mathematics On the Generalized Riesz Derivative ( ) s u(x) = C n,s P.V. Article

Similar documents
Lecture 13. Lecturer: Jonathan Kelner Scribe: Jonathan Pines (2009)

SAMPLE QUESTIONS FOR FINAL EXAM. (1) (2) (3) (4) Find the following using the definition of the Riemann integral: (2x + 1)dx

Theorems About Power Series

Research Article Sign Data Derivative Recovery

A probabilistic proof of a binomial identity

Our aim is to show that under reasonable assumptions a given 2π-periodic function f can be represented as convergent series

Sequences and Series

Infinite Sequences and Series

Class Meeting # 16: The Fourier Transform on R n

Convexity, Inequalities, and Norms

In nite Sequences. Dr. Philippe B. Laval Kennesaw State University. October 9, 2008

Annuities Under Random Rates of Interest II By Abraham Zaks. Technion I.I.T. Haifa ISRAEL and Haifa University Haifa ISRAEL.

Lecture 5: Span, linear independence, bases, and dimension

A note on the boundary behavior for a modified Green function in the upper-half space

Section 11.3: The Integral Test

Lecture 4: Cauchy sequences, Bolzano-Weierstrass, and the Squeeze theorem

Properties of MLE: consistency, asymptotic normality. Fisher information.

An Efficient Polynomial Approximation of the Normal Distribution Function & Its Inverse Function

SECTION 1.5 : SUMMATION NOTATION + WORK WITH SEQUENCES

Incremental calculation of weighted mean and variance

A Recursive Formula for Moments of a Binomial Distribution

INFINITE SERIES KEITH CONRAD

4.3. The Integral and Comparison Tests

Department of Computer Science, University of Otago

WHEN IS THE (CO)SINE OF A RATIONAL ANGLE EQUAL TO A RATIONAL NUMBER?

THE HEIGHT OF q-binary SEARCH TREES

Cooley-Tukey. Tukey FFT Algorithms. FFT Algorithms. Cooley

3. Greatest Common Divisor - Least Common Multiple

THIN SEQUENCES AND THE GRAM MATRIX PAMELA GORKIN, JOHN E. MCCARTHY, SANDRA POTT, AND BRETT D. WICK

Basic Elements of Arithmetic Sequences and Series

THE ABRACADABRA PROBLEM

Chapter 7 Methods of Finding Estimators

A Faster Clause-Shortening Algorithm for SAT with No Restriction on Clause Length

Asymptotic Growth of Functions

CME 302: NUMERICAL LINEAR ALGEBRA FALL 2005/06 LECTURE 8

Modified Line Search Method for Global Optimization

where: T = number of years of cash flow in investment's life n = the year in which the cash flow X n i = IRR = the internal rate of return

Vladimir N. Burkov, Dmitri A. Novikov MODELS AND METHODS OF MULTIPROJECTS MANAGEMENT

Chapter 5: Inner Product Spaces

Example 2 Find the square root of 0. The only square root of 0 is 0 (since 0 is not positive or negative, so those choices don t exist here).

Trigonometric Form of a Complex Number. The Complex Plane. axis. ( 2, 1) or 2 i FIGURE The absolute value of the complex number z a bi is

Lecture 4: Cheeger s Inequality

On Formula to Compute Primes. and the n th Prime


A sharp Trudinger-Moser type inequality for unbounded domains in R n

Partial Di erential Equations

Institute of Actuaries of India Subject CT1 Financial Mathematics

Entropy of bi-capacities

Irreducible polynomials with consecutive zero coefficients

Chapter 7 - Sampling Distributions. 1 Introduction. What is statistics? It consist of three major areas:

The analysis of the Cournot oligopoly model considering the subjective motive in the strategy selection

PROCEEDINGS OF THE YEREVAN STATE UNIVERSITY AN ALTERNATIVE MODEL FOR BONUS-MALUS SYSTEM

University of California, Los Angeles Department of Statistics. Distributions related to the normal distribution

Overview of some probability distributions.

Metric, Normed, and Topological Spaces

hp calculators HP 12C Statistics - average and standard deviation Average and standard deviation concepts HP12C average and standard deviation

AP Calculus AB 2006 Scoring Guidelines Form B

Chapter 6: Variance, the law of large numbers and the Monte-Carlo method

Perfect Packing Theorems and the Average-Case Behavior of Optimal and Online Bin Packing

Permutations, the Parity Theorem, and Determinants

Discrete Mathematics and Probability Theory Spring 2014 Anant Sahai Note 13

I. Chi-squared Distributions

GCE Further Mathematics (6360) Further Pure Unit 2 (MFP2) Textbook. Version: 1.4

Normal Distribution.

ON AN INTEGRAL OPERATOR WHICH PRESERVE THE UNIVALENCE

CS103A Handout 23 Winter 2002 February 22, 2002 Solving Recurrence Relations

Confidence Intervals for One Mean

A Note on Sums of Greatest (Least) Prime Factors

Building Blocks Problem Related to Harmonic Series

Soving Recurrence Relations

ON THE DENSE TRAJECTORY OF LASOTA EQUATION

Factors of sums of powers of binomial coefficients

Lecture 3. denote the orthogonal complement of S k. Then. 1 x S k. n. 2 x T Ax = ( ) λ x. with x = 1, we have. i = λ k x 2 = λ k.

SEQUENCES AND SERIES

NATIONAL SENIOR CERTIFICATE GRADE 12

Output Analysis (2, Chapters 10 &11 Law)

FOUNDATIONS OF MATHEMATICS AND PRE-CALCULUS GRADE 10

Analysis Notes (only a draft, and the first one!)

