ebook img

Generalized Hermite polynomials PDF

16 Pages·0.2 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Generalized Hermite polynomials

GENERALIZED HERMITE POLYNOMIALS 1 V.V. Borzov Department of Mathematics, St.Petersburg University of Telecommunications, 191065, Moika 61, St.Petersburg, Russia 1 The new method for obtaining a variety of extensions of Hermite polynomials is given. 0 As a first example a family of orthogonal polynomial systems which includes the general- 0 ized Hermite polynomials is considered. Apparently, either these polynomials satisfy the 2 differential equation of the second order obtained in this work or there is no differential n a equation of a finite order for these polynomials. J 5 KEYWORDS:orthogonalpolynomials,generalizedoscillatoralgebras,generalizedderiva- 2 tion operator. ] MSC (1991): 33C45, 33C80, 33D45, 33D80 A Q 1. Introduction . h t Inourformerwork([1])weconstructedanappropriateoscillator algebraA correspond- a µ m ing to the system of polynomials which are orthonormal with respect to a measure µ in [ the space Hx = L2(R1;µ(dx)). By a standard manner the energy operator (hamiltonian) 1 Hµ = Xµ2 +Pµ2 was defined. The position operator Xµ was introduced by the recurrent relations of the given polynomials system; the momentum operator P was determined as v µ 6 an unitary equivalent to the position operator Y in the dual space H = L2(R1;ν(dy)). µ y 1 In ([1]) it was proved that the usual differential equations for the classical polynomials 2 are equivalent to the equations of the form H ψ = λ ψ , where the eigenvalues of the 1 µ n n n 0 corresponding hamiltonian Hµ denote by λn. The central problem with a derivation of 1 the differential equations was findinga representation of the annihilation operator a (or µ− 0 another ”reducing” operator) of the algebra A by a differential operator in the space H . / µ x h Unfortunately, these formulas ([2]) are rather complicated in the general case. Therefore t a our interest is in describing such orthogonal polynomials systems for which appropriate m representations are simple. On this basis one can obtain some differential equations of : a finite order (it is desirable that we have to deal only with differential equations of the v i second order). X The results of this work may be thought of as a first step forward in this direction. r a From our point of view we consider a family of orthogonal polynomial systems which includes the generalized Hermite polynomials ([3]). These polynomials have been studied extensively in the monograph ([4]). Therefore the polynomials of the considered family is called Hermite-Chihara polynomials. The paper is organized as follows. A generalized derivation operator is introduced in Sec.2. By these operators a family of the Hermite-Chihara polynomials is determined in Sec.3. More exactly the annihilation operator a of the algebra A corresponding to −µ µ the system of the polynomials may be represented by a generalized derivation operator D . This operator is defined by a positive sequence ~v. In what follows we shall call this ~v sequence ~v a ”governing sequence”. In Sec.4 we construct the generators X , P and H µ µ µ of the algebra A corresponding to a system of the Hermite-Chihara polynomials. As µ an example of such polynomials we consider the ”classical” Hermite-Chihara polynomi- als ([4]) in Sec.5. Moreover, in this section a new derivation of the well-known ([3],[4]) 1This research was supported by RFFIgrant No00-01-00500 1 2 V.V.BORZOV differential equation for these polynomials is presented. Further, in Sec.6 we introduce a special family of orthogonal systems of Hermite-Chihara polynomials which includes the classical Hermite-Chihara polynomials. Furthermore, in this section we construct a ”gov- erning sequence” ~v of an appropriate generalized derivation operator D . Then in Sec.7 ~v we obtain a differential equation of the second order for above-mentioned polynomials by analogy with the derivation of the differential equation given in Sec.5. Finally, in the conclusion we consider the following conjecture. If the polynomials of a system of orthog- onal Hermite-Chihara polynomials satisfy a differential equation of the second order, then these polynomials belong to the special family of orthogonal systems of Hermite-Chihara polynomials introduced in Sec.6. Moreover, the other Hermite-Chihara polynomials do not satisfy any differential equation of a finite order. 