MPE Mathematical Problems in Engineering 1563-5147 1024-123X Hindawi Publishing Corporation 175769 10.1155/2013/175769 175769 Research Article New Inequalities between Information Measures of Network Information Content Popescu Pantelimon-George 0000-0002-4566-1545 Pop Florin Herişanu Alexandru Ţăpuş Nicolae Cattani Carlo 1 Computer Science Department Faculty of Automatic Control and Computers University Politehnica of Bucharest Splaiul Independentei 313 Bucharest 060042 Romania pub.ro 2013 20 3 2013 2013 12 12 2012 28 01 2013 2013 Copyright © 2013 Pantelimon-George Popescu et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

We refine a classical logarithmic inequality using a discrete case of Bernoulli inequality, and then we refine furthermore two information inequalities between information measures for graphs, based on information functionals, presented by Dehmer and Mowshowitz in (2010) as Theorems 4.7 and 4.8. The inequalities refer to entropy-based measures of network information content and have a great impact for information processing in complex networks (a subarea of research in modeling of complex systems).

1. Introduction

We first present the classical Shannon’s entropy; see [1, 2]. Let X be a discrete random variable taking values in A, and let p(xi)=Pr(X=xi) be the probability mass function of X; then, the Shannon entropy of X is defined by (1)H(X)=-xiAp(xi)log(p(xi)). We continue with some basic theoretical information about graphs; see [3, 4]. We consider G=(V,E),|V|< to be a finite undirected graph, where E(V2). G is connected if for any arbitrary vertices vi,vj exists an undirected path from vi to vj. Furthermore, we will consider 𝒢UC to be the set of finite undirected and connected graphs. This is usually used in strategies for improving transport efficiency in Network Information Content, including designing efficient routing strategies and making appropriate adjustments to the underlying network structure .

Now, taking into account the work presented in [6, 7], we define the concept of information functional. Let G=(V,E)𝒢UC, and let S initially be an abstract set; then, f:SR+ is called the information functional of G (we assume that f is monotonous). The abstract set S can be, for example, the vertex sets, a set of paths, certain subgraphs, and so forth. The set S is used to define the functional f that captures the structural information of graph G and has to be defined concretely.

Using different information functionals to measure the entropy of a graph, we obtain different probability distributions; so, the resulting graph entropies are also different. Further, we define, as in [6, 7], the entropy of a graph G=(V,E)𝒢UC with arbitrary vertex labels vi, using an arbitrary information functional f, to be (2)If(G)=-i=1|V|f(vi)j=1|V|f(vj)log(f(vi)j=1|V|f(vj)), and considering a vertex viV, pf(vi)=f(vi)/j=1|V|f(vj) interpreted as vertex probabilities (i=1|V|pf(vi)=1), we conclude that (3)If(G)=-i=1|V|pf(vi)log(pf(vi)).

2. Refinement of a Classical Logarithmic Inequality

It is well known from the literature that (4)logb(x)1ln(b)(x-1),x>0. Trying to refine this inequality, we consider the classical discrete Bernoulli inequality, as follows.

Theorem 1 (Bernoulli inequality).

Let ai0,xi>-1,i=1,,n and i=1nai1. Then, (5)i=1n(1+xi)ai1+i=1naixi, if ai1 or ai0, and if xi>0, or -1<xi<0,i=1,,n, then (6)i=1n(1+xi)ai1+i=1naixi. Now, using a factorization of a real number, one concludes as follows.

Theorem 2.

Let x>0 be a real number; then, (7)logb(x)1ln(b)i=1nxi1ln(b)(x-1), where n2 and x=i=1n(xi+1), with xi>0 if x>1, xi=0 if x=1, and 0>xi>-1 if 1>x>0, 1in.

Proof.

The inequality becomes an identity for x=1; that is, xi=0,1in. We will treat the other two cases, when x>1 and when 1>x>0 in the same manner, as follows: (8)logb(x)=logbi=1n(xi+1)=i=1nlogb(xi+1), for which applying the classical logarithmic inequality for xi+1>0,1in yields (9)logb(x)1ln(b)i=1nxi, and making use of the second part of the discrete Bernoulli inequality, previously presented, we obtain (10)1ln(b)i=1nxi1ln(b)[i=1n(xi+1)-1]=1ln(b)(x-1), and we are done.

