• Nenhum resultado encontrado

Some operational properties of the Laguerre transform

N/A
N/A
Protected

Academic year: 2021

Share "Some operational properties of the Laguerre transform"

Copied!
11
0
0

Texto

(1)

Some operational properties of the Laguerre transform

M. M. Rodrigues, V. N. Huy, and N. M. Tuan

Citation: 1798, 020130 (2017); doi: 10.1063/1.4972722 View online: http://dx.doi.org/10.1063/1.4972722

View Table of Contents: http://aip.scitation.org/toc/apc/1798/1

(2)

Some operational properties of the Laguerre transform

M. M. Rodrigues,

1,a)

, V.N. Huy

2,b)

and N.M. Tuan

3,c)

1CIDMA - Center for Research and Development in Mathematics and Applications Department of Mathematics,

University of Aveiro Campus Universit´ario de Santiago, 3810-193 Aveiro, Portugal.

2Department of Mathematics, College of Science, Vietnam National University, 334 Nguyen Trai, Thanh Xuan,

Hanoi, Vietnam.

3Dept. of Mathematics, College of Education, Vietnam National University, G7 Build, 144 Xuan Thuy Rd., Cau Giay,

Dist., Hanoi, Vietnam.

a)Corresponding author: mrodrigues@ua.pt b)nhat-huy85@yahoo.com

c)nguyentuan@vnu.edu.vn

Abstract.This paper is devoted to the study of some properties of the Laguerre transform. We define new properties of the Laguerre transform in a weighted L2−space. Moreover, we present some results concerning the action of this integral transform over some

class of polynomials.

INTRODUCTION

The Laguerre polynomials appear naturally in many branches of pure and applied mathematics and mathematical physics (see e.g. [1, 2, 3, 6]). Debnath [1] introduced the Laguerre transform and derived some of its properties. He also discussed the applications in study of heat conduction [3] and to the oscillations of a very long and heavy chain with variable tension [2]. Moreover, application of the integral Laguerre transforms for forward seismic modeling can be seen in [5].

The Laguerre transform of a function f (x) is denoted by ˜fα(n) and defined by the integral

L{ f (x)} = efα(n)= ∫

0

e−xxαLαn(x) f (x)dx, n = 0, 1, 2, . . . (0.1) provided the integral exists in the sense of Lesbegue, where Lαn(x) is a generalized Laguerre polynomial of degree n with orderα > −1, and satisfies the following differential equation

d dx [ e−xxα+1 d dxL α n(x) ] + ne−xxαLα n(x)= 0. (0.2)

The sequence of Laguerre polynomial (Lαn(x))n=0have the following property:

∞ ∫ 0 e−xxαLα n(x)L α m(x)dx= ( n+ α n ) Γ(α + 1)δnm, (0.3)

whereδnmis Kronecker function defined by

δnm= { 1, if n = m 0, if n , m (0.4) andΓ(α + 1) = ∫ 0 xαe−xdx.

ICNPAA 2016 World Congress

AIP Conf. Proc. 1798, 020130-1–020130-10; doi: 10.1063/1.4972722 Published by AIP Publishing. 978-0-7354-1464-8/$30.00

(3)

The inverse of the Laguerre transformation is then f (x)= ∞ ∑ n=0n)−1efα(n)Lαn(x) (0< x < ∞), whereδn= ( n+ α n ) Γ(α + 1).

This paper is devoted to the study of the generalized Laguerre transform and some operational properties.

MAIN RESULTS

In this section we define new properties for the Laguerre transform in a weighted L2−space. Moreover, we present

some results concerning the action of this integral transform over some class of polynomial. For 1≤ p ≤ ∞ the space Lp,αis defined via the following formula

Lp,α= { f : (0, ∞) → R : ∫ 0 | f (x)|pe−xxαdx< ∞}

with the norm

|| f ||Lp,α = ( ∫ ∞

0

| f (x)|pe−xxαdx)1/p

where the convention that

|| f ||Lp,α = (∫ 0 | f (x)|pe−xxαdx )1/p = ess supx∈R| f (x)| if p= ∞.