.04. This means $1000 is multiplied by 1.02 five times, once for each of the remaining sixmonth

Chapter 5 Unit 1. IET 350 Engineering Economics. Learning Objectives Chapter 5. Learning Objectives Unit 1. Annual Amount and Gradient Functions

Ekkehart Schlicht: Economic Surplus and Derived Demand

Solutions to Selected Problems In: Pattern Classification by Duda, Hart, Stork

Taking DCOP to the Real World: Efficient Complete Solutions for Distributed Multi-Event Scheduling

FIBONACCI NUMBERS: AN APPLICATION OF LINEAR ALGEBRA. 1. Powers of a matrix

BENEFIT-COST ANALYSIS Financial and Economic Appraisal using Spreadsheets

Subject CT5 Contingencies Core Technical Syllabus

CHAPTER 3 THE TIME VALUE OF MONEY

Factoring x n 1: cyclotomic and Aurifeuillian polynomials Paul Garrett <garrett@math.umn.edu>

3 Energy Non-Flow Energy Equation (NFEE) Internal Energy. MECH 225 Engineering Science 2

Problem Solving with Mathematical Software Packages 1

1. MATHEMATICAL INDUCTION

How To Solve The Phemean Problem Of Polar And Polar Coordiates

Degree of Approximation of Continuous Functions by (E, q) (C, δ) Means

Data Analysis and Statistical Behaviors of Stock Market Fluctuations

S. Tanny MAT 344 Spring be the minimum number of moves required.

A Combined Continuous/Binary Genetic Algorithm for Microstrip Antenna Design

Application and research of fuzzy clustering analysis algorithm under micro-lecture English teaching mode

MARTINGALES AND A BASIC APPLICATION

a 4 = = Which of the following sequences converge to zero? n 2 (a) n 2 (b) 2 n x 2 x = lim x n = lim x

BINOMIAL EXPANSIONS In this section. Some Examples. Obtaining the Coefficients

Transcription:

mathematics Article O the Geeralized Riesz Derivative Chekua Li * ad Joshua Beaudi Departmet of Mathematics ad Computer Sciece, Brado Uiversity, Brado, MB R7A 6A9, Caada; beaudijd3@bradou.ca * Correspodece: lic@bradou.ca Received: 3 Jue ; Accepted: July ; Published: 3 July Abstract: The goal of this paper is to costruct a itegral represetatio for the geeralized Riesz derivative RZ Dx s ux) for k < s < k + with k =,,, which is proved to be a oe-to-oe ad liearly cotiuous mappig from the ormed space W k+ R) to the Baach space CR). I additio, we show that RZ Dx s ux) is cotiuous at the ed poits ad well defied for s = + k. Furthermore, we exted the geeralized Riesz derivative RZ Dx s ux) to the space C k R ), where k is a -tuple of oegative itegers, based o the ormalizatio of distributio ad surface itegrals over the uit sphere. Fially, several examples are preseted to demostrate computatios for obtaiig the geeralized Riesz derivatives. Keywords: Riesz derivative; fractioal Laplacia; ormalizatio; distributio; Gamma fuctio. Itroductio Durig the past few decades, fractioal calculus 4 has bee a emerget tool which uses fractioal differetial ad itegral equatios to develop more sophisticated mathematical models that ca accurately describe complex systems. Fractioal powers of the Laplacia operator arise aturally i the stu of partial differetial equatios related to aomalous diffusio, where the fractioal operator plays a similar role to that of the iteger-order Laplacia for ordiary diffusio 5,6. By replacig Browia motio of particles with Lévy flights 7, oe obtais a fractioal diffusio equatio i terms of the fractioal Laplacia operator 8 of order s, ) via the Cauchy pricipal value P.V. for short) itegral 9, give as ) s ux) = C,s P.V. where = / x + + / x, ad the costat C,s is give by ) cos y C,s = R y +s = π / ux) uζ) dζ, ) R x ζ +s +s s Γ ) s. ) Γ s) Let x = x, x,, x ) R. For a give -tuple α = α, α,, α ) of oegative itegers or called a multi-idex), we defie α = α + α + + α, α! = α!α! α! x α = x α xα xα, α u = α α α u = x α xα α u. xα The Schwartz space SR ) space of rapidly decreasig fuctios o R ) is the fuctio space defied as SR ) = { ux) C R ) : ux) α,k C α,k cost) α, k N }, Mathematics, 8, 89; doi:.339/math8789 www.mdpi.com/joural/mathematics

Mathematics, 8, 89 of where N = {} N is the set of oegative itegers ad ux) α,k = sup x α k ux). x R Let r = x + x + + x. The fuctio space C k R ) is defied i Referece as follows. C k R ) = { ux) is bouded ad k ux) is cotiuous o R : M k cost) >, such that k ux) M } k as r where k = k, k,, k ) is a -tuple of oegative itegers. Applyig the ormalizatio i distributio theory, Pizzetti s formula, ad surface itegrals o R, Li very recetly exteded the fractioal Laplacia ) s ux) over the space C k R ) which cotais SR ) as a proper subspace) for all s > ad s =,,, ad obtaied Theorem below. Theorem. Let i =,, ad i < s < i +. The the geeralized fractioal Laplacia ) s is ormalized over the space C k R ) as ) s ux) = C,s r s Sr) r Ω ux) r i Ω i ux) i i! + ) + i ) dr, 3) where Ω = π / /Γ ) is the area of the uit sphere Ω R, k = k, k,, k ) with k + + k = i +, ad Sr) = ux + rσ) ux) + ux rσ)dσ. I particular for =, we have the followig. Ω Theorem. Let k < s < k + ad k =,,,. The the fractioal Laplacia operator ) s is ormalized over C k+ R) as ) s ux) = C,s y s Sy) u ) x)y yk k)! uk) x), 4) r where Sy) = ux + y) ux) + ux y). Defiitio. For a sufficietly ice fuctio ux) defied o R the left- ad right- sided Riema-Liouville derivatives of order α, with m < α < m N, give by RLD α,xux) = Γm α) d m x dx m ut) dt, t x) α m+ ad respectively. RLD α x,ux) = )m Γm α) d m dx m x ut) dt t x) α m+

