Henstock-Kurzweil Laplace Transform

Relaterede dokumenter
Hermite-Hadamard-Fejer Type Inequalities for s Convex Function in the Second Sense via Fractional Integrals

Fractional Wavelet Transform in Terms of Fractional Convolution

Exercise 6.14 Linearly independent vectors are also affinely independent.

A known plaintext attack on the ISAAC keystream generator

Pontryagin Approximations for Optimal Design of Elastic Structures

On Hermite-Hadamard-Fejér type inequalities for convex functions via fractional integrals

Large time behavior of solutions for a complex-valued quadratic heat equation

AALBORG UNIVERSITY. Spectral Analysis of a δ-interaction in One Dimension Defined as a Quadratic Form

SKRIFTLIG EKSAMEN I NUMERISK DYNAMIK Bygge- og Anlægskonstruktion, 7. semester Torsdag den 19. juni 2003 kl Alle hjælpemidler er tilladt

MONOTONE POSITIVE SOLUTIONS FOR p-laplacian EQUATIONS WITH SIGN CHANGING COEFFICIENTS AND MULTI-POINT BOUNDARY CONDITIONS

Critical exponent for semilinear wave equation with critical potential

Kurver og flader Aktivitet 15 Geodætiske kurver, Isometri, Mainardi-Codazzi, Teorema Egregium

Sign variation, the Grassmannian, and total positivity

Basic statistics for experimental medical researchers

Computing the constant in Friedrichs inequality

Chapter 6. Hydrogen Atom. 6.1 Schrödinger Equation. The Hamiltonian for a hydrogen atom is. Recall that. 1 r 2 sin 2 θ + 1. and.

INTERVAL VALUED FUZZY IDEALS OF GAMMA NEAR-RINGS

University of Copenhagen Faculty of Science Written Exam - 3. April Algebra 3

264.. Cox, Daio Jang (23) Grandell (1976). 1.1 (Ω, F, {F, [, ]}, P). N λ, λ F, 1 2 u R, λ d < a... E{e iu(n 2 N 1 ) F λ 2 } = e {(eiu 1) 2 1 λ d}, F λ

Nøgleord og begreber. l Hospitals regel 2. Test l Hospitals regel. Uegentlige integraler 2. Test uegentlige integraler. Sammenligning.

University of Copenhagen Faculty of Science Written Exam April Algebra 3

Eric Nordenstam 1 Benjamin Young 2. FPSAC 12, Nagoya, Japan

The LWR Model in Lagrangian coordinates

One-dimensional chemotaxis kinetic model

x t X t, t 1 τ =0 A t,τ (ξ t )x t τ = h t (ξ t )

Aktivering af Survey funktionalitet

Deepak Kumar Dubey, V. K. Jain RATE OF APPROXIMATION FOR INTEGRATED SZASZ-MIRAKYAN OPERATORS

Outline. Chapter 6: (cont d) Qijin Chen. November 21, 2013 NH = =6 CH = 15 4

SKRIFTLIG EKSAMEN I NUMERISK DYNAMIK Bygge- og Anlægskonstruktion, 8. semester Fredag den 30. juni 2005, kl Alle hjælpemidler er tilladt

= λ([ x, y)) + λ((y, x]) = ( y ( x)) + (x y) = 2(x y).

Nodal set of strongly competition systems with fractional Laplacian

what is this all about? Introduction three-phase diode bridge rectifier input voltages input voltages, waveforms normalization of voltages voltages?

yt () p0 cos( t) OPGAVE 1

Danish and English. Standard Field Analysis (Diderichsen) Standard Field Analysis (Diderichsen)

Multivariate Extremes and Dependence in Elliptical Distributions

CHAPTER 6 Derive differential Continuity, Momentum and Energy equations form Integral equations for control volumes.

Probabilistic properties of modular addition. Victoria Vysotskaya

A C 1,α partial regularity result for non-autonomous convex integrals with discontinuous coefficients

Positive solutions of Kirchhoff type elliptic equations involving a critical Sobolev exponent

University of Copenhagen Faculty of Science Written Exam - 8. April Algebra 3

DoodleBUGS (Hands-on)

Global attractor for the Navier Stokes equations with fractional deconvolution

since a p 1 1 (mod p). x = 0 1 ( 1) p 1 (p 1)! (mod p) (p 1)! 1 (mod p) for p odd and for p = 2, (2 1)! = 1! = 1 1 (mod 2).

Eksamen i Signalbehandling og matematik

Løsningsforslag til opgavesæt 5

Fremkomsten af mængdelæren. Stig Andur Pedersen

Gribskov kommune Tisvilde By, Tibirke

Uniform central limit theorems for kernel density estimators

Løsningsforslag til opgavesæt 5

Asymptotic analysis of a non-linear non-local integro-differential equation arising from bosonic quantum field dynamics

Besvarelser til Lineær Algebra Reeksamen Februar 2017

Generalized Probit Model in Design of Dose Finding Experiments. Yuehui Wu Valerii V. Fedorov RSU, GlaxoSmithKline, US

OPGAVE 1. Spørgsmål 1 (15%) Opstil bevægelsesligningen med tilhørende begyndelsesbetingelser for massen i punkt D.

MATEMATIK 11 Eksamensopgaver Juni 1995 Juni 2001, 3. fjerdedel

Integration m.h.t. mål med tæthed

Spørgsmål 1 (5%) Forklar med relevant argumentation, at den stationære temperaturfordeling i områdets indre er bestemt ved følgende randværdiproblem

Semi-smooth Newton method for Solving Unilateral Problems in Fictitious Domain Formulations

UGESEDDEL 9 LØSNINGER. Sydsæter Theorem 1. Sætning om implicitte funktioner for ligningen f(x, y) = 0.