Clearly, if f is an arbitrary polynomial then f ∈ Lp,α. We also define

lp,α=f = (f(n))n=0: ∞ ∑ n=0 | f (n)|p ( n+ α n ) < ∞ with the norm

|| f ||lp,α =    ∞ ∑ n=0 | f (n)|p ( n+ α n ) 1/p . We define the differential operator R via the formula

R[ f (x)] = exx−α d

dx[e−xxα+1 ddxf (x)]

= x f′′

(x)+ (α + 1 − x) f(x) and then for n= 0, 1, ..., we have

Rn[ f (x)]= R[Rn−1[ f (x)]].

Let P(x) be a polynomial. The differential operator P(R) is obtain from P(x) by substituting x → R. i.e., if P(x)=∑nk=0akxkthen P(R)[ f (x)]= nk=0 akRk[ f (x)]. Moreover, we denote by supp efα= {n ∈ Z+: efα(n), 0} the support of the Laguerre transform of f .

(4)

Theorem 0.1 Let 1 ≤ p < ∞, P(x) be a polynomial. Then for an arbitrary infinitely differentiable function f ∈ Lp, there exist the following limit

lim m→∞||P m(R)[ f (x)]||1/m Lpand lim m→∞||P m(R)[ f (x)]||1/m Lp= sup{|P(−n)| : n ∈ supp efα}. Proof: From the definition of differential operator R, we get

L{R[f(x)]} =∫∞ 0 e −xxαLα n(x) exx−α ddx [ e−xxα+1 d fdx]dx =∫0Lαn(x)dxd [e−xxα+1 d fdx]dx. Using the integral by part, we have

L{R[f(x)]} = [ e−xxα+1Lαn(x)d f dx ] 0 − ∫ 0 e−xxα+1dL α n(x) dx d f dxdx = − ∫ 0 e−xxα+1dL α n(x) dx d f dxdx. (0.5) We see that − ∫ 0 e−xxα+1dL α n(x) dx d f dxdx = [ e−xxα+1dL α n(x) dx f (x) ]∞ 0 + ∫ 0 d dx [ e−xxα+1dL α n(x) dx ] f(x) dx = ∫ 0 d dx [ e−xxα+1dL α n(x) dx ] f(x) dx. (0.6)

Then, from (0.2) it follows − ∫ 0 e−xxα+1dL α n(x) dx d f dxdx= −n 0 e−xxαf(x) Lαn(x) dx. Hence L{R[f(x)]}= −nefα(n). (0.7) Similarly, we get L{Rm[f(x)]}= (−1)mnmefα(n). (0.8) Hence, from this and the definition of Pm(R) we obtain

L{Pm(R)[f (x)]}= Pm(−n)efα(n). (0.9) Now, let considerσ an arbitrary number in suppefα. Then

0

e−xxαLασ(x)Pm(R)[ f (x)]dx= Rm(−σ)efα(σ). (0.10) Applying H¨older inequality, we obtain

0e−xxαLα σ(x)P m(R)[ f (x)]dx ≤ (∫0e−xxα|Lα σ(x)| qdx )1/q × (∫ 0 e−xxα|Pm(R)[ f (x)]|pdx )1/p . (0.11)

(5)

Therefore, it follows from (0.10) and (0.11) that |Pm(−σ)ef α(σ)| ≤ (∫ 0 e−xxα|Lασ(x)|qdx )1/q 0 e−xxα|Pm(R)[ f (x)]|pdx. (0.12) That mean |Pm(−σ)ef α(σ)| ≤ (∫ 0 e−xxα|Lασ(x)|qdx )1/q ∥Pm(R)[ f (x)] Lp,α. (0.13)

As Lασis a polynomial, we obtain Lασ ∈ Lq,αand then

(∫

0

e−xxα|Lασ(x)|qdx )1/q

< ∞. It follows from efα(σ) , 0 and (0.13) that

|P(−σ)| ≤ limm→∞∥P

m(R)[ f (x)]1/m

Lp,α. (0.14)

Since (0.14) holds for allσ ∈ suppefα, we obtain

sup{|P(−σ)| : σ ∈ suppefα} ≤ limm→∞∥Pm(R)[ f (x)]∥1L/m

p,α. (0.15)

