WEAKLY TIGHT FUNCTIONS AND THEIR DECOMPOSITION MONA KHARE AND BHAWNA SINGH Received 2 September 2004 and in revised form 30 December 2004

MONA KHARE AND BHAWNA SINGH

Received 2 September 2004 and in revised form 30 December 2004

The present paper deals with the study of a weakly tight function and its relation to tight functions. We obtain a Jordan-decomposition-type theorem for a locally bounded weakly tight real-valued function defined on a sublattice ofI^X, followed by the notion of a total variation.

1. Introduction

The notion of a signed measure arises if a measure is allowed to take on both positive and negative values. A set that is both positive and negative with respect to a signed measure is termed as a null set. Some concepts in measure theory can be generalized by means of classes of null sets. An abstract formulation and proof of the Lebesgue decomposition theorem using the concept of null sets is given by Ficker [5]. A real-valued function satis- fying certain properties that can be expressed as a difference of two nonnegative functions possessing the same properties is called “decomposable.” Several Jordan-decomposition- type theorems are exhibited in [3]. Faires and Morrison [4] exposed conditions on a vector-valued measure that ensure vector-valued Jordan-decomposition-type theorem to hold. For a signed null-additive fuzzy measure, a Jordan-decomposition-type theorem is investigated by Pap in [11].

The problem of generation of measures by tight functions defined on a lattice of sets has been taken up by several authors [1,2,6,8,9]. Nayak and Srinivasan [10] initiated a weaker form of tightness for a real-valued function µdefined on a lattice of sets to decomposeµas a differenceµ⁺−µ⁻and then extended it to a countably additive measure.

InSection 2, we have defined and studied the notions of measuring envelopes, modu- lar functions, and additive functions. The notions of superadditive and subadditive func- tions are also given with the help of pointwise addition of elements in I^X. The lower envelopeβ_∗of a superadditive functionβdefined on a sublatticeKofI^Xturns out to be superadditive. InSection 3, we introduce the notion of a weakly tight functionβ:K→R, whereKis a sublattice ofI^Xcontaining 0 and 1 (cf. [10]). The condition imposed on the [0, 1]-valued functionβto be a weakly tight function is less restrictive than that for being a tight function. It is proved that a superadditive, monotone, and weakly tight function

Copyright©2005 Hindawi Publishing Corporation

International Journal of Mathematics and Mathematical Sciences 2005:18 (2005) 2991–2998 DOI:10.1155/IJMMS.2005.2991

βis tight (thereK is taken to be closed under addition). InSection 4, our main result states that a locally bounded, weakly tight real-valued functionβdefined onKhas a rep- resentation of the formβ⁺−β⁻; bothβ⁺andβ⁻are nonnegative monotone (and hence, locally bounded) functions defined onK. If, in addition,βis additive and modular, then the decomposed partsβ⁺andβ⁻preserve superadditive and supermodular properties.

The total variation|β|ofβis defined as the sum ofβ⁺andβ⁻, following the terminology in the classical measure theory (cf. [12]); few properties of|β|are noted.

Notations. Throughout this paper,Xis a nonempty set andI≡[0, 1] is the unit interval of the real lineR;Cdenotes a subfamily ofI^Xof all functions fromXtoI;Kstands for a sublattice ofI^Xcontaining the least element 0 and the greatest element 1, where 0 and 1 are constant functions sending eachx∈Xto 0 and 1, respectively. We will denote byβa function fromKtoIsatisfyingβ(0)=0.

2. Measuring envelopes

LetCbe a sublattice ofI^Xand letξ:C→Ibe a function. Thenξis calledmonotoneif f, g∈C,g≤f ⇒ξ(g)≤ξ(f). The mappingξis calledsupermodular(submodular, resp.) if for f,g∈C,ξ(f) +ξ(g)≤ξ(f ∨g) +ξ(f ∧g)(ξ(f) +ξ(g)≥ξ(f ∨g) +ξ(f ∧g), resp.);

ξ is said to be modularif it is both supermodular and submodular. The mappingξ is said to besuperadditive(subadditive, resp.) if for f1,f2∈Csuch that f1+f2∈C, ξ(f1+ f2)≥ξ(f1) +ξ(f2) (ξ(f1+f2)≤ξ(f1) +ξ(f2), resp.);ξ is said to beadditiveif it is both superadditive and subadditive. If we restrict ourselves to disjoint crisp sets inC, then the condition of being additive forξcoincides with that of [10].