Mathematics, 8, 89 3 of From itegratio by parts we have α m RL D,xux) α = u m) x), ad α m ) + RL D,xux) α = u m ) x). Defiitio. The α-order Riesz derivative of a fuctio ux) x R) is defied as RZD α xux) = Ψ α RLD α,x + RL D α x,) ux) where for α =, 3,. Ψ α = cos απ I geeral, the followig defiitio regardig the Riesz derivative o R ca be give. Defiitio 3. The Riesz fractioal derivative is defied for suitably smooth fuctio ux) x R ) i arbitrary dimesios by, RZDxux) α = l yu)x) d,l α) R y +α, < α < l where l is a arbitrary iteger bigger tha α, ad l yu)x) deotes either the cetred differece or o-cetred differeces l yu)x) = l yu)x) = l k= ) k ) l ux + l/ k)y), k l k= ) k ) l ux ky). k The d,l α) are ormalizig costats which are idepedet of the choice of l > α, ad are aalytic fuctios with respect to the parameter α by d,l α) = α π +/ Γ + α ) ) A lα) + α πα ), Γ si ad l k= ) k ) l k α, i the case of o-cetred differece, k A l α) = l k= ) k ) ) l l α k k, i the case of cetred differece, for a eve umber l > α. It is well kow that the Riesz derivative plays a importat role i aomalous diffusio 3 5 ad space of fractioal quatum mechaics. For example, the Riesz derivative satisfies the fractioal diffusio equatio, which has lots of physical applicatios 3: P L x, t; α) t σ α RZ D α x P L x, t; α) =,

Mathematics, 8, 89 4 of where P L x, t; α) is the α-stable Lévy distributio ad α, < α, is called the Lévy idex. There are also may studies, icludig umerical aalysis 6 9, scietific computig ad Fourier trasform methods,, o differetial equatios ivolvig the Riesz derivative with applicatios i several fields, icludig mathematical physics ad egieerig. It is widely cosidered that the Riesz derivative is equivalet to the fractioal Laplacia i arbitrary dimesios 4. Cai ad Li 5 showed that for s, ) ) s ux) = RZ Dx s ux), ux) SR) ad s = /, ) s ux) = RZ Dx s ux), ux) SR ) with >. Furthermore, o page 5 ad 6 i the same referece they stated i) for the case with α = 3, 5,, the Riesz derivative of the give fuctio ux) x R) ca be defied i the form ) RZDxux) α = α αγ +α ux + y) ux) + ux y) π Γ α ) y +α, which is suitable for positive values of α =, 4, 6,. ii) For k =,,. The, α 4k+ RZ Dxux) α = 4k 4k + )Γk + ) ) π / Γ 4k ad for k =,,,, α 4k+3 RZ Dxux) α = 4k+ 4k + 3)Γk + ) ) π / Γ +4k ux + y) ux) + ux y) y 4k+, ux + y) ux) + ux y) y 4k+4. We would like to recosider cases i) ad ii) i this paper as the itegrals o the right-had side do ot exist eve for a sufficietly good fuctio ux) SR). Ideed, by Taylor s expasio ux + y) ux) + ux y) = u x + θy) + u x θy) y u x)y as y +! where θ, ). This clearly makes all the itegrals o the right-had side diverget ear the origi. As outlied i the abstract, we establish a itegral represetatio for the geeralized Riesz derivative RZ Dx s ux) for k < s < k + with k =,,, as a liearly cotiuous mappig from the ormed space W k+ R) to the Baach space CR). The we stu the geeralized Riesz derivative i arbitrary dimesios ad further show that RZ Dx s ux) is cotiuous at the ed poits based o the ormalizatio of distributio ad the surface itegrals. I particular, the derivative RZ Dx s ux) is well defied for all s = k +, which exteds Defiitio.. The Geeralized Riesz Derivative o R Let CR) be the space of cotiuous fuctios o R give as CR) = {ux) : ux) is cotiuous o R ad u < } where u = sup ux). x R

Mathematics, 8, 89 5 of Clearly, CR) is a Baach space. The followig space will play a importat role i defiig the geeralized Riesz derivative o R. Let k =,,. We defie the ormed space W k R) as where Clearly, ux) = W k R) = u k = max { } ux) : u k) x) is cotiuous o R ad u k < { sup x R xux), sup x R x x + W kr) but ux) = } xu x), sup x + )u k) x). x x + x R / SR), ad SR) W k R) C k R) CR) for all k =,,. We are rea to prove the followig theorem which establishes a iitial equivalece betwee the Riesz derivative ad the fractioal Laplacia o the space W R). Theorem 3. Let u W R). The both RZ D s x ux) ad ) s ux) exist ad for < s <. RZD s x ux) = ) s ux) = C,s ux + y) ux) + ux y) y +s Proof of Theorem 3. Makig the variable chage z = x ζ, we derive from Equatio ) that for = ) ) s ux) ux z) ux) = C,s P.V. R z +s dz. Settig w = z o the right-had side of the above equality, we come to P.V. R ux) ux z) z +s dz = P.V. R ux) ux + w) w +s dw. Therefore, P.V. = P.V. R = P.V. ux) ux z) dz R z +s ux) ux z) z +s dz + P.V. R ux + y) ux) + ux y) y +s R ux) ux + w) dw w +s after relabelig y = z ad y = w. This implies that ) s ux) = C,s P.V. R ux + y) ux) + ux y) y +s. Note that the above itegral is well defied for ux) W R). Ideed, a secod order Taylor expasio ifers ux + y) ux) + ux y) y +s sup y R u y) y s.