Now, we will prove that

sup{|P(−σ)| : σ ∈ suppefα} ≥ limm→∞∥Pm(R)[ f (x)]∥1L/m

p,α. (0.16)

Indeed, if supp efαis an unbounded set then

sup{|P(−σ)| : σ ∈ suppefα} = ∞

and (0.16) holds. Now, we only need to prove (0.16) for the case that supp efα is a bounded set, that mean, f is a polynomial and M := sup{|n| : n ∈ suppefα} < ∞. Hence Pm(R)[ f (x)]= Mn=−Mn)−1Pm(−n)efα(n)Lαn(x) (0< x < ∞) and ∥Pm(R)[ f (x)] Lp,α ≤ ∑M n=−Mn)−1Pm(−σ)efα(n)∥LαnLp,α (0.17) ≤ [sup{|P(−σ)| : n ∈ suppefα}]mMn=−Mn)−1efα(n)∥LαnLp,α (0.18) which gives limm→∞∥Pm(R)[ f (x)]∥ 1/m Lp≤ sup{|P(−σ)| : σ ∈ suppefα}. Therefore, (0.16) have been proved. From (0.15) and (0.16), it is easy to see

lim

m→∞||P

m(R)[ f (x)]||1/m

Lp= sup{|P(−n)| : n ∈ supp efα}, and then the proof is complete.

 Let P(x)= x. Then, according Theorem 0.1 we obtain

(6)

Corollary 0.2 Let 1 ≤ p ≤ ∞, P(x) be a polynomial. Then for an arbitrary arbitrary infinitely differentiable function f ∈ Lp, there exist the following limit

lim m→∞||R m[ f (x)]||1/m Lpand lim m→∞||R m[ f (x)]||1/m Lp= deg( f ).

Now, let consider an infinitely differentiable function f : (0, ∞) → R. Assume that we can define the sequence of function (Sm[ f (x)])m=1satisfying the condition

R(Sm)[ f (x)]= Sm−1[ f (x)], ∀m ∈ N.

We also see that, for an arbitrary polynomial f we get efα(0)= 0 if and only if the sequence of function (Sm[ f (x)])m=1 is well defined and

Sm[ f (x)]= ∞ ∑ n=1 (δn)−1 ( −1 n )m efα(n) Lαn(x). Next, we state the following theorem

Theorem 0.3 Let consider P(x) a polynomial and 1 ≤ p ≤ ∞. Then for an arbitrary infinitely differentiable function f ∈ Lpsatisfying efα(0)= 0, we have

lim m→∞||P m (S )[ f (x)]||1L/m 2,α ≥ sup { P(−1n ) : n ∈ suppefα } . Moreover, if f is a polynomial then

lim m→∞||P m(S )[ f (x)]||1/m L2,α = sup { P(−1n ) : n ∈ suppefα } . Proof: We have known that

L{ f (x)} = L{R(S )[ f (x)]} = nL{S [ f (x)]}. (0.19) We deduce that L{S[f(x)]}= −1 n efα(n). (0.20) Similarly, we get L{Sm[f(x)]}= ( −1 n )m efα(n). (0.21)

From this and the definition of Pm(S ), we obtain

L{Pm(S )[f(x)]}= Pm ( −1 n ) efα(n). (0.22)

Now, we considerσ an arbitrary number in suppefα. Then ∫ 0 e−xxαLασ(x)Pm(S )[ f (x)]dx= Rm ( −1 σ ) efα(σ). (0.23)

Applying H¨older inequality we obtain ∫0e−xxαLα σ(x)P m(S )[ f (x)]dx ≤ (∫0e−xxα|Lα σ(x)| qdx )1/q × (∫ 0 e−xxα|Pm(S )[ f (x)]|pdx )1/p . (0.24)

(7)

Therefore, it follows from (0.23) and (0.24) that |Pm ( −1 σ ) efα(σ)| ≤ (∫ 0 e−xxα|Lασ(x)|qdx )1/q 0 e−xxα|Pm(S )[ f (x)]|pdx. That mean |Pm ( −1 σ ) efα(σ)| ≤ (∫ 0 e−xxα|Lασ(x)|qdx )1/q ∥Pm(S )[ f (x)] Lp,α. (0.25)

