In this paper, given such a D and 1 \leq k < l \leq n, we present a characterization of the set of all unit spherical EDMs whose entries agree with those of D except possibly with the entry in the klth and lkth positions. As a re…
▽ More
A unit spherical Euclidean distance matrix (EDM) D is a matrix whose entries can be realized as the interpoint (squared) Euclidean distances of n points on a unit sphere. In this paper, given such a D and 1 \leq k < l \leq n, we present a characterization of the set of all unit spherical EDMs whose entries agree with those of D except possibly with the entry in the klth and lkth positions. As a result, we show that this set can be discrete, consisting of one or two elements, or it can be continuous. The results are derived using two alternative approaches, the second of which is based on Cayley-Menger matrices.
△ Less
Submitted 8 April, 2019; v1 submitted 18 March, 2019;
originally announced March 2019.
-
On Representations of Graphs as Two-Distance Sets
Authors:
A. Y. Alfakih
Abstract:
Let a \neq b be two positive scalars. A Euclidean representation of a simple graph G in R^r is a map** of the nodes of G into points in R^r such that the squared Euclidean distance between any two points is a if the corresponding nodes are adjacent and b otherwise. A Euclidean representation is spherical if the points lie on an (r-1)-sphere, and is J-spherical if this sphere has radius 1 and a=2…
▽ More
Let a \neq b be two positive scalars. A Euclidean representation of a simple graph G in R^r is a map** of the nodes of G into points in R^r such that the squared Euclidean distance between any two points is a if the corresponding nodes are adjacent and b otherwise. A Euclidean representation is spherical if the points lie on an (r-1)-sphere, and is J-spherical if this sphere has radius 1 and a=2 < b. Let dim_E(G), dim_S(G) and dim_J(G) denote, respectively, the smallest dimension r for which G admits a Euclidean, spherical and J-spherical representation.
In this paper, we extend and simplify the results of Roy[18] and Nozaki and shinohara[17] by deriving exact simple formulas for dim_E(G) and dim_S(G) in terms of the eigenvalues of V^TAV, where A is the adjacency matrix of G and V is the matrix whose columns form an orthonormal basis for the orthogonal complement of the vector of all 1's.
We also extend and simplify the results of Musin [16] by deriving explicit formulas for determining the J-spherical representation of G and for determining dim_J(G)in terms of the largest eigenvalue of \bar{A}, the adjacency matrix of the complement graph \bar{G}. As a byproduct, we obtain several related results and in particular we answer a question raised by Musin in [16].
△ Less
Submitted 17 August, 2018;
originally announced August 2018.
-
On yielding and jointly yielding entries of Euclidean distance matrices
Authors:
A. Y. Alfakih
Abstract:
An $n \times n$ matrix D is a Euclidean distance matrix (EDM) if there exist $p^1, \ldots, p^n$ in some Euclidean space such that $d_{ij} = || p^i - p^j||^2$ for all $i,j=1,\ldots,n$. Let D be an EDM and let $E^{ij}$ be the $n \times n$ symmetric matrix with 1's in the $ij$th and $ji$th entries and 0's elsewhere. We say that $[l_{ij},u_{ij}]$ is the yielding interval of entry $d_{ij}$ if it holds…
▽ More
An $n \times n$ matrix D is a Euclidean distance matrix (EDM) if there exist $p^1, \ldots, p^n$ in some Euclidean space such that $d_{ij} = || p^i - p^j||^2$ for all $i,j=1,\ldots,n$. Let D be an EDM and let $E^{ij}$ be the $n \times n$ symmetric matrix with 1's in the $ij$th and $ji$th entries and 0's elsewhere. We say that $[l_{ij},u_{ij}]$ is the yielding interval of entry $d_{ij}$ if it holds that $D+t E^{ij}$ is an EDM iff $l_{ij} \leq t \leq u_{ij}$. If the yielding interval of entry $d_{ij}$ has length 0, i.e., if $l_{ij}=u_{ij}$, then $d_{ij}$ is said to be unyielding. Otherwise, if $l_{ij} \neq u_{ij}$, then $d_{ij}$ is said to be yielding. Let $d_{ij}$ and $d_{ik}$ be two unyielding entries of $D$. We say that $d_{ij}$ and $d_{ik}$ are jointly yielding if $D+t_1 E^{ij} + t_2 E^{ik}$ is an EDM for some nonzero scalars $t_1$ and $t_2$. In this paper, we characterize the yielding and the jointly yielding entries of an EDM D in terms of Gale transform of $p^1,\ldots,p^n$. Moreover, for each yielding entry, we present explicit formulae of its yielding interval. Finally, we specialize our results to the case where $p^1,\ldots,p^n$ are in general position.
△ Less
Submitted 5 July, 2018; v1 submitted 22 September, 2016;
originally announced September 2016.