The familyC⊆I^X is said to beclosed under additionif for f,g∈Cwith f +g∈I^X, we have f +g∈C, and is said to beclosed under addition modulo1 (orclosed under⊕) if

f,g∈Cimplies that f⊕g∈C, where f ⊕g=(f+g)∧1.

A functionξ:C→Iis calledsubadditive modulo1 if forf1,f2∈Cwithf1⊕f2∈C, we haveξ(f1⊕f2)≤ξ(f1) +ξ(f2); hereC⊆I^X.

If f,g∈I^X and f +g∈I^X, then f ⊕g= f +g. Thus ifCis closed under addition modulo 1, thenCis closed under addition. Also, ifξ is subadditive modulo 1, thenξ is subadditive.

The definition for a functionξ:C→Ito be supermodular (submodular, superaddi- tive, subadditive, resp.) continues to hold for a real-valued functionξdefined onC.

Definition 2.1. Let β:K→I be a function satisfying β(0)=0. Defineβ∗:I^X→I and β^∗:I^X→Iby

β∗(f)=supβ(g) :g≤ f,g∈K,

β^∗(f)=infβ(g) :g≥f,g∈K, f ∈I^X. (2.1) β_∗andβ^∗are called thelower envelopeand theupper envelopeofβ, respectively.

We obtain

(i)β∗(0)=0=β^∗(0);

(ii) bothβ∗andβ^∗are monotone;

(iii)β^∗|K≤β≤β∗|K;

(iv)βis monotone if and only ifβ∗|K=β=β^∗|K.

Proposition2.2. (i)Ifβis supermodular, thenβ∗is supermodular.

(ii) Ifβis submodular, thenβ^∗is submodular.

Proof. (i) Let f1,f2∈I^X. Letε >0. Then there existg1,g2∈K, f1≥g1, f2≥g2such that β∗

f1

−ε 2< βg1

, β_∗f2

−ε 2< βg2

. (2.2)

It follows that

β_∗f1

+β_∗f2

−ε < βg1∨g2

+βg1∧g2

. (2.3)

Consequently,

β∗ f1

+β∗ f2

−ε < β∗ f1∨f2

+β∗ f1∧f2

. (2.4)

Sinceεis arbitrary, we get β_∗f1

+β_∗f2

≤β_∗f1∨f2

+β_∗f1∧f2

. (2.5)

Proof of (ii) follows analogously.

Proposition2.3. (i)IfKis closed under addition andβis superadditive, thenβ∗is super- additive.

(ii) IfK is closed under addition modulo1andβis subadditive modulo 1, thenβ^∗is subadditive modulo1, and henceβ^∗is subadditive.

Proof. (i) Letf1and f2be inI^Xsuch that f1+ f2∈I^X. Letε >0. Then there existg1,g2∈K withg1≤ f1andg2≤ f2such thatβ(g1)> β∗(f1)−ε/2 andβ(g2)> β∗(f2)−ε/2. Hence, β_∗(f1) +β_∗(f2)−ε < β(g1) +β(g2). Since, for anyx∈X, 0≤(g1+g2)(x)≤(f1+f2)(x)≤ 1, we getg1+g2∈I^X and sog1+g2∈K. Sinceβis superadditive,β∗(f1) +β∗(f2)−ε <

β(g1+g2)≤β∗(f1+f2) and hence, we obtainβ∗(f1) +β∗(f2)≤β∗(f1+f2).

(ii) Let f1,f2∈I^X. Let ε >0. Then there exist g1,g2∈K with f1≤g1 and f2≤g2

such that

β^∗f1

+β^∗f2

+ε > βg1

+βg2

. (2.6)

Sinceg1,g2∈KandKis closed under⊕, we getg1⊕g2∈K. Also,f1⊕f2≤g1⊕g2. Hence (2.6) gives

β^∗f1

+β^∗f2

+ε > βg1⊕g2

≥β^∗f1⊕f2

. (2.7)

Sinceεis arbitrary, we obtain thatβ^∗is subadditive modulo 1.

Definitions 2.4. Letβ1andβ2be real-valued functions defined onK. Define (β1+β2)(f)= β1(f) +β2(f), f ∈K, and (β1−β2)(f)=β1(f)−β2(f), f ∈K. Likewise, forβ:K→R andλ∈R, define (λβ)(f)=λβ(f), f ∈K.

Proposition2.5. (i)Ifβ1 andβ2are supermodular (submodular, resp.), thenβ1+β2 is supermodular (submodular, resp.).