As Lασis a polynomial, we obtain Lασ ∈ Lq,αand then

(∫ 0 e−xxα|Lα σ(x)| qdx )1/q < ∞. Then it follows from efα(σ) , 0 and (0.25) that

|P ( −1 σ ) | ≤ limm→∞∥P m(S )[ f (x)]1/m Lp,α. (0.26)

Since (0.26) holds for allσ ∈ suppefα, we obtain

sup{|P ( −1 σ ) | : σ ∈ suppefα} ≤ limm→∞∥P m(S )[ f (x)]1/m Lp,α. (0.27)

Now, we will prove

sup{|P ( −1 σ ) | : σ ∈ suppefα} ≥ limm→∞∥Pm(S )[ f (x)]∥1L/m p,α (0.28)

in the case where f is a polynomial. Indeed, put M := deg( f ). Hence,

Pm(S )[ f (x)]= Mn=−Mn)−1Pm ( −1 n ) efα(n)Lαn(x) (0< x < ∞) and ∥Pm(S )[ f (x)] Lp,α ≤ ∑M n=−Mn)−1Pm (−1 σ ) efα(n)∥LαnLp≤ [sup{|P(−1 σ)| : n ∈ suppefα}]mM n=−Mn)−1efα(n)∥LαnLp,α which gives limm→∞∥Pm(S )[ f (x)]∥1L/mp ≤ sup{|P ( −1 σ ) | : σ ∈ suppefα}.

That mean (0.28) have been proved. From (0.27) and (0.28), it is easy to see

lim m→∞||P m (S )[ f (x)]||1L/m p= sup{|P ( −1 n ) | : n ∈ supp efα} and the proof is complete.

 Let P(x)= x, then we have the following result

(8)

Corollary 0.4 Let 1 ≤ p ≤ ∞ and let f an arbitrary polynomial satisfying efα(0) = 0. Then, there exist the following limit lim m→∞||S m[ f (x)]||1/m Lpand lim m→∞||S m[ f (x)]||1/m Lp,α = 1 deg1( f ), where deg1( f )= inf{n ∈ Z+: efα(n), 0}.

We denote by Tmthe set of the algebra polynomials of degree≤ m, and the error of approximation Em( f ) of f by

elements from Tmis

Em( f )= inf P∈Tm

∥ f − P∥L2,α.

Take it in mind, we have the following theorem

Theorem 0.5 Assume that f (x)∈ L2. Then L{ f (x)} ∈ l2and

|| f ||L2,α = √ Γ(α + 1)||L{ f (x)}||l2,α. Moreover, Em( f )= || f ||L2,αmn=0 ( efα(n))2Γ(α + 1) ( n+ α n ) . Proof: We see that

|| f (x) −m n=0(δn)−1c(n)Lαn(x)||L2,α = ∫ 0 e −xxα(f (x)m n=0(δn)−1c(n)Lαn(x) )2 dx (0.29) =∫0e−xxα( f (x))2dx− 2∫∞ 0 e−xxαf (x)m n=0(δn)−1c(n)Lαn(x)dx (0.30) +∫0e−xxα(∑nm=0n)−1c(n)Lαn(x) )2 dx. (0.31) Note that ∫ 0 e−xxαf (x) mn=0n)−1c(n)Lαn(x)dx = ∑m n=0 ∫ 0 e−xxαf (x)(δn)−1c(n)Lnα(x)dx (0.32) =∑m n=0(δn)−1c(n) efα(n). (0.33) Using (0.3) we obtain ∫ 0 e−xxα (∑m n=0 (δn)−1c(n)Lαn(x) )2 dx =∑1≤i, j≤m∫∞ 0 e −xxα((δ n)−2(c(n))2Lαi(x)Lαj(x) ) dx =∑1≤i= j≤m 0 e −xxα((δ n)−2(c(n))2Lαi(x)Lαj(x) ) dx =∑m n=0(δn)−1(c(n))2. (0.34)

Using (0.29), (0.32) and (0.34) we get || f (x) − mn=0 (δn)−1c(n)Lαn(x)||L2,α = || f ||2L2,α − 2 ∑m n=0(δn)−1c(n) efα(n)+ ∑m n=0(δn)−1(c(n))2 = || f ||2 L2,α− ∑m n=0(δn)−1( efα(n))2+ ∑m n=0(δn)−1( efα(n)− c(n))2 (0.35) which gives inf (c0,...,cm)∈Rm || f (x) − mn=0 (δn)−1c(n)Lαn(x)||L2,α = || f ||2L2,αmn=0 (δn)−1( efα(n))2.