-
Universal Rigidity of Bar Frameworks via the Geometry of Spectrahedra
Authors:
A. Y. Alfakih
Abstract:
A bar framework (G,p) in dimension r is a graph G whose vertices are points p^1,...,p^n in R^r and whose edges are line segments between pairs of these points. Two frameworks (G,p) and (G,q) are equivalent if each edge of (G,p) has the same (Euclidean) length as the corresponding edge of (G,q). A pair of non-adjacent vertices i and j of (G,p)is universally linked if ||p^i-p^j||=||q^i-q^j|| in ever…
▽ More
A bar framework (G,p) in dimension r is a graph G whose vertices are points p^1,...,p^n in R^r and whose edges are line segments between pairs of these points. Two frameworks (G,p) and (G,q) are equivalent if each edge of (G,p) has the same (Euclidean) length as the corresponding edge of (G,q). A pair of non-adjacent vertices i and j of (G,p)is universally linked if ||p^i-p^j||=||q^i-q^j|| in every framework (G,q) that is equivalent to (G,p). Framework (G,p) is universally rigid iff every pair of non-adjacent vertices of (G,p) is universally linked. In this paper, we present a unified treatment of the universal rigidity problem based on the geometry of spectrahedra. A spectrahedron is the intersection of the positive semidefinite cone with an affine space. This treatment makes it possible to tie together some known, yet scattered, results and to derive new ones. Among the new results presented in this paper are: (i) A sufficient condition for a given pair of non-adjacent vertices of (G,p) to be universally linked. (ii) A new, weaker, sufficient condition for a framework (G,p) to be universally rigid thus strengthening the existing known condition. An interpretation of this new condition in terms of the Strong Arnold Property and transversal intersection is also presented.
△ Less
Submitted 2 April, 2015;
originally announced April 2015.
-
Graph connectivity and universal rigidity of bar frameworks
Authors:
A. Y. Alfakih
Abstract:
Let $G$ be a graph on $n$ nodes. In this note, we prove that if $G$ is $(r+1)$-vertex connected, $1 \leq r \leq n-2$, then there exists a configuration $p$ in general position in $R^r$ such that the bar framework $(G,p)$ is universally rigid. The proof is constructive and is based on a theorem by Lovasz et al concerning orthogonal representations and connectivity of graphs [12,13].
Let $G$ be a graph on $n$ nodes. In this note, we prove that if $G$ is $(r+1)$-vertex connected, $1 \leq r \leq n-2$, then there exists a configuration $p$ in general position in $R^r$ such that the bar framework $(G,p)$ is universally rigid. The proof is constructive and is based on a theorem by Lovasz et al concerning orthogonal representations and connectivity of graphs [12,13].
△ Less
Submitted 15 August, 2014; v1 submitted 8 July, 2014;
originally announced July 2014.
-
On Farkas Lemma and Dimensional Rigidity of Bar Frameworks
Authors:
A. Y. Alfakih
Abstract:
We present a new semidefinite Farkas lemma involving a side constraint on the rank. This lemma is then used to present a new proof of a recent characterization, by Connelly and Gortler, of dimensional rigidity of bar frameworks.
We present a new semidefinite Farkas lemma involving a side constraint on the rank. This lemma is then used to present a new proof of a recent characterization, by Connelly and Gortler, of dimensional rigidity of bar frameworks.
△ Less
Submitted 9 May, 2014;
originally announced May 2014.
-
On affine motions and universal rigidity of tensegrity frameworks
Authors:
A. Y. Alfakih,
Viet-Hang Nguyen
Abstract:
Recently, Alfakih and Ye [Lin. Algebra Appl. 438:31--36, 2013] proved that if an $r$-dimensional bar framework $(G,p)$ on $n \geq r+2$ nodes in general position in $\R^r$ admits a positive semidefinite stress matrix with rank $n-r-1$, then $(G,p)$ is universally rigid. In this paper, we generalize this result in two directions. First, we extend this result to tensegrity frameworks. Second, we repl…
▽ More
Recently, Alfakih and Ye [Lin. Algebra Appl. 438:31--36, 2013] proved that if an $r$-dimensional bar framework $(G,p)$ on $n \geq r+2$ nodes in general position in $\R^r$ admits a positive semidefinite stress matrix with rank $n-r-1$, then $(G,p)$ is universally rigid. In this paper, we generalize this result in two directions. First, we extend this result to tensegrity frameworks. Second, we replace the general position assumption by the weaker assumption that in configuration $p$, each point and its neighbors in $G$ affinely span $\R^r$.
△ Less
Submitted 25 May, 2013;
originally announced May 2013.
-
A Remark on the Manhattan Distance Matrix of a Rectangular Grid
Authors:
A. Y. Alfakih
Abstract:
Consider the Quadratic Assignment Problem (QAP): given two matrices A and D, minimize {trace AXDX^T: X is a permutation matrix}. New lower bounds were obtained recently (Mittelmann and peng [8]) for the QAP where D is either the Manhattan distance matrix of a rectangular grid, or the Hamming distance of a hypercube. In this note, we show that the results in [8,11] extend to the case where D is a s…
▽ More
Consider the Quadratic Assignment Problem (QAP): given two matrices A and D, minimize {trace AXDX^T: X is a permutation matrix}. New lower bounds were obtained recently (Mittelmann and peng [8]) for the QAP where D is either the Manhattan distance matrix of a rectangular grid, or the Hamming distance of a hypercube. In this note, we show that the results in [8,11] extend to the case where D is a spherical Euclidean distance matrix, which includes the Manhattan distance matrix and the Hamming distance matrix as special cases.