Mathematics, 8, 89 6 of Hece, it is absolutely itegrable ear the origi. Furthermore, ux) W R) implies that there exists a costat C > such that y + )u y) C as y. as This idicates that the itegral is absolutely itegrable at ifiity. I summary, ) s ux) = C,s R = C,s ux + y) ux) + ux y) y +s ux + y) ux) + ux y) y +s, ux + y) ux) + ux y) y +s is a eve fuctio with respect to y. Assume < s < /. Itegratio by parts yields ux + y) ux) + ux y) y +s = ux + y) ux) + ux y) s y s + y= s = d ux + y) s dx y s d ux y) s dx y s by applyig the facts that all four itegrals ux + y) y s, ux y) y s, ad u x + y) y s are uiformly coverget with respect to x usig the coditios are bouded. Sice we come to From the formula 6 d dx d dx sup x R ) s ux) = C,sΓ s) s C,s = xux) ad sup xu x) x R u x + y) u x y) y s u x y) y s ux + y) y s = Γ s) RL Dx,ux), s ad ux y) y s = Γ s) RL D,xux) s ) RLDx,ux) s + RL D,xux) s. cos y y +s = Γ s) cosπs), s we have ) ) s ux) = Ψ s RLD,x s + RL Dx, s ux) = RZ Dx s ux).

Mathematics, 8, 89 7 of Fially we assume / < s <. Applyig we deduce that ) s ux) = = = d dx d dx ux + y) y s = Γ s) RL Dx,ux) s ad ux y) y s = Γ s) RL D,xux), s C,s ss ) d I particular for s = /, we have u x + y) + u x y) y s C,s ux + y) ss ) dx y s + d dx C,s ss ) Γ s) RLD,x s + RL Dx, s = C,s s Γ s) RLD s,x + RL D s x, RZD xux) = ) / ux) = C,/ = π which is well defied ad exteds Defiitio to the value α =. Remark. ux) ux y) y s ux) = RZ D s x ux). ux + y) ux) + ux y) y ux + y) ux) + ux y) y, a) Usig the formula Γz)Γ z + ) = π z Γz) for z =,,,, we have for u W R) that RZDx s ux) = ) s si πs ux) = Γ + s) π ux + y) ux) + ux y) y +s b) for < s <. This expressio has symbolically appeared i several existig literatures, such as Refereces 3,,4, for a suitable smooth fuctio ux). Cai ad Li preseted Theorem 3 i Referece 5 uder the coditios that ux) SR) which is a proper subspace of W R), ad s, ) with s = /. I order to stu the geeralized Riesz derivative, we briefly itroduce the followig basic cocepts i distributio ad the ormalizatio of x+ λ. Let DR) be the Schwartz space 7 of ifiitely differetiable fuctios or so-called the Schwartz space of testig fuctios) with compact support i R, ad D R) be the space of distributios liearly cotiuous fuctioals) defied o DR). Furthermore, we shall defie a sequece φ x), φ x),, φ m x), which coverges to zero i DR) if all these fuctios vaish outside a certai fixed ad bouded smooth set i R ad coverge uiformly to zero i the usual sese) together with their derivatives of ay order. We further assume that D R + ) is the subspace of D R) with support cotaied i R +. The fuctioal δ is defied as δ, φ) = φ), where φ DR). Clearly, δ is a liear ad cotiuous fuctioal o DR), ad hece δ D R).

Mathematics, 8, 89 8 of Let f D R). The the distributioal derivative f o D R) is defied as: f, φ) = f, φ ) for φ DR). I particular, δ m) x), φx)) = ) m φ m) ), where m is a oegative iteger. The distributio x+ λ o DR) is ormalized i Referece 7 as: x λ +, φx)) = x λ φx) φ) xφ ) xm m )! φm ) ) dx, 5) where m < λ < m m N) ad φ DR). Let τx) be a ifiitely differetiable fuctio o, +) R satisfyig the followig coditios: i) τx), ii) τx) = if x /, iii) τx) = if x. Let r = x + x + + x. We costruct the sequece I m r) for m =,, as: if r m, I m r) = m m ) τ + m +m r mm+ + m +m if r > m. Clearly, I m r) is ifiitely differetiable with respect to x, x,, x ad r, ad I m r) = if r m + m m, as m m + m +m m + m m ) mm+ =. + m+m Furthermore, I m r). Applyig Equatio 5) ad the idetity sequece I m r) for m =,,, Li established Theorems ad outlied i the itroductio. Based o Theorems ad 3, the geeralized Riesz derivative o R is well defied, for k < s < k + with k =,,,, as RZDx s ux) = C,s y s Sy) u ) x)y yk k)! uk) x), where ux) W k+ R), ad Sy) = ux + y) ux) + ux y). The followig theorem is to costruct a relatioship betwee the ormed space W k+ R) ad the Baach space CR) by the geeralized Riesz derivative. Theorem 4. Let k < s < k + with k =,,,. The the geeralized Riesz derivative RZ Dx s RZDx s ux) = C,s y s Sy) u ) x)y yk k)! uk) x) is a oe-to-oe ad liearly cotiuous mappig from W k+ R) to CR). give by