3. Refinement of Dehmer and Mowshowitz Inequalities between Information Measures

In the work of Dehmer and Mowshowitz , by Theorems4.7 and 4.8 are presented two information inequalities derived assuming only the characteristic properties of the functions involved. Here, we refine those results by using the previous theorem, as follows.

Theorem 3.

Let f and f* be information functionals. Then (all logarithms are in base 2), (11)If(G)-k=1|V|pf(vk)log(pf*(vk))-1ln(2)i=1npikk=1|V|pf(vk)-k=1|V|pf(vk)log(pf*(vk))-1ln(2)k=1|V|(pf(vk))2-pf(vk)pf*(vk)pf*(vk),If*(G)1ln(2)i=1npikk=1|V|pf*(vk)-k=1|V|pf*(vk)log(pf(vk))1ln(2)k=1|V|[(pf(vk))pf(vk)-pf*(vk)-ln(2)pf*(vk)log(pf(vk))], where n2 and pf(vk)/pf*(vk)=i=1n(pik+1),1k|V|, with 1k|V|, pik>0 if pf(vk)>pf*(vk), pik=0 if pf(vk)=pf*(vk), and 0>pik>-1 if pf(vk)<pf*(vk), 1in.

Proof.

Applying Theorem 2 for x=pf(vk)/pf*(vk)>0 yields that (12)log(pf(vk))-log(pf*(vk))1ln(2)i=1npik1ln(2)pf(vk)-pf*(vk)pf*(vk),1k|V|. By multiplying this inequality first with -pf(vk) and then with pf*(vk), we obtain (13)-pf(vk)log(pf(vk))-1ln(2)i=1npik·pf(vk)-pf(vk)log(pf*(vk))-1ln(2)(pf(vk))2-pf(vk)pf*(vk)pf*(vk)-pf(vk)log(pf*(vk)),-pf*(vk)log(pf*(vk))1ln(2)i=1npik·pf*(vk)-pf*(vk)log(pf(vk))1ln(2)(pf(vk)-pf*(vk))-pf*(vk)log(pf(vk)),1k|V|. Finally, summating after k yields the wanted results.

4. Conclusions

As a main result of this paper, we improved two inequalities between entropy-based measures, this giving us a way of bounding measures of network properties. A good example for this purpose is presented into , where, using concrete information functionals, the corresponding entropies for the graphs in a set of 2265 nonisomorphic chemical graphs, that is, in MS2265, are computed. Here, new stronger lower and, respectively, upper bounds for entropy-based measures are given and thus, by knowing the limitations of such measures, we narrow down the set of expected values. The utility of inequalities between measures of structural information content is found in problem solving through problem transformation, as is also spotted in .

Acknowledgment

This work was supported and founded by the project “ERRIC-Empowering Romanian Research on Intelligent Information Technologies/FP7-REGPOT-2010-1”, ID: 264207.

Ash R. Information Theory 1965 New York, NY, USA Interscience MR0229475 Cover T. M. Thomas J. A. Elements of Information Theory 2006 John Wiley & Sons MR2239987 Buckley F. Harary F. Distance in Graphs 1990 Addison-Wesley MR1045632 Harary F. Graph Theory 1969 Reading, Mass, USA Addison-Wesley MR0256911 Chen S. Huang W. Cattani C. Altieri G. Traffic dynamics on complex networks: a survey Mathematical Problems in Engineering 2012 2012 23 732698 10.1155/2012/732698 MR2837454 Dehmer M. Mowshowitz A. Inequalities for entropy-based measures of network information content Applied Mathematics and Computation 2010 215 12 4263 4271 10.1016/j.amc.2009.12.051 MR2596104 ZBL1220.05125 Dehmer M. Information processing in complex networks: graph entropy and information functionals Applied Mathematics and Computation 2008 201 1-2 82 94 10.1016/j.amc.2007.12.010 MR2432584 ZBL1152.05361