If bothβ1andβ2are modular, then so areβ1+β2andβ1−β2.

(ii) Ifβ is supermodular (submodular, resp.) then λβ is supermodular (submodular, resp.), whereλis a nonnegative real number.

(iii) IfK is closed under addition, andβ1,β2are superadditive (subadditive, resp.), then β1+β2is superadditive (subadditive, resp.). If bothβ1andβ2are additive, then so areβ1+β2

andβ1−β2.

(iv) IfKis closed under addition, andβis superadditive (subadditive, resp.), thenλβis superadditive (subadditive, resp.), whereλis a nonnegative real number.

Proof. We will prove only (i) and (iii).

(i) Letβ1andβ2be supermodular. Letf,g∈K. Then, β1+β2

(f) +β1+β2

(g)=β1(f) +β2(f) +β1(g) +β2(g)

≤β1(f∨g) +β1(f ∧g) +β2(f ∨g) +β2(f∧g)

=β1+β2

(f∨g) +β1+β2 (f ∧g).

(2.8)

Ifβ1andβ2are submodular, then, by similar arguments,β1+β2is submodular.

(iii) LetK be closed under addition. Letβ1 and β2 be superadditive. Let f,g∈K.

Then,

β1+β2

(f) +β1+β2

(g)=β1(f) +β2(f) +β1(g) +β2(g)

≤β1(f+g) +β2(f +g)=β1+β2

(f +g). (2.9)

3. Weakly tight functions

Definition 3.1. Letβ:K→Iwithβ(0)=0. Thenβis calledtight(cotight, resp.) if βf2

=βf1

+β∗ f2−f1

, f1,f2∈K, f1≤ f2, βf2

=βf1

+β^∗f2−f1

, f1,f2∈K, f1≤f2, resp.. (3.1)

Ifβis tight (or cotight), then βis modular and monotone. Furthermore, ifβis tight (cotight, resp.), then β∗ (β^∗, resp.) is an extension ofβ. A detailed study of tight and cotight functions is made in [7,13].

Definition 3.2. Letβ:K→Rbe a function withβ(0)=0. Thenβis calledweakly tight if for every pair f1,f2∈K with f1≤ f2 and for anyε >0, there exists f ∈K such that

f ≤f2−f1and

βf2

−βf1

−β(f)< ε. (3.2)

Proposition3.3. Letβ:K→I be a function satisfyingβ(0)=0. Ifβis tight, thenβis weakly tight.

Proof. Let f1,f2∈Kwith f1≤f2. Letε >0. Sinceβis tight, there exists f ∈K with f ≤ f2−f1such that

β∗ f2−f1

−ε=βf2

−βf1

−ε < β(f) (3.3)

or

βf2

−βf1

−β(f)< ε. (3.4)

Sinceβis monotone and f ≤ f2−f1, we obtain β(f)=β∗(f)≤β∗

f2−f1

=βf2

−βf1

, (3.5)

and so

βf2

−βf1

−β(f)≥0. (3.6)

Thus,|β(f2)−β(f1)−β(f)|< ε.

Proposition3.4. LetKbe closed under addition. Letβ:K→Ibe superadditive, monotone, and weakly tight. Thenβis tight.

Proof. Let f1,f2∈K with f1≤ f2. Letε >0. Sinceβis weakly tight, there exists f ∈K such that f ≤ f2−f1and

βf2

−βf1

−β(f)< ε. (3.7)

Consequently,

βf2

−βf1

< β(f) +ε≤β∗ f2−f1

+ε. (3.8)

Sinceεis arbitrary, we getβ(f2)−β(f1)≤β_∗(f2−f1).

Since, byProposition 2.3(i),β∗is superadditive, we getβ(f2)=β∗(f2)≥β∗(f2−f1) + β∗(f1), which yields thatβ∗(f2−f1)≤β∗(f2)−β∗(f1)=β(f2)−β(f1).

Thusβ_∗(f2−f1)=β(f2)−β(f1), that is,βis tight.

4. A Jordan-decomposition-type theorem

In this section,βis a real-valued function defined on a sublatticeK ofI^X containing 0 and 1. Also, it is assumed throughout this section thatβislocally bounded, that is, for any f inK, sup{β(g) :g≤ f,g∈K}exists. For a locally bounded real-valued functionβ, the definitions of lower and upper envelopes,β∗andβ^∗, ofβmay be given in the same way.