Mathematics, 8, 89 9 of Proof of Theorem 4. From the above itegral expressio, the geeralized Riesz derivative RZ Dx s is a liear mappig o the space W k+ R). Let u m x) W k+ R) ad u m x) i W k+ R). It follows from Taylor s expasio that S m y) = u m x + y) u m x) + u m x y) = u ) m x)y + + yk k)! uk) m x) + yk+ k + )! where θ, ). Clearly, u k+) m x + θy) + u m k+) ) x θy), RZD s x u m x) = = C,s k + )! C,s k + )! y s+k+ u k+) m y s+k y u k+) m x + θy) + u k+) ) m x θy) x + θy) + y u k+) ) m x θy). Therefore, RZ Dx s u m x) C,s k + )!k + s) sup u k+) m y) y R C +,s k + )!s k) sup y u k+) m y), y R u k+) which coverges to zero, as u m x) k+ implies that both sup y R m y) ad y sup y R u k+) m y) go to zero as m. It remais to show that RZ Dx s is oe-to-oe from W k+ R) to CR). Assume u x), u x) W k+ R) such that RZDx s u x) = RZ Dx s u x). This ifers that = Usig the formula 8 we arrive at = y s y s S y) u ) x)y yk k)! uk) x) S y) u ) x)y yk k)! uk) x) s k+) y s Γ s) = δk+) y), δ k+) y) S y) u ) x)y yk k)! uk) x) δ k+) y) S y) u ) x)y yk k)! uk) x).. Hece, S k+) ) = S k+) ),

Mathematics, 8, 89 of by otig that Evidetly, δ k+) y) which further claims that S y) u ) x)y yk k)! uk) x) S k+) ) = u k+) x) = u k+) x) = S k+) ), u x) = u x) + P k+ x), = S k+) ). where P k+ x) is a polyomial of degree k + i the space W k+ R), which must be zero due to the coditio sup xp k+ x) <. x R Remark. At this momet, we are uable to describe a subspace say C s R)) of CR) such that the geeralized Riesz derivative RZ Dx s is bijective ad liearly cotiuous mappig from W k+ R) to C s R). This further stu is of iterest sice we ca defie a iverse operatio of the Riesz derivative o C s R) if it exists. I additio, we have the followig theorem regardig the its at the ed poits for the geeralized Riesz derivative RZ D s x ux) over the space W k+ R). Theorem 5. Let ux) W k+ R) ad k < s < k + with k =,,,. The, s k+) RZ Dx s ux) = ) k u k+) x), ad s k + RZ Dx s ux) = ) k+ u k) x) i the space CR). I particular, for all k =,,. s k RZ Dx s ux) = ) k+ u k) x) Proof of Theorem 5. Let k < s < k + with k =,,,. The, RZ D s s k+) x ux) ) k u k+) x) = sup s k+) x R C,s y s Sy) u ) x)y yk k)! uk) x) ) k u k+) x). Usig s k+) Γ Γ s) Γ s) = k! ) k+, ad k + )! ) + k + k + )! = π, k+ k + )!

Mathematics, 8, 89 of we derive that ) C,sΓ s) = π / k+) k + )Γ s k+) + k + k! ) k+ k + )! = )k. Furthermore, the itegral coverges uiformly with respect to s. Hece, y s Sy) u ) x)y yk Γ s) k)! uk) x) s k+) = δ k+) y) y s Γ s) = S k+) ) = u k+) x). Sy) u ) x)y yk k)! uk) x) Sy) u ) x)y yk k)! uk) x) I summary, we get RZ D s s k+) x ux) ) k u k+) x) =, which implies that s k+) RZ Dx s ux) = ) k u k+) x) i the space CR). O the other had, RZ D s s k + x ux) ) k+ u k) x) = sup s k + x R C,s y s Sy) u ) x)y yk k)! uk) x) ) k+ u k) x). Usig we derive that Γ s) k )! ) = k, ad s k + Γ s) k)! ) Γ + k = k)! π k k! ) k )! C,sΓ s) = π / k kγ s k + + k ) k k)! = )k+.

Mathematics, 8, 89 of Thus, from y s + Γ s) = δs) y), s y s u ) x)y + + yk k)! uk) x) = Sy) u ) x)y + + yk k)! uk) x) δ s) y) s k + = s k + Ss) ) = S k) ) = u k) x) for s > k, ad it follows that Therefore, i the space CR). RZ D s s k + x ux) ) k+ u k) x) =. s k + RZ Dx s ux) = ) k+ u k) x) Remark 3. a) From Theorem 5, we have s k+ RZ Dx s ux) = s k+) + RZ Dx s ux) = s k+) RZ Dx s ux) = u 4k+) x) for all k =,,,, ad for all k =,,. b) Clearly for k =,,, s k RZ Dx s ux) = s k) + RZ Dx s ux) = RZDx k+ ux) = k+ k + /)k)! 4) k π y k s k) RZ Dx s ux) = u 4k) x) ux + y) ux) + ux y) u ) x)y yk k)! uk) x) usig the idetity Γ k + ) = 4)k k! π. k)! I particular, RZD xux) = π RZD 3 xux) = 6 π ux + y) ux) + ux y), ad y y 4 ux + y) ux) + ux y) u ) x)y. To ed off this sectio, we use the followig example to demostrate computatios of the geeralized Riesz derivative.