Additive Property of Drazin Invertibility of Elements in a Ring

Hydrogen Burning in Stars-II

Some results for the weighted Drazin inverse of a modified matrix

BJB T e l: E-m a il: in n ie u w la n d.b e - W e b s it e : - Fa x :

Combined concave convex effects in anisotropic elliptic equations with variable exponent

Engelsk. Niveau D. De Merkantile Erhvervsuddannelser September Casebaseret eksamen. og

A B C D E Hjemmeværnmuseet's arkiv/depot Søgaard Distrikter - LMD. Reol/hylde Region/distrikt/m.m. Kasse nr. Indhold 2C3 Flyverhjemmeværne 1

On the complexity of drawing trees nicely: corrigendum

The Size of the Selberg Zeta-Function at Places Symmetric with Respect to the Line Re(s)= 1/2

DM549 Diskrete Metoder til Datalogi

8 Regulære flader i R 3

On the Fučik spectrum of non-local elliptic operators

Gamle eksamensopgaver (MASO)

Formale Systeme 2 Sommer 2012 Prof. P. H. Schmitt

Nonlinear Nonhomogeneous Dirichlet Equations Involving a Superlinear Nonlinearity

On the Relations Between Fuzzy Topologies and α Cut Topologies

Rotational Properties of Bose - Einstein Condensates

Funktionsundersøgelse. Rasmus Sylvester Bryder

Decomposition theorem for the cd-index of Gorenstein* posets

The X Factor. Målgruppe. Læringsmål. Introduktion til læreren klasse & ungdomsuddannelser Engelskundervisningen

Vores mange brugere på musskema.dk er rigtig gode til at komme med kvalificerede ønsker og behov.

Du kan efter ønske opfatte integralet som et Riemann-integral eller et Lebesgue-integral (idet de to er identiske på C([a, b], C) jf. Theorem 11.8.

PARALLELIZATION OF ATTILA SIMULATOR WITH OPENMP MIGUEL ÁNGEL MARTÍNEZ DEL AMOR MINIPROJECT OF TDT24 NTNU

Engelsk. Niveau C. De Merkantile Erhvervsuddannelser September Casebaseret eksamen. og

Symplectic and contact properties of the Mañé critical value of the universal cover

Eksamen i Mat F, april 2006

User Manual for LTC IGNOU

Frequency Dispersion: Dielectrics, Conductors, and Plasmas

The fundamental solution of compressible and incompressible fluid flow past a rotating obstacle

19.3. Second Order ODEs. Introduction. Prerequisites. Learning Outcomes

Erratum to: Closed-Range Composition Operators on A 2 and the Bloch Space

Satisability of Boolean Formulas

Adaptive Algorithms for Blind Separation of Dependent Sources. George V. Moustakides INRIA, Sigma 2

p-laplacian problems with nonlinearities interacting with the spectrum

On a multidimensional moving boundary problem governed by anomalous diffusion: analytical and numerical study

ATEX direktivet. Vedligeholdelse af ATEX certifikater mv. Steen Christensen

E-PAD Bluetooth hængelås E-PAD Bluetooth padlock E-PAD Bluetooth Vorhängeschloss

Our activities. Dry sales market. The assortment

Integration m.h.t. mål med tæthed

Strings and Sets: set complement, union, intersection, etc. set concatenation AB, power of set A n, A, A +

DM547 Diskret Matematik

MM537 Introduktion til Matematiske Metoder

Transkript:

Chper 5 Hensock-Kurzweil Lplce Trnsform 5.1 Inroducion The Lplce rnsform of funcion f : [, ) R, s C, is defined s he inegrl [15] e s f() d. (5.1) Mny uhors, in some previous decdes, showed heir ineres in sudying he Lplce rnsform in clssicl sense; by clssicl mening Lebesgue inegrl nd/or Riemnn inegrl on he rel line. In his chper, we focus on n inegrl rnsform h generlizes he clssicl Lplce rnsform. Here he generlizion is given by he replcemen, in he definiion of Lplce rnsform, i.e., in (5.1), of Lebesgue inegrl wih Hensock-Kurzweil (HK) inegrl. The Hensock-Kurzweil inegrl is defined in erms of Riemnn sums bu wih sligh modificion, ye i includes he Riemnn, improper Riemnn, nd Lebesgue inegrls s specil cses. This inegrl is equivlen o he Denjoy nd Perron inegrls. 117

The firs quesion rises is h wheher (or in which condiions) he Lplce rnsform when reed s HK inegrl mke sense. The sufficien condiion for he exisence of Lplce rnsform is h he funcion f is loclly inegrble on [, ), becuse he produc of loclly inegrble funcion nd bounded mesurble funcion is loclly inegrble. Since Hensock- Kurzweil inegrl llows nonbsolue convergence, i mkes n idel seing for he Lplce rnsform. I ws proved [9] h, on compc inervl, he produc of nonbsolue inegrble funcion nd bounded vriion funcion is sill nonbsolue inegrble nd his remins vlid in cse of n unbounded inervl [1]. We observed in (5.1) h he kernel funcion e s, s C, is no bounded vriion funcion on [, ) hence we cnno clim (s in he clssicl sense) h he HK Lplce rnsform exiss for ny HK inegrble funcion. The convoluion plys n imporn role in pure nd pplied mhemics, pproximion heory, differenil nd inegrl equions nd mny oher res. Lplce rnsform hs he propery h i cn inerc wih convoluions so we obin vrious resuls on convoluion. We give necessry nd sufficien condiions so h he convoluion operor s HK inegrl is coninuous. Finlly we solve he problem of inversion by using generlized differeniion. The resuls in his chper re ppered in [, 14]. 5. Exisenil Condiions In his secion, we ckle he problem of exisence of Lplce rnsform s HK inegrl. According o Zyed [16], he clssicl sufficien condiion for he exisence 118

of Lplce rnsform is h he funcion f is loclly inegrble on [, ), i.e., f L loc [, ). This is becuse he muliplier for he Lebesgue inegrble funcions is bounded mesurble funcion nd he funcion e s, s C, is bounded mesurble funcion. We know h he mulipliers for HK inegrble funcions re precisely he funcions of bounded vriion [1]. Noe h in (5.1), he funcion e s, s C, is no of bounded vriion on [, ) bu i is of bounded vriion on ny compc inervl J R. Hence we cn no consider (5.1) s HK inegrl direcly. So we cnno ge ny specific condiion for he exisence of Lplce rnsform s HK inegrl. However we do hve he following differen condiions. Theorem 5..1. Le f : [, ) R be ny coninuous funcion such h F (x) = x f, x <, is bounded on [, ). Then he Lplce rnsform L{f}(s), i.e., L{f}(s) = e s f() d, Re.s > exiss. Proof. Le < x <, s = σ + iω, σ, ω R. We hve (F () e s ) = s F () e s + f() e s. Since e s is of bounded vriion on ny compc inervl, sy [, b] nd F () is bounded on [, ). Then F () e s is HK inegrble on [, b]. Therefore s F () e s is inegrble over I = [, b]. Now (e s ) d = s e s d = = σ + ω e σ d σ + ω [e σ e σb ]. σ Wihou loss of generliy le us ke =. Therefore (e s ) σ + ω d = [1 e σb ] <. σ 119