(9)

Then, it follows from∑mn=0n)−1c(n)Lαn(x) which is the algebra polynomial of degree m Em( f )= || f ||2L2,αmn=0 (δn)−1( efα(n))2= || f ||L2,αmn=0 ( efα(n))2Γ(α + 1) ( n+ α n ) . Let c(n)= efα(n) for all n= 0, 1, . . . , m, we obtain

|| f (x) − mn=0 (δn)−1efα(n)Lnα(x)||L2,α = || f ||2L2,αmn=0 (δn)−1( efα(n))2.

Therefore, since lim

m→∞|| f (x) − mn=0 (δn)−1efα(n)Lαn(x)||L2,α = 0, we have L{ f (x)} ∈ l2,αand || f ||2 L2,α = mn=0 (δn)−1( efα(n))2, and then Γ(α + 1)||L{ f (x)}||l2,α = || f ||L2,α. The proof is complete.

 We are now able to present the following result

Theorem 0.6 Let P(x) be the polynomial. Then for an arbitrary polynomial f , we always have ||Pm(R)[ f (x)]|| L2,α≤ sup{|P(−n)| : n ∈ suppefα}||Pm−1(R)[ f (x)]||L2,α, (0.36) ||Pm(R)[ f (x)]|| L2,α ≥ inf { |P(−n)| : n ∈ suppefα } ||Pm−1(R)[ f (x)]|| L2,α (0.37) and lim m→∞||P m(R)[ f (x)]||1/m L2,α = deg( f ). (0.38)

Proof: We have known that

L{Pm(R)[f(x)]}= Pm(−n) efα(n). Then it follows from Theorem 0.5 that

||Pm(R)[ f (x)]|| L2,α = √ Γ(α + 1)||L{Pm(R)[f(x)]}|| l2,α = √Γ(α + 1) ( ∑ n=0 ( Pm(−n)ef α(n))2 ( n+ α n ))1/2 . Hence ||Pm(R)[f(x)]|| L2,α = √ Γ(α + 1)    ∑ n∈suppefα ( Pm(−n)efα(n))2 ( n+ α n ) 1/2 ≤ √Γ(α + 1) sup{|P(−n)| : n ∈ suppefα } ×    ∑ nsuppefα ( Pm−1(−n)efα(n))2 ( n+ α n ) 1/2 = sup{|P(−n)| : n ∈ suppefα } ||Pm−1(R)[f (x)]|| L2,α

(10)

and (0.36) just been proved.

Now, we will prove (0.37). Indeed, we have ||Pm(R)[f (x)]|| L2,α= √ Γ(α + 1)    ∞ ∑ n=0 (Pm(−n)efα(n))2 ( n+ α n ) 1/2 which gives ||Pm(R)[f(x)]|| L2,α = √ Γ(α + 1)    ∑ n∈suppefα (Pm(−n)efα(n))2 ( n+ α n ) 1/2 . It follows that ||Pm(R)[f(x)]|| L2,α ≥ √ Γ(α + 1) inf{|P(−n)| : n ∈ suppefα} ×    ∑ n∈suppefα (Pm−1(−n)efα(n))2 ( n+ α n ) 1/2 = inf{|P(−n)| : n ∈ suppefα}||Pm−1(R)[f(x)]||L2,α, and (0.37) just been proved. Note that, we check (0.37) from Theorem 0.1. The proof is complete.