Mathematics, 8, 89 3 of Theorem 6. Let s > ad s =,,. The, Furthermore, ) Γ + s RZDx s e x = s π Γ s) se x j= x) j Γj s). j)! s + RZ Dx s e x = e x ad RZDx s e x k+ dk = ) s k dx k e x, where k =,,. Proof of Theorem 6. We first assume < s < 3. Lettig ux) = e x we come to By Theorem 4 as e x W 3 R)), RZD s x e x = C,s = C,s u ) x) = e x 4x e x, ad u 4) x) = e x 48x e x + 6x 4 e x. y s Sy) u ) x)y y4 4! u4) x) y s { e x+y) e x + e x y) u ) x)y y4 4! u4) x) }. Clearly, e x+y) e x + e x y) u ) x)y y4 4! u4) x) = e x+y) e x + e x y) e x 4x y + e x y y 4 e x + 4y 4 x e x 4 3 y4 x 4 e x = e x e y + y y4 ) + 4x y e x e y + y ) + 4 3 x4 y 4 e x e y ) + e x e y j=3 xy) j j)! usig e xy + e xy = xy) j j)! j= = + 4x y + 4 3 x4 y 4 + j=3 xy) j. j)! Makig the variable chage u = y, y s e y + y ) y4 = u s e u + u ) u du.

Mathematics, 8, 89 4 of Usig itegratio by parts, we get by otig that u s e u + u ) u du = e u + u u s u s + s u= u s e u + udu = u s e u + udu s e = u + u s s + ) u s + u s+ e u du s s + ) e = u s s + ) s + ) u s+ + u s+3 e u du ss ) s + ) = Γ s + 3) ss ) s + ) Γ s) = ss ), e u + u u u u s e u + u u u if < s < 3. Similarly, we obtai u s e u u s+ = u + e u + u u u s =, e = u + u u + u s =, ad = u + e u u s+ = y s e y + y ) Γ s) = s ), ) y 3 s e y = y j s e y = Γ s), Γj s), for j = 3, 4,.

Mathematics, 8, 89 5 of Hece, RZD s x e x = C,s e x y s e y + y y4 ) +4C,s x e x y s e y + y ) + 4 3 C,sx 4 e x y 3 s e y ) +C,s e x j=3 x) j j)! y j s e y = s π Γ + s ) e x + s+ s π Γ + s ) x e x ) + s+ π Γ s) s 3 Γ s) Γ + s x 4 e x ) Γ + s + s π Γ s) se x j=3 ) Γ + s = s π Γ s) se x j= x) j Γj s) j)! x) j Γj s). j)! Clearly, the series s π ) Γ + s Γ s) se x j= x) j Γj s) j)! ca be exteded to all values of s > ad s =,,. For example, a similar calculatio leads to RZDx s e x = s Γs + )e x + s+ sγs + ) x e x π π + s sγs + ) Γ s) π e x j= ) Γ + s = s π Γ s) se x x) j Γj s) j)! j= x) j Γj s) j)! if < s <. I additio, ) Γ + s s + RZ Dx s e x = s π s + Γ s) se x Γ s) + ) Γ + s = s π s + Γ s) se x Γ s) Γ + s + s π s + ) Γ s) se x j= x) j j)! Γj s) = e x, j= x) j Γj s) j)! by applyig the formula sγ s) = Γ s).

Mathematics, 8, 89 6 of Clearly for j =, 3,, k, Γj s) s k Γ s) = s k j s)j s) s)γ s) Γ s) = ) j k )k ) k j + ). Hece for k =,,, ) ) s k RZ Dx s e x = k π Γ + k e x + k+ k π Γ + k x e x ) k π k Γ + k x) e x j j)! )j kk ) k j + ) j= = ) k+ dk e x k dx by Theorem 5, which ca be verified directly by mathematical iductio. Remark 4. From the physicists Hermite polyomials give by d H x) = ) e x dx e x, we derive s k RZ Dx s e x = ) k+ e x H k x). 3. The Geeralized Riesz Derivative o R with I this sectio, we begi to stu the geeralized Riesz derivative RZ D s x ux) for s > o R, ad obtai its itegral represetatio usig Theorem metioed i the itroductio. I particular, we derive explicit itegral expressios for RZ D k+ x ux) whe k =,,,. Theorem 7. Let < s < ad k = k, k,, k ) be a -tuple of oegative itegers with k + + k =. The for ux) C k R ) defied i the itroductio), RZDx s ux) = ) s ux) = C,s r s Sr)dr 6) where Sr) is the surface itegral o the uit sphere Ω R, give by Sr) = Ω ux + rσ) ux) + ux rσ)dσ. Proof of Theorem 7. We let l = i the case of cetred differece from Defiitio 3 ad derive that yu)x) = k= ) k ) ux + k)y) = ux + y) ux) + ux y) k ad direct computatio implies that d,l s) = s sγ + s ) π Γ s) = C,s by makig use of the idetity Γ z)γz) = π si πz