Hence (e s ) is bsoluely inegrble over [, b]. Now by Hke s heorem [1], we cn wrie lim (e s ) σ + ω d = lim [1 e σb ] b b σ σ + ω =, σ > σ σ + ω =, Re.s >. σ This shows h (e s ) is bsoluely inegrble on [, ). Agin lim e s =, Re.s >. Hence he funcion e s : [, ) R is coninuous on [, ) wih lim e s = nd (e s ) is bsoluely inegrble over [, ) nd we hve ssumed h f : [, ) R is coninuous funcion such h F (x) = x f, x <, is bounded on [, ). Therefore by Dedekind s es [1], we cn sy h he inegrl e s f() d exiss for Re.s >. Remrk 5... The bove resul cn lso be proved by using Du-Bois Reymond s es [1] s follows: Tke φ() = e s, s = σ + iω, J = [, ). Then clerly φ() is differenible nd φ L 1 (J). Since F is bounded, we hve lim F () e s =, Re.S >. Hence F (x) = x f, x < is bounded on J, e s is differenible on J, (e s ) L 1 (J) nd F () e s s, Re.s >. Therefore by Du-Bois Reymond s es, f()e s HK(J). Th is, he Lplce rnsform of f(), e s f() d exiss, Re.s >. 1

Theorem 5..3. If he Lplce rnsform of f(), e s f() d exiss for Re.s >, hen he funcion f HK loc. Proof. Suppose he inegrl e s f() d exiss for Re.s >. Noe h he funcion e s, s C, s funcion of rel vrible, is no of bounded vriion on [, ) bu is of bounded vriion on ny compc inervl I = [, b]. By Hke s heorem [1], we cn wrie e s f() d = lim b e s f() d. Since he limis on R.H.S. exis, we cn sy h he HK inegrl e s f()d exiss, Re.s > for some compc inervl [, b]. Hence he funcion f() is HK inegrble necessrily on he compc inervl I = [, b], h is, f HK loc. Theorem 5..4. Le f HK([, )). Define F (x) = f() d. Then x L{f}(s) exiss if nd only if e s F () d exiss s C. Proof. Le T >. The funcion e s, s C, is of bounded vriion on compc inervl [, T ] nd he funcion f is HK inegrble on [, ). Hence e s f(), s C, is HK inegrble on [, T ] nd by inegring by prs, we ge T T e s f() d = e st F (T ) F () + s e s F () d. Since he funcion f is HK inegrble on [, ), F () is finie, sy A nd by [9.1(), [5]], F is coninuous funcion such h lim F (T ) =. T Therefore by Hke s heorem [1], T { T } lim e s f() d = lim e st F (T ) F () + s e s F () d. T T 11

Hence e s f() d = s e s F () d A. This shows h L{f}(s) exiss if nd only if e s F () d exiss s C, Re.s >. Theorem 5..5. (Uniqueness heorem) Le L{f}(s) = e s f() d nd L{g}(s) = e s g() d. If f() = g().e. on [, ), hen L{f}(s) = L{g}(s). 5.3 Bsic Properies The usul elemenry properies such s lineriy, dilion, modulion, rnslion re remins sme for he HK Lplce rnsform, he proofs of which re quie similr o h of clssicl ones. Also here re some oher resuls h re nlogous o he clssicl one wih some modificion in he hypohesis which we discuss below. Theorem 5.3.1. (HK Lplce rnsform of derivive) Le f : [, ) f() R be funcion which is in ACG δ (R) such h lim =. Then for Re.s > e s boh L{f}(s) nd L{f }(s) fil o exis or L{f }(s) = s L{f}(s) f(). Proof. Le T >. Consider he inegrl J = T e s f () d. This inegrl is vlid since he funcion e s, s C, is bounded vriion on ny compc inervl [, T ] nd he funcion f is HK inegrble on R. By inegring by prs, we ge By Hke s heorem [1], T lim T J = e st f(t ) f() + s T e s f() d. T e s f () d = lim T e st f(t ) f() + lim s e s f() d T 1

Therefore = s e s f () d = s Th is, L{f }(s) = s L{f}(s) f(). e s f() d f(). e s f() d f(). Remrk 5.3.. The bove resul cn be generlize by imposing he condiions on f s: For k =, 1,,..., (n 1), f (k) f re in ACG δ (R) nd lim (k) =, Re.s >. e s Then L{f (n) }(s) nd L{f}(s) fils o exis or L{f (n) }(s) = s n L{f}(s) s n 1 f() s n f ()... f (n 1) (). Theorem 5.3.3. (Differeniion of HK Lplce rnsform) Le f : [, ) R be funcion such h L{f}(s) exiss on compc inervl, sy J = [α, β], α, β R. Define g() = f() nd ssume h g HK(R + ). Then L{g}(s) exiss nd L{g}(s) = (L{f}(s)).e. Here s R. Proof. Suppose L{f}(s) = Define g() = f(). e s f() d exiss on [α, β], α, β R. We know he necessry nd sufficien condiion for differeniing under he HK inegrl is h for ll [, b] [α, β] [1]. Now e s f() ds d = = s= s= = e s f() ds d = = s= = e s f() d ds (5.) (e e b )f() d 13