Definition 4.1. For f ∈K, defineβ⁺(f) andβ⁻(f) as follows:

β⁺(f)=supβ(g) :g≤f,g∈K,

β⁻(f)= −infβ(g) :g≤ f,g∈K. (4.1)

Remarks 4.2. (i)β⁺=β∗|K. (ii) β⁻=(−β)⁺;β⁺=(−β)⁻.

(iii) Bothβ⁺andβ⁻are nonnegative, monotone (and hence, locally bounded).

(iv) −β⁻≤β≤β⁺.

Theorem4.3. Letβbe a weakly tight real-valued function defined onK. Thenβ=β⁺−β⁻. Proof. Letε >0. Let f ∈K. Then there exists f1∈Ksuch that f1≤ f and

β⁺(f)< βf1

+ε

2. (4.2)

Sinceβis weakly tight, there exists f2∈Ksuch thatf2≤f −f1and β(f)−

βf1

+βf2<ε

2. (4.3)

This implies that

βf1

+βf2

< β(f) +ε

2. (4.4)

Also, f2≤ f − f1≤ f yields that −β⁻(f)≤β(f2). Hence, using (4.2) and (4.4), we get β⁺(f)−β⁻(f)< β(f1) +ε/2 +β(f2)< β(f) +ε. Sinceεis arbitrary, we get

β⁺−β⁻≤β. (4.5)

Replacingβin (4.5) by−β, we get

(−β)⁺−(−β)⁻≤ −β, (4.6)

or

β⁻−β⁺≤ −β, (4.7)

or

β⁺−β⁻≥β. (4.8)

Thus,β⁺−β⁻=β.

Proposition4.4. LetKbe closed under addition. Ifβis additive, then bothβ⁺andβ⁻are superadditive.

Proof. The proof follows fromProposition 2.3(i).

Proposition4.5. Ifβis modular, then bothβ⁺andβ⁻are supermodular.

Proof. The proof follows fromProposition 2.2(i).

The results obtained in this section may be summarized as follows.

Theorem4.6 (Jordan-decomposition-type theorem). LetKbe a sublattice ofI^Xcontain- ing0and1. Ifβ:K→Ris locally bounded and weakly tight, thenβcan be written as

β=β⁺−β⁻, (4.9)

where bothβ⁺andβ⁻are nonnegative and monotone (and hence, locally bounded) functions defined onK. Furthermore, ifβis modular (βis additive andK is closed under addition), then the decomposed partsβ⁺andβ⁻are supermodular (superadditive).

Definition 4.7. For a functionβ:K→R, define thetotal variationofβ, written as|β|, by

|β| =β⁺+β⁻. (4.10)

Theorem4.8. Letβ:K→Rbe a locally bounded function.

(i) Ifβis weakly tight, thenβ=0⇔ |β| =0.

(ii) For eachf ∈K,|β(f)| ≤ |β|(f).

(iii) Ifβis modular, then|β|is supermodular.

(iv) LetKbe closed under addition. Thenβbeing additive implies that|β|is superaddi- tive.

Proof. (i) Ifβ=0, thenβ⁺=0=β⁻and so|β| =0. Conversely, if|β| =0, then bothβ⁺ andβ⁻vanish. Sinceβis weakly tight, byTheorem 4.3,β=β⁺−β⁻. Henceβ=0.

(ii) Let f ∈K. Then byRemark 4.2(iv),β(f)≤β⁺(f), and−β(f)≤β⁻(f).

Ifβ(f)>0, then

|β|(f)=β⁺(f) +β⁻(f)≥β(f)=β(f). (4.11) Ifβ(f)<0, then

β(f)= −β(f)≤β⁻(f)≤ |β|(f). (4.12) (iii) Follows fromProposition 4.5andProposition 2.5(i).

(iv) Follows fromProposition 4.4andProposition 2.5(iii).

Remark 4.9. If we restrict ourselves to{0, 1}-valued functions inK, thenKmay be viewed as a sublattice ofP(X). For a [0,∞]-valued functionβdefined on this restrictedK over P(X), the definitions of lower and upper envelopes,β∗andβ^∗, reduce to the correspond- ing definitions in classical theory given by Adamski [1]. In this manner, the present study generalizes the theory in [1].

References

[1] W. Adamski,Tight set functions and essential measure, Measure Theory, Oberwolfach 1981 (Oberwolfach, 1981) (D. K¨olzow and D. Maharam-Stone, eds.), Lecture Notes in Math., vol. 945, Springer, Berlin, 1982, pp. 1–14.