Mathematics, 8, 89 7 of for ay o-iteger z. Hece, RZDx s ux) = C,s ux + y) ux) + ux y) R y +s, 7) which is well defied for ux) C k R ). Ideed, a secod order Taylor expasio derives ux + y) ux) + ux y) y +s D u L, < s <, y +s which is itegrable ear zero. Furthermore, ux) C k R ) implies that sup y D uy) y R is bouded as y. This deduces that the itegral coverges at ifiity. Usig the spherical coordiates below y = r cos θ y = r si θ cos θ y 3 = r si θ si θ cos θ 3 y = r si θ si θ cos θ y = r si θ si θ si θ, where the agles θ, θ,, θ rage over, π ad θ rages over, π. The Equatio 7) turs out to be where Clearly, the itegral RZDx s ux) = C,s Sr) = Ω r Sr) r +s dr = C,s ux + rσ) ux) + ux rσ)dσ. Sr) r +s dr = Sr) S) r +s dr Sr) dr, r+s coverges as S) = ad Sr) is a eve fuctio with respect to r. It follows from Theorem for < s < that RZDx s ux) = ) s ux) = C,s r s Sr)dr. Remark 5. There is a sig differece betwee Defiitio ad Defiitio 3 for =. Ideed for u W R) ad < s <, from Defiitio, ad RZD s x ux) = ) s ux) = C,s RZD s x ux) = ) s ux) = C,s ux + y) ux) + ux y) y +s ux + y) ux) + ux y) y +s

Mathematics, 8, 89 8 of by Equatio 7), which is directly from Defiitio 3. Let i =,, ad i < s < i +. Applyig Theorem 7 ad Theorem, we ca exted the geeralized Riesz derivative RZ Dx s over the space C k R ) as RZD s r s x ux) = ) s ux) = C,s Sr) r Ω ux) r i Ω i ux) i i! + ) + i ) where k = k, k,, k ) is a -tuple of oegative itegers with k + + k = i +. I particular, RZD xux) = Γ ) + RZDxux) 3 = 3Γ +3 RZD k+ x π + ) π + Sr) r dr, r 4 Sr) r Ω ux) ux) = k k + )k)!γ + + k π + 4) k k! Sr) r Ω ux) ) The followig theorem ca be foud i Referece. dr, r s r k Ω k ux) k k! + ) + k ) Theorem 8. Let ux) C k R ) with > ad i < s < i + for i =,,. The, dr, 8) dr. ux) = ) i+ i+ ux), s i+) )s ux) = ) i i ux) s i + )s ad where k = k, k,, k ) is a -tuple of oegative itegers ad k + k + + k = i +. From Theorem 8 we have s i+) RZ Dx s ux) = ) i+ i+ ux), s i + RZ Dx s ux) = ) i i ux). ad Hece, RZDx s ux) = ) i i ux). s i for i =,,. A a example, we are goig to compute RZ Dxux), where ux) = e x x. from Referece that Sr) = Ω ux + rσ) ux) + ux rσ)dσ = 4πe x x e r k= r k x + x )k k!). It follows

Mathematics, 8, 89 9 of usig The, RZDxux) = Sr) 4π r dr = e x x r e r + e r k= = e x x e x x k= e r r dr x + x )k k!) = πe x x e x x k= e r r k dr r k x + x )k k!) x + x )k k!) Γ k ) = π x πe x x e x x + x )k k )! k!) k= 4 k k )! Γ e r r dr = Γ /) = π, ) e r r k dr = Γ k, k ) k )! = π. 4 k k )! Let ux) C R ). The ux)i m r) has a compact support ad belogs to the space C k R ) for all -tuple of oegative itegers k where the idetity sequece I m r) is give i Sectio. Let i < s < i + with i =,,, ad set S m r) = Sr)I m r). Applyig Equatio 8) we ca defie the geeralized Riesz derivative RZ Dx s C R ) as over the space RZD s x ux) = ) s ux) = C,s m S m r) r Ω ux) m+m m r s if the it exists. To complete this sectio, we preset the followig theorem. Theorem 9. Let s > ad >. The RZ D s x x x ) = o R. r i Ω i ux) i i! + ) + i ) Proof of Theorem 9. We first ote that the fuctio x x C R ), but ot bouded. Clearly, x x ) = / x + + / x )x x ) = x, ) x x ) =. dr, 9)

Mathematics, 8, 89 of ad Assume < s < first. The from Equatio 9), RZDx s x x ) = m+m m C,s m To compute S m r) we come to Clearly, = C,s m m+m m r s S m r) r Ω ux) r s S m r) x r Ω dr. S m r) = I m r) ux + rσ) ux) + ux rσ)dσ Ω Ω ux + rσ) ux) + ux rσ) = x + rσ ) x + rσ ) x x + x rσ ) x rσ ) = 4x r σ σ + x r σ. x r σ dσ = x r where V is the volume of the uit ball i R. Furthermore, Ω Ω σ σ dσ = σ dσ = x r V = x r Ω, Ω σ σ dσ dσ = due to the itegral cacellatio over the uit sphere. Hece, dr Sr) = x r Ω, ad RZDx s x x ) = m C,s m C,s m r s Sr) x r Ω m+m m m = x Ω C,s m dr r s I m r)sr) x r Ω m+m m m r s I m r)dr =. dr It follows from ) x x ) = ) 3 x x ) = = that the result still holds for s >. 4. Coclusios A itegral represetatio is costructed for the geeralized Riesz derivative RZ Dx s ux) for k < s < k + with k =,, i arbitrary dimesios by applyig the ormalizatio of distributio ad the surface itegrals. We further show that RZ Dx s ux) is cotiuous at the ed poits ad well defied for s = + k. I additio, several examples are preseted to demostrate computatios for obtaiig the geeralized Riesz derivatives.