= L{f}() L{f}(b) which exiss. Therefore he double HK inegrl = s= e s f() ds d exiss. Hence by [lemm 5(), [13]], he equliy in (5.) holds nd we cn differenie under he HK inegrl. By (5.), we hve s= = e s f() d ds = L{f}() L{f}(b) = (L{f}(s)) ds. Therefore e s f() d ds = s= = for ll [, b] [α, β]. Which implies h (L{f}(s)) ds e s f() d = (L{f}(s)).e. on [α, β]. Th is, L{ f()}(s) = (L{f}(s)).e. on [α, β]. Hence he derivive of he HK Lplce rnsform of f() exiss.e. on [α, β]. Remrk 5.3.4. Le G(s) = L{ f()}(s) nd H(s) = L{f()}(s). Then (H(s)) = G(s).e. on [α, β]. And, we know h [5], A funcion f : [, b] R is HK inegrble on [, b] if nd only if here exiss n ACG δ funcion F on [, b] such h F = f.e. on [, b]. Agin, by fundmenl heorem of clculus [1], (H(s)) is HK inegrble, i.e., every derivive is HK inegrble. Hence G(s) is lso HK inegrble. Consequenly, he funcion H(s) is ACG δ on [α, β]. Coninuing in his wy n-imes, we cn sy h if n f() is HK inegrble, hen ny order of derivive of L{f}(s) exiss.e. So L{f}(s) is nlyic.e. 14

Theorem 5.3.5. (Muliplicion by 1 ) Le f : [, ) R be funcion { } such h L{f}(s) exiss nd ssume h f() HK(R + ). Then L f() (s) exiss nd Here s R. { } f() L (s) = Proof. Assume h f() HK(R + ). s L{f}(u) du. Since he funcion e s is of bounded vriion on ny compc inervl I = [, b], we hve e s f() HK([, b]). Th is, e s f() d exiss. Wihou loss of generliy, le us ke = so h e s f() d exiss. By Hke s heorem [1], lim } Hence L (s) exiss. { f() Now consider he inegrls I 1 = s b e s f() d exiss. e u f() du d, I = I 1 = = Therefore I 1 exiss on R R. [e s ] f() d s f() e d which exiss. s e u f() d du. Also, noe h he funcion e s is of bounded vriion on ny compc inervl J R. Therefore R V J(e s ) d < M J, for some consn M J >. Hence by [lemm 5(), [13]], I exiss on R R nd I 1 = I on R R. So Th is, L s f() e d = = s { } f() (s) = L{f}(u) du. s s e u f() d du L{f}(u) du. 15

Theorem 5.3.6. (HK Lplce rnsform of inegrl) Le f : [, ) R { } be HK inegrble funcion such h L{f}(s) exiss. Then L f(u) du (s) { } exiss nd L f(u) du (s) = 1 L{f}(s). s Proof. Define F () = f(u) du, R+. Then F () = f().e. on R + [9.1(b), [5]]. Since f HK(R + ), he inegrl Th is, lim Hence lim F () is bounded on R +. Now for T >, consider T f(u) du exiss on R +. f(u) du exiss on R +. e s f() d = T e s F () d. This inegrl exiss since e s is of bounded vriion on [, T ] nd he funcion f() is HK inegrble on R +. By inegring by prs, we ge T T e s f() d = e st F (T ) + s e s F () d. Since lim F (T ) is bounded on R, se lim F (T ) = A. T T By Hke s heorem [1], we cn wrie T e s f() d = lim e s f() d T = lim T e st F (T ) + s lim = s T T e s F () d Re.s >. e s F () d 16

Therefore Th is, e s F () d = 1 s e s f() d, Re.s >. e s f(u) du d = 1 L{f}(s), Re.s >. s { } Hence L f(u) du = 1 L{f}(s), Re.s >. s 5.4 Convoluion In his secion we shll discuss some clssicl resuls of Lplce convoluion in he HK inegrl seing, clled HK-convoluion. Firs we give wo resuls bou he exisence of HK-convoluion. Theorem 5.4.1. Le f HK(R + ) nd g BV(R + ). Then we hve f g exiss on R +. Proof. By definiion [4], f g() = f( τ) g(τ) dτ which exiss s HK-inegrl by muliplier heorem [1]. By inegring by prs, we ge Hence f g exiss on R +. f g() A f { } inf g + V R + [,] g(τ). Theorem 5.4.. Le f HK loc nd g BV such h he suppor of g is in he compc inervl I = [, b]. Then } f g() A f [ b, ] {inf g + V [,b] g(τ). [,b] 17

Proof. Suppose supp(g) [, b]. Then we hve f g() = f( τ)g(τ) dτ. Inegre by prs, we ge f g() g(τ) b f( τ) dτ + f( τ) dg(τ) inf g(τ) b b τ [,b] f( τ) dτ + f( τ) dτ V [,b] g(τ) { } inf g(τ) + V [,b] g(τ) sup τ [,b] f(u) du. [ b, ] b Therefore } f g() A f [ b, ] {inf g + V [,b] g(τ). [,b] The following resul shows h he HK-convoluion is bounded. Theorem 5.4.3. Le f HK(R + ) nd g L 1 (R + ) BV(R + ). Then we hve A f g A f g 1. Proof. For < b, we hve Le f g() d = f( τ)g(τ) dτ d. I 1 = Consider f( τ)g(τ) dτ d nd I = I = f( τ)g(τ) d dτ. Since f HK(R + ), we hve f( τ) d exiss nd finie. Le b f(τ) dτ = M. 18 f( τ)g(τ) d dτ.

Therefore I M g(τ) dτ. Since g L 1 (R + ), we hve for some finie K, g(τ) dτ = K. So I MK nd hence I exiss on R + R +. Now s g L 1 (R + ), we hve g 1 K 1 nd lso V R + [,] g M 1. Therefore by [lemm 5, [13]], I 1 = I on R + R +. Hence f g() d sup b f(u) du g(τ) dτ [, b] A f [, b] g 1. Therefore sup [,b] Th is, A f g A f g 1. f g() d A f [, b] g 1. Now we give some bsic elemenry properies of HK-convoluion. Theorem 5.4.4. Suppose f HK(R + ) nd g BV(R + ). Then he following holds: I) f g() = g f() for ll R +. II) (λf) g = f (λg) = λ(f g) for some λ C. III) (f g) x () = f x g() = f g x (). IV) h (f + g)() = h f() + h g() for some h HK(R + ). Proof. The proof is similr o h of he clssicl cse. 19