△ Less
Submitted 25 August, 2012;
originally announced August 2012.
-
On stress matrices of chordal bar frameworks in general position
Authors:
A. Y. Alfakih
Abstract:
A bar framework in R^r, denoted by G(p), is a simple connected graph G whose vertices are points p^1,...,p^n in R^r that affinely span R^r, and whose edges are line segments between pairs of these points. In this paper, we use stress matrices to characterize the universal and global rigidities of chordal bar frameworks in general position in R^r, i.e., bar frameworks where graph G is chordal and t…
▽ More
A bar framework in R^r, denoted by G(p), is a simple connected graph G whose vertices are points p^1,...,p^n in R^r that affinely span R^r, and whose edges are line segments between pairs of these points. In this paper, we use stress matrices to characterize the universal and global rigidities of chordal bar frameworks in general position in R^r, i.e., bar frameworks where graph G is chordal and the points p^1,...,p^n are in general position in R^r. We also prove that if a chordal bar framework in R^r admits a stress matrix of rank n-r-1 with generic rank profile, then it admits a positive semidefinite stress matrix of rank n-r-1.
△ Less
Submitted 17 May, 2012;
originally announced May 2012.
-
Universal rigidity of bar frameworks in general position: a Euclidean distance matrix approach
Authors:
A. Y. Alfakih
Abstract:
A configuration p in r-dimensional Euclidean space is a finite collection of labeled points p^1,p^2,...,p^n in R^r that affinely span R^r. Each configuration p defines a Euclidean distance matrix D_p = (d_ij) = (||p^i-p^j||^2), where ||.|| denotes the Euclidean norm. A fundamental problem in distance geometry is to find out whether or not, a given proper subset of the entries of D_p suffices to un…
▽ More
A configuration p in r-dimensional Euclidean space is a finite collection of labeled points p^1,p^2,...,p^n in R^r that affinely span R^r. Each configuration p defines a Euclidean distance matrix D_p = (d_ij) = (||p^i-p^j||^2), where ||.|| denotes the Euclidean norm. A fundamental problem in distance geometry is to find out whether or not, a given proper subset of the entries of D_p suffices to uniquely determine the entire matrix D_p. This problem is known as the universal rigidity problem of bar frameworks. In this chapter, we present a unified approach for the universal rigidity of bar frameworks, based on Euclidean distance matrices (EDMs), or equivalently, on projected Gram matrices. This approach makes the universal rigidity problem amenable to semi-definite programming methodology. Using this approach, we survey some recently obtained results and their proofs, emphasizing the case where the points p^1,...,p^n are in general position.
△ Less
Submitted 14 January, 2012;
originally announced January 2012.
-
On affine motions and bar frameworks in general position
Authors:
A. Y. Alfakih,
Yinyu Ye
Abstract:
A configuration p in r-dimensional Euclidean space is a finite collection of points (p^1,...,p^n) that affinely span R^r. A bar framework, denoted by G(p), in R^r is a simple graph G on n vertices together with a configuration p in R^r. A given bar framework G(p) is said to be universally rigid if there does not exist another configuration q in any Euclidean space, not obtained from p by a rigid m…
▽ More
A configuration p in r-dimensional Euclidean space is a finite collection of points (p^1,...,p^n) that affinely span R^r. A bar framework, denoted by G(p), in R^r is a simple graph G on n vertices together with a configuration p in R^r. A given bar framework G(p) is said to be universally rigid if there does not exist another configuration q in any Euclidean space, not obtained from p by a rigid motion, such that ||q^i-q^j||=||p^i-p^j|| for each edge (i,j) of G. It is known that if configuration p is generic and bar framework G(p) in R^r admits a positive semidefinite stress matrix S of rank n-r-1, then G(p) is universally rigid. Connelly asked whether the same result holds true if the genericity assumption of p is replaced by the weather assumption of general position. We answer this question in the affirmative in this paper.
△ Less
Submitted 16 September, 2010;
originally announced September 2010.
-
Toward the Universal Rigidity of General Frameworks
Authors:
Abdo Y. Alfakih,
Nicole Taheri,
Yinyu Ye
Abstract:
Let (G,P) be a bar framework of n vertices in general position in R^d, d <= n-1, where G is a (d+1)-lateration graph. In this paper, we present a constructive proof that (G,P) admits a positive semi-definite stress matrix with rank n-d-1. We also prove a similar result for a sensor network where the graph consists of m(>= d+1) anchors.
Let (G,P) be a bar framework of n vertices in general position in R^d, d <= n-1, where G is a (d+1)-lateration graph. In this paper, we present a constructive proof that (G,P) admits a positive semi-definite stress matrix with rank n-d-1. We also prove a similar result for a sensor network where the graph consists of m(>= d+1) anchors.
△ Less
Submitted 6 January, 2011; v1 submitted 6 September, 2010;
originally announced September 2010.