Mathematics, 8, 89 of Author Cotributios: Coceptualizatio, C.L. ad J.B.; methodology, C.L.; software, C.L.; validatio, C.L. ad J.B.; formal aalysis, C.L.; resources, C.L.; writig origial draft preparatio, C.L.; writig review ad editig, C.L.; visualizatio, C.L. All authors have read ad agreed to the published versio of the mauscript. Fudig: This work is supported by NSERC Caada 9-397) ad BURC. Ackowledgmets: The authors are grateful to the academic editor ad reviewers for the careful readig of the paper with productive suggestios ad correctios. Coflicts of Iterest: The authors declare o coflict of iterest. Refereces. Kilbas, A.A.; Srivastava, H.M.; Trujillo, J.J. Theory ad Applicatios of Fractioal Differetial Equatios; Elsevier: North-Hollad, The Netherlads, 6.. Goreflo, R.; Maiardi, F. Fractioal Calculus: Itegral ad Differetial Equatios of Fractioal Order. I Fractals ad Fractioal Calculus i Cotiuum Mechaics; Spriger: New York, NY, USA, 997; pp. 3 76. 3. Podluby, I. Fractioal Differetial Equatios; Academic Press: New York, NY, USA, 999. 4. Srivastava, H.M.; Buschma, R.G. Theory ad Applicatios of Covolutio Itegral Equatios; Kluwer Academic Publishers: Dordrecht, The Netherlads; Bosto, MA, USA; Lodo, UK, 99. 5. Metzler, R.; Klafter, J. The radom walk s guide to aomalous diffusio: A fractioal amics approach. Phys. Rep., 339, 77. CrossRef 6. Metzler, R.; Klafter, J. The restaurat at the ed of the radom walk: Recet developmets i the descriptio of aomalous trasport by fractioal amics. J. Phys. A Math. Ge. 4, 37, R6. CrossRef 7. Madelbrot, B. The Fractal Geometry of Nature; Hery Holt ad Compay: New York, NY, USA, 98. 8. Kwaśicki, M. The equivalet defiitios for the fractioal Laplacia operator. Fract. Calc. Appl. Aal. 7,, 7 5. CrossRef 9. Saichev, A.I.; Zaslavsky, G.M. Fractioal kietic equatios: Solutios ad applicatios. Chaos Iterdiscip. J. Noliear Sci. 997, 7, 753 764, doi:.63/.667. CrossRef. Barros-Neto, J. A Itroductio to the Theory of Distributios; Marcel Dekker, Ic.: New York, NY, USA, 973.. Li, C. O the geeralized fractioal Laplacia. Submitted.. Samko, S.G.; Kilbas, A.A.; Marichev, O.I. Fractioal Itegrals ad Derivatives: Theory ad Applicatios; Gordo ad Breach: Philadelphia, PA, USA, 993. 3. Bayı, S.Ş. Defiitio of the Riesz derivative ad its applicatio to space fractioal quatum mechaics. J. Math. Phys. 6, 57, 35. CrossRef 4. Cushma, J.H. Dyamics of Fluids i Hierarchical Porous Media; Academic Press: Lodo, UK, 99. 5. Beso, D.A. The Fractioal Advectio-Dispersio Equatio: Developmet ad Applicatio. Ph.D. Thesis, Uiversity of Nevada, Reo, NV, USA, 998. 6. Dig, H.F.; Li, C.P.; Che, Y.Q. High-order algorithms for Riesz derivaive ad their applicatios I). Abstr. Appl. Aal. 4, 4, 653797. CrossRef 7. Dig, H.F.; Li, C.P. High-order algorithms for Riesz derivative ad their applicatios V). Numer. Meth. Part. Differ. Equ. 7, 33, 754 794. CrossRef 8. Yag, Q.; Liu, F.; Turer, I. Numerical methods for fractioal partial differetial equatios with Riesz space fractioal derivatives. Appl. Math. Model., 34, 8. CrossRef 9. Huag, Y.; Oberma, A. Numerical methods for the fractioal Laplacia: A fiite differece-quadrature approach. SIAM J. Numer. Aal. 4, 5, 356 384. CrossRef. Muslih, S.I.; Agrawal, O.P. Riesz fractioal derivatives ad fractioal dimesioal space. It. J. Theor. Phys., 49, 7 75. CrossRef. Maiardi, F.; Luchko, Y.; Pagii, G. The fudametal solutio of the space-time fractioal diffusio equatio. Fract. Calc. Appl. Aal., 4, 53 9.. Miller, K.S.; Ross, B. A Itroductio to the Fractioal Calculus ad Fractioal Differetial Equatios; Wiley: Hoboke, NJ, USA, 993. 3. Pozrikidis, C. The Fractioal Laplacia; CRC Press: Boca Rato, FL, USA, 6. 4. Li, C.; Li, C.P.; Humphries, T.; Plowma, H. Remarks o the geeralized fractioal Laplacia operator. Mathematics 9, 7, 3. CrossRef 5. Cai, M.; Li, C.P. O Riesz derivative. Fract. Calc. Appl. Aal. 9,, 87 3. CrossRef

Mathematics, 8, 89 of 6. Gradshtey, I.S.; Ryzhik, I.M. Tables of Itegrals, Series, ad Products; Academic Press: New York, NY, USA, 98. 7. Gel fad, I.M.; Shilov, G.E. Geeralized Fuctios; Academic Press: New York, NY, USA, 964; Volume I. 8. Li, C. Several results of fractioal derivatives i D R + ). Fract. Calc. Appl. Aal. 5, 8, 9 7. CrossRef c by the authors. Licesee MDPI, Basel, Switzerlad. This article is a ope access article distributed uder the terms ad coditios of the Creative Commos Attributio CC BY) licese http://creativecommos.org/liceses/by/4./).