The ssociive propery of HK-convoluion is given in he nex heorem. Theorem 5.4.5. If f HK(R + ), g BV(R + ) nd h L 1 (R + ), hen (f g) h() = f (g h)(), R +. Proof. Consider (f g) h() = τ By chnging he order of inegrion, we ge Now Consider nd Le (f g) h() = I 1 = I = = ξ= ξ= f(ξ) f(ξ) g(τ ξ) h( τ) dξ dτ. ξ f(ξ) g h( ξ) dξ = f (g h)(). τ=ξ ξ= J = ξ= g(u) h( u ξ) du dξ f(ξ) g(τ ξ) h( τ) dξ dτ f(ξ) g(τ ξ) h( τ) dτ dξ. τ=ξ τ=ξ g(τ ξ) h( τ) dτ. Inegre by prs, we ge J g( ξ) u h( τ) dτ + h( τ) dτ dg(τ ξ) τ=ξ τ=ξ τ=ξ inf g [ξ,] h( τ) dτ + h( τ) dτ V [ξ,] g(τ ξ) inf [ξ,] g τ=ξ τ=ξ h( τ) dτ + τ=ξ τ=ξ inf [ξ,] g h 1 + h 1 V [ξ,] g(τ ξ) 13 h( τ) dτ V [ξ,] g(τ ξ)

Therefore } h 1 {inf g + V [ξ,] g(τ ξ). [ξ,] I sup R + f(ξ) dξ { h 1 { inf A f h 1 {inf [ξ,] g + V [ξ,] g(τ ξ) }} g + V [ξ,] g(τ ξ) [ξ,] }. Hence I exiss on R + R +. And, by [lemm 5, [13]] we re hrough. Now we show h he HK-Lplce rnsform of convoluion of wo funcions is he produc of heir HK-Lplce rnsforms. Theorem 5.4.6. Le f 1 HK(R + ), f L 1 (R + ) BV(R + ). Then f g exiss nd if L {f 1 ()} nd L {f ()} exis s C, hen L {f 1 f ()} exiss s C nd L {f 1 f ()} (s) = L {f 1 ()} (s) L {f ()} (s). Proof. Consider L {f 1 f ()} (s) = = e s f 1 f () d By inerchnging he iered inegrls, we hve L {f 1 f ()} (s) = τ= Bu by hypohesis boh he inegrls = e sτ f 1 (τ)e s( τ) f ( τ) dτ d. e sτ f 1 (τ) e sτ f 1 (τ) dτ =τ e s( τ) f ( τ) d dτ e su f (u) du. e sτ f 1 (τ) dτ nd e sτ f (τ) dτ exis s C. Therefore L {f 1 f ()} exiss s C nd 131

L {f 1 f ()} (s) = L {f 1 ()} (s) L {f ()} (s). Now i remins o show he jusificion for inerchnging he iered inegrls. Since he funcion f L 1 (R + ), K I such h f 1 < K I, where I = [, b] R + is ny compc inervl. Clerly, V [,b] e s( τ) f ( τ) e σu V [ b, ] f (τ) + f (τ) e σr for some rels u nd r nd so V [ b, ] e sτ f (τ) dτ e σu V [ b, ] f (τ) dτ +e σr f (τ) dτ. Therefore V [ b, ] e sτ f (τ) dτ e σu V [ b, ] f (τ) dτ + K I e σr. Since f BV(R + ), here exiss consn M I such h Then V [ b, ] f (τ) dτ < M I. V [ b, ] e sτ f (τ) dτ e σu M I + K I e σr. Thus for ech compc inervl I = [, b] R +, he inegrl is finie nd f 1 K I. Hence he inegrl V [ b, ] e sτ f (τ) dτ τ= =τ e s f 1 (τ) f ( τ) d dτ exiss on R + R +. And, by [lemm 5, [13]], we cn inerchnge he iered inegrls. 13

Theorem 5.4.7. In ddiion o he hypohesis of he bove heorem if f 1 f () is in HK(R + ) nd f is coninuous, hen f 1 f () is coninuous wih respec o he Alexiewicz norm. Proof. Choose ny δ > such h < δ 1 (The cse for negive δ is nlogous). Define D(, δ) = f 1 f ( + δ) f 1 f () = I 1 + I. Where I 1 = f 1 (τ) [f ( + δ τ) f ( τ)] dτ, I = Since f BV(R + ), f is bounded on R +. +δ So here exiss M such h f (x) M, for ll x R +. Then +δ I M f 1 (τ) dτ. Bu f 1 HK(R + ) implies h lim δ Therefore I s δ. Now consider I 1 = +δ f 1 (τ) dτ. f 1 (τ) [f ( + δ τ) f ( τ)] dτ. f 1 (τ)f (+δ τ) dτ. By inegring by prs, we ge I 1 [f ( + δ τ) f ( τ)] f 1 (τ) dτ u + f 1 (τ) dτ d(f (u + δ τ) f (u τ)) f ( + δ τ) f ( τ) sup f 1 (τ) dτ R + u + sup f 1 (τ) dτ d(f (u + δ τ) f (u τ)) u R + 133

= f ( + δ τ) f ( τ) sup f 1 (τ) dτ R + u + sup f 1 (τ) dτ f (u + δ τ) f (u τ). u R + Therefore I 1 f ( + δ τ) f ( τ) sup f 1 (τ) dτ R + u + sup f 1 (τ) dτ f (u + δ τ) f (u τ). u R + Since f is coninuous on R +, for given ɛ > here exiss η > such h I 1 ɛ, δ < η. Therefore D(, δ) I 1 + I < ɛ, δ < η. Th is, for given ɛ > here exiss η > such h D(, δ) < ɛ, δ < η. Hence f 1 f is coninuous R +. Now, le α, β R +. Consider β α [f 1 f ( + δ) f 1 f ()] d = α+δ α f 1 f () d β+δ β f 1 f () d. Wrie F 1, (x) = x f 1 f () d. Then β sup [f 1 f ( + δ) f 1 f ()] d sup F 1, (α + δ) F 1, (α) α, β R + α R + α + sup β R + F 1, (β + δ) F 1, (β). Since F 1, is n indefinie HK inegrl of f 1 f on R +, i is coninuous on R + [5]. Therefore for given ɛ > here exiss η > such h Hence we hve F 1, (ξ + δ) F 1, (ξ) < ɛ, δ < η nd for ll ξ R+. sup α, β R + β α [f 1 f ( + δ) f 1 f ()] d < ɛ, δ < η. 134