[2] J. R. Choksi,On compact contents, J. London Math. Soc.33(1958), 387–398.

[3] J. Diestel and B. Faires,On vector measures, Trans. Amer. Math. Soc.198(1974), 253–271.

[4] B. Faires and T. J. Morrison,The Jordan decomposition of vector-valued measures, Proc. Amer.

Math. Soc.60(1976), 139–143.

[5] V. Ficker,An abstract formation of the Lebesgue decomposition theorem, Proc. Amer. Math. Soc.

60(1976), 101–105.

[6] J. L. Kelley and T. P. Srinivasan,Pre-measures on lattices of sets, Math. Ann.190(1970/1971), 233–241.

[7] M. Khare and B. Singh,Cotight functions and extension to quasi∗-measure, in preparation.

[8] J. Kisy ´nski,On the generation of tight measures, Studia Math.30(1968), 141–151.

[9] P. Morales,Extension of a tight set function with values in a uniform semigroup, Measure Theory, Oberwolfach 1981 (Oberwolfach, 1981) (D. K¨olzow and D. Maharam-Stone, eds.), Lecture Notes in Math., vol. 945, Springer, Berlin, 1982, pp. 282–290.

[10] M. K. Nayak and T. P. Srinivasan,Scalar and vector valued premeasures, Proc. Amer. Math. Soc.

48(1975), 391–396.

[11] E. Pap,Null-Additive Set Functions, Mathematics and Its Applications, vol. 337, Kluwer Aca- demic Publishers, Dordrecht, 1995.

[12] H. L. Royden,Real Analysis, 3rd ed., Macmillan Publishing, New York, 1981.

[13] P. Srivastava, M. Khare, and B. Singh,Premeasure spaces, tight functions and extension to quasi∗- measure, in preparation.

Mona Khare: Department of Mathematics and Statistics, Faculty of Science, University of Allah- abad, Allahabad 211002, Uttar Pradesh, India

E-mail address:raghav9802@rediffmail.com

Bhawna Singh: Allahabad Mathematical Society, 10 CSP Singh Marg, Allahabad 211001, Uttar Pradesh, India

E-mail address:[email protected]

Special Issue on

Time-Dependent Billiards

Call for Papers

This subject has been extensively studied in the past years for one-, two-, and three-dimensional space. Additionally, such dynamical systems can exhibit a very important and still unexplained phenomenon, called as the Fermi acceleration phenomenon. Basically, the phenomenon of Fermi accelera- tion (FA) is a process in which a classical particle can acquire unbounded energy from collisions with a heavy moving wall.

This phenomenon was originally proposed by Enrico Fermi in 1949 as a possible explanation of the origin of the large energies of the cosmic particles. His original model was then modified and considered under different approaches and using many versions. Moreover, applications of FA have been of a large broad interest in many different fields of science including plasma physics, astrophysics, atomic physics, optics, and time-dependent billiard problems and they are useful for controlling chaos in Engineering and dynamical systems exhibiting chaos (both conservative and dissipative chaos).

We intend to publish in this special issue papers reporting research on time-dependent billiards. The topic includes both conservative and dissipative dynamics. Papers dis- cussing dynamical properties, statistical and mathematical results, stability investigation of the phase space structure, the phenomenon of Fermi acceleration, conditions for having suppression of Fermi acceleration, and computational and numerical methods for exploring these structures and applications are welcome.

To be acceptable for publication in the special issue of Mathematical Problems in Engineering, papers must make significant, original, and correct contributions to one or more of the topics above mentioned. Mathematical papers regarding the topics above are also welcome.

Authors should follow the Mathematical Problems in Engineering manuscript format described at http://www .hindawi.com/journals/mpe/. Prospective authors should submit an electronic copy of their complete manuscript through the journal Manuscript Tracking System athttp://

mts.hindawi.com/according to the following timetable:

Manuscript Due December 1, 2008 First Round of Reviews March 1, 2009 Publication Date June 1, 2009

Guest Editors

Edson Denis Leonel,Departamento de Estatística, Matemática Aplicada e Computação, Instituto de Geociências e Ciências Exatas, Universidade Estadual Paulista, Avenida 24A, 1515 Bela Vista, 13506-700 Rio Claro, SP, Brazil ; [email protected]

Alexander Loskutov,Physics Faculty, Moscow State University, Vorob’evy Gory, Moscow 119992, Russia;

[email protected]

Hindawi Publishing Corporation http://www.hindawi.com