 Since for all a polynomial f and n > deg( f ) then efα(n) = 0, we obtain suppefα ⊂ {0, 1, . . . deg( f )}. Then, by applying Theorem 0.6 we have the following corollary

Corollary 0.7 Let P(x) be the polynomial. Then for an arbitrary polynomial f , we always have ||Pm(R)[ f (x)]|| L2,α ≤ sup{|P(−n)| : 0 ≤ n ≤ deg( f )}||Pm−1(R)[ f (x)]||L2,α, ||Pm (R)[ f (x)]||L2,α ≥ inf { |P(−n)| : 0 ≤ n ≤ deg( f )}||Pm−1 (R)[ f (x)]||L2,α and lim m→∞||P m(R)[ f (x)]||1/m L2,α = deg( f ).

Let P(x)= x, then it follows from Theorem 0.6

Corollary 0.8 Let P(x) be the polynomial. Then for an arbitrary polynomial f , we always have ||Rm[ f (x)]|| L2,α ≤ deg( f )||P m−1(R)[ f (x)]|| L2,α, ||Rm[ f (x)]|| L2,α ≥ deg1( f )||P m−1(R)[ f (x)]|| L2,α,

where deg1( f )= inf{n ∈ Z+: efα(n), 0} and lim

m→∞||R

m[ f (x)]||1/m

L2,α = deg( f ).

Remark.Since f is the polynomial with degree d, f (x) can express via f (x)= dn=0 anLnα(x). Then {n ∈ Z+: efα(n), 0} = {n ∈ N : an, 0} and C= inf{|P(−n)| : n ∈ Z+, efα(n), 0} = inf{|P(−n)| : n ∈ Z+, an, 0}.

(11)

ACKNOWLEDGMENTS

The work of the first author was supported by Portuguese funds through the CIDMA - Center for Research and Development in Mathematics and Applications, and the Portuguese Foundation for Science and Technology (“FCT– Fundac¸˜ao para a Ciˆencia e a Tecnologia”), within project UID/MAT/ 0416/2013. The work of V.N.Huy was supported by Vietnam National University under grant number QG.16.08.

REFERENCES

[1] L. Debnath, On Laguerre Transforms, Bull Calcutta Math . Soc 55, 69–77 (1960).

[2] L. Debnath, Applications of Laguerre transform to the problem of oscillations of a very long and heavy chain, Annali dell’ Univ. di Ferrara, Sezione VII-Scienze Mathematiche, IX, 149–151 (1961).

[3] L. Debnath, Application of Laguerre transform to heat conduction problem (1962), Annali dell’ Univ. di Ferrara, Sezione VII-Scienze Mathematiche,X, 17–19 (1962).

[4] L. Debnath and D. Bhatta Integral transform and their applications, 2nd edition, Chapman & Hall/CRC Press, Boca Raton, FL, 2007.

[5] V. Konyukh, B. G. Mikhailenko, and A. A. Mikhailov, Application of the integral Laguerre transforms for forward seismic modeling,J. Comp. Acous, 9(4), 1523–1541 (2001).

[6] J. McCully, The Laguerre transform,SIAM Review2, 185–191 (1960). [7] I. N. Sneddon, The use of Integral Transforms, McGraw-Hill, New York, 1972.

Referências

Documentos relacionados

We also determined the critical strain rate (CSR), understood as the tangent of the inclination angle between the tangent to the crack development curve and the crack development

The iterative methods: Jacobi, Gauss-Seidel and SOR methods were incorporated into the acceleration scheme (Chebyshev extrapolation, Residual smoothing, Accelerated

The probability of attending school four our group of interest in this region increased by 6.5 percentage points after the expansion of the Bolsa Família program in 2007 and

The resulting multilayer problem for the coefficients of the eigenfunction series expansion of the approximate solution is then solved in the Laplace space.. Again, the approximation

Evolução das divulgações do goodwill pós adoção das IFRS ao nível de cumprimento das divulgações dos testes de imparidade, especialmente no período de crise financeira, ou

E eu que- ria dizer que o Brasil somente ofertará esse setor de serviços educacionais se vocês, aqui, da Associação, chegarem a um consenso: “Queremos que o setor seja incluído

E tal como Alfama, também a Mouraria “não constitui uma circunscrição administrativa” (Firmino da Costa, 1999:67). Contudo, podemos identificar o bairro no mapa da cidade