Thus for given ɛ > here exiss η > such h A f 1 f (+δ) f 1 f () < ɛ, δ < η. This shows h f 1 f () is coninuous R + wih respec o he Alexiewicz norm. The following resul concerns wih necessry nd sufficien condiions on f n nd g n : [, ] R so h he HK-convoluion s n operor is coninuous. I involves eiher uniform boundedness or uniform convergence of he indefinie HK inegrl of f n. Theorem 5.4.8. Le {f n } be sequence of HK-inegrble funcions, f n : [, ] R, n such h f n = f s n for some HK-inegrble funcion f : [, ] R. Define F n (x) = x f n, F (x) = x f. Le {g n } be sequence of uniform bounded vriion funcions, g n : [, ] R, n, such h g n g poinwise on [, ], where g : [, ] R. Then f n g n () f g() s n if nd only if i) F n F uniformly on [, ] s n. ii) F n F poinwise on [, ], {F n } is uniformly bounded nd V (g n g). Proof. i) Suppose F n F uniformly on [, ] s n. By ssumpion nd muliplier heorem [1], f n g n () exiss R + nd f n g n () = f n (u) g n ( u) du nd f g() = Consider f n g n () f g() = I 1 + I, where I 1 = [f n (u) f(u)] g n ( u) du, I = f(u) g( u) du. [g n ( u) g( u)] f(u) du. 135

By inegring by prs, we ge Then I 1 = g n ( u) [f n (u) f(u)] du = g n ( u)[f n () F ()] I 1 Mx u F n(u) F (u) s n Hence I 1 s n. And by inegring by prs, we ge y [F n (u) F (u)] dg n. [f n (u) f(u)] dg n dg n + M F n () F () (Since F n F uniformly). I = {g n ( u) g( u)} f F dg n + F dg. By ssumpion he firs erm ends o nd since F is coninuous nd ech g n is of bounded vriion wih g n g s n, so we hve F dg n F dg. Hence I s n. Therefore f n g n () f g() s n. Now for he converse pr, suppose h eiher F n F on [, ] or F n F no uniformly on [, ]. Then here is sequence {x i } i Λ in [, ] on which F n F uniformly. So, by Bolzno-Weiersrss heorem [7], we cn find subsequence {y i } i Γ of {x i } i Λ, Γ Λ, such h F n (y i ) F (y i ) for ll i nd y i y s i, i.e., F n (y i ) F (y i ). Le us ssume wihou loss of generliy h < y n y. Le H be Heviside sep funcion. Define H(u y n ), g n ( u) = H(u y), n Γ oherwise 136

nd g( u) = H(u y). Then we hve, for n Γ, f n g n () = F n () F n (y n ) nd f g() = F () F (y). Since F n (y n ) F (y n ) s n, y n y s n nd F is coninuous, we hve F n (y n ) F (y). Bu F n () F () s n. Therefore f n g n () f g() s n which is conrdicion. Hence we mus hve F n F uniformly on [, ]. ii) Suppose F n F poinwise on [, ], F n is uniformly bounded nd V (g n g). Consider f n g n () f g() = I 1 + I, where I 1 = f n (u) [g n ( u) g( u)] du, I = By inegring by prs, we ge I 1 = [g n ( u) g( u)] f n = [g n ( u) g( u)] F n () Therefore y [f n (u) f(u)] g( u) du. f n d(g n g) F n (y) d(g n g). I 1 g n ( u) g( u) F n () + F n (y) V (g n g). Since {F n } is uniformly bounded, we hve F n M, n nd for some consn M. Therefore I 1 g n ( u) g( u) M + M V (g n g). By ssumpion g n g s n nd lso V (g n g). Hence I 1 s n. Agin by inegring by prs, we wrie I = g( u) {F n () F ()} 137 (F n F ) dg. (5.3)

By ssumpion F n () F (), so he firs erm ends o. And, since F n M, F n F nd g is of bounded vriion, by he domined convergence heorem for Riemnn-Sieljes inegrl [6], we hve F n dg F dg s n. Therefore by (5.3), I s n. Hence f n g n () f g() s n. Now for he converse pr suppose here is c (, b) such h F n (c) F (c) s n. Le g n ( x) = g( x) = H(x c). Then f n (u) g n ( u) du = F n () F n (c), f(u) g( u) du = F () F (c). Since F n () F () nd F n (c) F (c), we hve F n () F n (c) F () F (c) s n. Th is, f n g n () f g() s n which is conrdicion. Hence we mus hve F n F poinwise on [, ]. Now if {F n } is no uniformly bounded, hen here is sequence {x i } i Λ in [, ] on which F n. Therefore by Bolzno-Weiersrss heorem [7], we cn find subsequence {y i } i Γ of {x i } i Λ, Γ Λ, such h F n (y i ) 1 for ll n nd F n (y i ) nd y i y. Wihou loss of generliy le us ke y i y. Define g n ( u) = H(u y i). Fn(yi ) Then V (g n ) 1, g = nd V (g n g) = nd f n (u) g n ( u) du = 138 f n (u) H(u y i) Fn (y i ) du

f n (u) = y i Fn (y i ) du = F n() Fn (y i ) F n (y i ). Therefore nd f n (u)g n ( u) du s n f(u)g( u) du =. Hence f n g n f g s n which is conrdicion nd so we mus hve {F n } is uniformly bounded sequence. 5.5 Inversion Since in cse of Hensock-Kurzweil inegrl, he inegrion process nd differeniion process re inverse of ech oher. Here we shll esblish he inverse of Hensock-Kurzweil Lplce rnsform by using Pos s generlized differeniion mehod [8]. For his we shll need following lemms: Lemm 5.5.1. Le η be such h < η < b, nd h(x) C ( x + η), h () =, h () <, h(x) is nonincresing on (, b]. Then ( π e k h(x) dx e k h() k h () k. Proof. Since h(x) C ( x + η), h (x) is coninuous. Le ɛ > be such h < ɛ < h (). Choose δ >, δ < η such h h (x) h () < ɛ, x + δ. Th is, h () ɛ < h (x) < h () + ɛ, x + δ. (5.4) 139

Consider he inegrl We wrie I k = I k + I k, where I k = +δ I k = e k [h(x) h()] dx. e k [h(x) h()] dx, I k = e k [h(x) h()] dx. +δ Since h is nonincresing on (, b], we hve h(x) h( + δ) for ll x > + δ. Therefore nd Hence I k Now consider s k. I k e k [h(+δ) h()] dx +δ I k e k [h(+δ) h()] (b δ). I k = +δ By Tylor s series wih reminder, we hve Therefore h(x) h() h (ξ) So by (5.4), we cn wrie +δ Th is, I k = e k [h () ɛ] (x ) dx +δ +δ e k [h(x) h()] dx. (x ), ξ + δ. e k h (ξ) (x ) dx. +δ e k h (ξ) (x ) dx e k [h ()+ɛ] (x ) dx. Consider +δ +δ e k [h () ɛ] (x ) dx I k e k [h ()+ɛ] (x ) dx. (5.5) J = +δ e k [h () ɛ] (x ) dx. 14

Afer subsiuing x = u, we ge Therefore +δ Similrly, +δ J = = = δ e k [h () ɛ] u du 1 k (h () ɛ) k 1 (h () ɛ) 1 π k (h () ɛ) ( e k [h () ɛ] (x ) dx = ( e k [h ()+ɛ] (x ) dx = δ k (h () ɛ) Therefore (5.5) becomes ( ( π I k (h k () ɛ) Since ɛ > is rbirry, we hve ( π I k h k () Hence So Th is, ( I k π k h () ( e k [h(x) h()] dx e u du e u du s k s k. π k (h () ɛ) π k (h () + ɛ) ( π k h () ( e k h(x) dx e k h() π k h () π k (h () + ɛ) π. k h () s k. s k. s k.. s k. s k. 141

Lemm 5.5.. Le η be such h < η < b, nd h(x) C (b η x b), h (b) =, h (b) <, h(x) is nonincresing on [, b). Then ( e k h(x) dx e k h(b) π s k. k h (b) Proof. The proof is similr o h of previous lemm. Lemm 5.5.3. Le η be such h < η < b, nd h(x) C ( x + η), h () =, h () <, h(x) is nonincresing on (, b]. Suppose φ(x) HK([, b]). Then ( φ(x) e k h(x) dx φ() e k h() π k h () s k. Proof. Since h(x) C ( x + η), we hve h (x) is coninuous. Le ɛ > be such h < ɛ < h (). Choose δ >, δ < η such h h (x) h () < ɛ, x + δ. Th is, Consider he inegrl h () ɛ < h (x) < h () + ɛ, x + δ. I k = We show h I k s k. Le We clim h g k s k. [φ(x) φ()] e k [h(x) h()] dx. e k [h(x) h()], x [, b] g k (x) =, x =. Since h is nonincresing on (, b], we hve h(x) h() < for ll x (, b]. Therefore g k (x) = e k [h(x) h()] s k. Se g(x) = for ll x [, b]. Then g k (x) = e k [h(x) h()] = g(x) for ll x [, b], s k. 14

Since φ(x) HK([, b]), he inegrl [φ(x) φ()] dx exiss. Observe h he sequence {g k (x)} is of uniform bounded vriion on [, b] wih g k (x) s k. Then by [Cor.3., [11]], we hve [φ(x) φ()] e k [h(x) h()] dx [φ(x) φ()] dx s k s k. Therefore Th is, φ(x) e k [h(x) h()] dx = φ() Bu by lemm (5.5.1), we hve Therefore φ(x) e k h(x) dx = φ() ( π e k h(x) dx e k h() k h () e k [h(x) h()] dx. e k h(x) dx. ( π φ(x) e k h(x) dx φ() e k h() k h () s k. s k. Lemm 5.5.4. Le η be such h < η < b, nd h(x) C (b η x b), h (b) =, h (b) <, h(x) is nonincresing on [, b) nd suppose φ(x) HK([, b]). Then ( π φ(x) e k h(x) dx φ(b) e k h(b) k h (b) 143 s k.

Proof. The proof is similr o h of he previous lemm. Now we re redy o obin he inversion heorem for Hensock-Kurzweil Lplce rnsform. Theorem 5.5.5. (Inversion Theorem) If f() HK(R + ) nd if he HK- Lplce rnsform of f(), L{f}(s) exiss, hen lim k 1 k! ( ) k+1 k e ku u k f(u) du = f(). Proof. The inversion heorem follows immediely, if we cn prove he following: I) II) lim k lim k I) We hve he relion 1 k! 1 k! ( ) k+1 k e ku u k f(u) du = f() (5.6) ( ) k+1 k e ku u k f(u) du = f(). (5.7) ( π f(x) e k φ(x) dx f() e k φ() k φ () Tke = nd φ(x) = ln(x) x Therefore Th is, f(x) e k [ln(x) x ] dx f() e k [ln() 1] ( s k. for ll x [, ). Then we hve π k ( 1 ) ( f(x) x k e kx π dx f() k e k k s k. s k. f(x) x k e kx dx f() k+1 e k ( π k s k. 144

Therefore s k, we hve 1 k! ( ) k+1 k b f(x) x k e kx dx = 1 k! = ( k e ( ) k+1 k ( π f() k+1 e k k ) k k ( π k! f(). k The Sirling s formul is given by ( ) n n e πn n!, using his we cn wrie 1 k! ( ) k+1 k b f(x) x k e kx dx = k f() πk = f() ( π k s k. s k This is rue for every b R, b >. Therefore by Hke s heorem [1], we hve lim k 1 k! ( ) k+1 k f(x) x k e kx dx = f(). II) Agin we hve he relion ( π f(x) e k φ(x) dx f(b) e k φ(b) k φ (b) s k. Tke b = nd φ(x) = ln(x) x, for ll x (, ]. Then we ge lim k 1 k! By Hke s heorem [1], we hve lim k 1 k! ( ) k+1 k f(x) x k e kx dx = f(). ( ) k+1 k f(x) x k e kx dx = f(). Therefore from (5.6) nd (5.7), we cn wrie lim k 1 k! ( ) k+1 k f(x) x k e kx dx = f(). 145

5.6 Exmples Here we shll give some exmples of HK-Lplce rnsformble funcions whose clssicl Lplce rnsform do no exis. However we could no ble o find he exc form of he HK-Lplce rnsform of hese funcions. Exmple 5.6.1. Consider he funcion g() = e s sin ln(+), s C. Le f() = e s sin nd φ() = 1 ln(+). Observe h he funcion φ() is differenible on R + nd φ L 1 (R + ) nd if F () = e su sin u du, Re.s >, hen F () M for ll (, ), Re.s > for some consn M >. Now Then { e s (s sin cos ) F () φ() = s + 1 + 1 } s + 1 1 ln( + ). { e s (s sin cos ) lim F () φ() = lim + 1 } 1 s + 1 s + 1 ln( + ). Therefore lim F () φ() exiss. Hence by Du-Bois-Reymond s es [1], we hve e s sin ln(+) HK(R + ), Re.s >. Th is, e s sin ln( + ) d exiss, Re.s >. Hence he HK-Lplce rnsform of he funcion g() =. Exmple 5.6.. Consider he funcion f() = cos. We hve F () = cos u u du = cos u du. sin ln(+) exiss, Re.s > 146

Le φ() = e s, s C. Then φ () = s e s L 1 (R + ), Re.s >. Now Then F () φ() = e s cos u du. lim F () φ() = lim e s = lim e s lim =, Re.s >. cos u du cos u du Therefore by Du-Bois-Reymond s es [1], we hve f() φ() HK(R + ), Re.s >. Hence he HK-Lplce rnsform of he funcion f() = cos exiss for Re.s >. Exmple 5.6.3. Le φ() = e σ, R +, σ >. This funcion is coninuous on R + nd φ( + 1) = 1 ( + 1) e < 1 = φ() for ll σ(+1) R+ eσ 1 nd lso lim φ() = lim =, σ >. e σ Therefore φ() is decresing o zero s. Observe h he inegrl diverges. e σ d, σ > Now le n N be such h < (1 + 4n ) π 4ω. For R, we hve sin ω 1 if nd only if ω n=n [(1 + 4n) π 4, (3 + 4n) π 4 ]. 147

If n N, hen (3 + 4n) π < (1 + n) π. 4 (1+n)π e σ sin d = n (3+4j) π 4 e σ j=n (1+4j) π 4 n (3+4j) π 4 e σ j=n 1 = 1 (1+4j) π 4 n (3+4j) π 4 j=n n (1+4j) π 4 j=n 4 sin ω d 1 d 1 (3 + 4j) π 4 eσ(3+4j) π 4 (3 + 4j) π e s(3+4j) π 4 π n 1 (1 + j) π e. σ(1+j)π j=n π d On he oher hnd, we hve (1+n)π e σ Bu since e σ d = = = n π e σ n π e σ n π e σ n π e σ HK(R + ), we hve Therefore j=n e σjπ jπ =. So e σ sin ω / L 1 (R + ), σ >. Similrly, e σ cos ω / L 1 (R + ), σ >. Now le f() = sin ω, φ() = e σ, R. d + d + d + (1+n)π e σ d n π n (1+j)π e σ j=n jπ n (1+j)π e σjπ j=n jπ n d + e σjπ jπ π. j=n e σ d =. Observe h φ() is monoone funcion nd φ() s. And sin ωu du, ω for ll ω R+. Th is, F is bounded. Therefore by Chrier-Dirichle s es [1], we hve e σ sin ω HK(R + ). 148 jπ d d

Similrly, e σ cos ω HK(R + ). Hence e σ {cos ω i sin ω} HK(R + ). Th is, Thus e s e σ e iω d exiss, σ >. d exiss, Re.s > s HK inegrl, where s = σ + iω. 149

References [1] Brle R. G., A Modern Theory of Inegrion, Grd. Sud. Mh. 3, Amer. Mh. Soc. Providence, 1. [] Chudhry M. S., Tikre S. A., On Guge Lplce Trnsform, In. Journl of Mh. Anlysis, Vol. 5, No. 35 (11), 1733-174. [3] Erdelyi A., Tbles of Inegrl rnsforms, Vol. 1, McGrw Hill Book Compny, Inc. 1954. [4] Gsque C., P. Wiomski P., Fourier nlysis nd pplicions, Springer- Verlg, New York, Inc. 1999. [5] Gordon R. A., The Inegrls of Lebesgue, Denjoy, Perron, nd Hensock, Grd. Sud. Mh. 4, Amer. Mh. Soc. Providence, 1994. [6] McLeod R. M., The Generlized Riemnn Inegrl, Crus Mh. Monogrphs, No., The Mhemicl Associion of Americ, Wshingon DC, 198. [7] Nnson I. P., Theory of funcions of rel vrible, Vol. 1, Frederick Unger Publishing Co. New York, 1964. [8] Pos E. L., Generlized differeniion, Trns. Amer. Mh. Soc. Vol. 3 (193), 73-781. 15

[9] Sks S., Theory of he Inegrl, Dover Publicions, New York, 1964. [1] Swrz C., Inroducion o Guge Inegrls, World Scienific Publishing Co. Singpore, 1. [11] Tlvil E., Limis nd Hensock inegrls of producs, Rel Anl. Exchnge 5 (1999-), 97-918. [1] Tlvil E., Necessry nd sufficien condiions for differeniing under he inegrl sign, Amer. Mh. Monhly, 18 (1), 544-548. [13] Tlvil E., Hensock-Kurzweil Fourier rnsforms, Ill. Jour. Mh. Vol. 46, No. 4(), 17-16. [14] Tikre S. A., Chudhry M. S., On some resuls of HK-convoluion, In. Elecron. J. Pure Appl. Mh. Vol., No. (1), 93-1. [15] Widder D. V., The Lplce rnsform, Princeon Universiy Press, Princeon, 1946. [16] Zyed A. I., The Hndbook of Funcion nd Generlized Funcion rnsformions, CRC Press, Inc. 1996. 151