2. Generalized derivation operator In this section we introduce a new class of differential operators (they are the infinite order in general case) which play a large role in the construction of the Hermite-Chihara polynomials. Let ~v = {vn}∞n=0 be a monotone nondecreasing sequence: 1= v v v v .... (2.1) 0 1 2 n ≤ ≤ ≤ ··· ≤ ≤ This sequence ~v define a linear operator D by the relations: ~v D x0 = 0, D xn = v xn 1, n= 1,2..., (2.2) ~v ~v n 1 − − on the set of the formal power series of real argument x. We will seek for the operator D of the type ~v ∞ dm D = a xn . (2.3) ~v nm dxm n,m=0 X Substituting (2.3) in (2.2), we get the following formula: ∞ dk D = ε xk 1 , (2.4) ~v k − dxk k=0 X The coefficients {εk}∞k=0 are defined by the recurrent relations: v ε ε ε1 = v0 = 1, εk = k−1 εk 1 k−2 1 , k = 1,2.... (2.5) k! − − − 2! −···− (k 1)! − Definition 2.1. A differential operator D determined by formulas (2.4),(2.5) is called a ~v generalized derivation operator induced of the sequence ~v. Lemma 2.2. For the order of a generalized derivation operator D defined by formulas ~v (2.4),(2.5) to be finite it is necessary and sufficient that the following equalities: ε = ε = = 0. (2.6) k+1 k+2 ··· was valid. To take three examples of generalized derivation operators of a finite order. 1. Let k = 1. There exist a unique solution of the system (2.6): ~v = {n+1}∞n=0. (2.7) The generalized derivation operator D corresponding to ~v take the following form: ~v d D = . (2.8) ~v dx GENERALIZED HERMITE POLYNOMIALS 3 2. Let k = 2 and let v to be a number such that v 1. There exist a one-parameter 1 1 ≥ family of the solution ~v = {vn}∞n=0 of the system (2.6): v = 1, v = C2 v n2+1, n= 1,2.... (2.9) 0 n n+1 1− The generalized derivation operator D corresponding to ~v take the following form: ~v d v d2 1 D = +x( 1) . (2.10) ~v dx 2 − dx2 If v = 4, then from (2.9),(2.10) we have 1 d d2 ~v = (n+1)2 ∞n=0, D~v = dx +xdx2 . (2.11) 3. Let k = 3 and let v (cid:8),v to be(cid:9)some number such that 1 v v . There exist a 1 2 1 2 ≤ ≤ two-parameter family of the solution ~v = {vn}∞n=0 of the system (2.6): (n+1)n(n 2) v = 1 v , v = C3 v − v + 0 ≤ 1 n n+1 2− 2 1 (n+1)(n 1)(n 2) + − − , n= 1,2.... (2.12) 2 The generalized derivation operator D corresponding to ~v take the following form: ~v d v d2 v 3v +3 d3 D = +x( 1 1) +x2 2− 1 . (2.13) ~v dx 2 − dx2 3! dx3 If v = 8, v = 27, then from (2.12) and (2.13) we have 1 2 d d2 d3 ~v = (n+1)3 ∞n=0, D~v = dx +xdx2 +x2dx3 . (2.14) (cid:8) (cid:9) 3. Hermite-Chihara polynomials Let µ be a symmetric probability measure, i.e. the all odd moments of the measure µ are vanish and ∞ µ(dx) = 1. In this section we consider a system of polynomials , which are orthonorm−∞al with respect to the measure µ , such that there is a representation R of the annihilation operator of the oscillator algebra A corresponding to this system by µ a generalized derivation operator. Recall ([1]) that the recurrent relations of a canonical orthonormal polynomials system {ψn(x)}∞n=0 take the following form: xψ (x) = b ψ (x)+b ψ (x), n 1, (3.1) n n n+1 n 1 n 1 − − ≥ x ψ (x) = 1, ψ (x) = . (3.2) 0 1 b 0 In ([1])it was described how to get the positive sequence {bn}∞n=0 from the given sequence {µ2n}∞n=0 of even moments of a symmetric positive measure µ. The question we are interested now is when for a canonical orthonormal polynomials system {ψn(x)}∞n=0 there are two sequences such that: 1. a positive sequence ~v = {vn}∞n=0 which satisfies (2.1); 2. a real sequence ~γ = {γn}∞n=0 for which are hold the following relations: D ψ = 0, D ψ = γ ψ , n = 1,2..., (3.3) ~v 0 ~v n n n 1 − where the generalized derivation operator D is determined by formulas (2.4),(2.5). ~v We denote by [n] the following symbol: b2 [0] = 0, [n]= n−1, n = 1,2.... (3.4) b2 0 4 V.V.BORZOV Let J be a symmetric Jacobi matrix J = {bij}∞i,j=0 whichhasthepositiveelements b = b , i = 0,1,... onlydistinctfromzero. Then i,i+1 i+1,i the polynomials of the first kind can be represented in the form ([5]): ǫ(n) 2 ( 1)m ψn(x) = − b20m−nα2m 1,n 1xn−2m, (3.5) [n]! − − m=0 X where the greatest integer function ispdenoted byǫ(α) . The coefficients α for any 2m 1,n 1 n 1, ǫ(n) m 1 are defined by the following equalities: − − ≥ 2 ≥ ≥ n−1 k1−2 km−1−2 α = 0, α = [k ] [k ] [k ]. (3.6) 1,n 1 2m 1,n 1 1 2 m − − − − ··· k1=X2m−1 k2=X2m−3 kXm=1 Substituting (3.6),(3.5) and (3.4) into (3.3), it is easy to prove the following theorem ([2]). Theorem 3.1. Lettheorthonormal polynomial system{ψn(x)}∞n=0 isdefinedby(3.6),(3.5) and (3.4). For existence two sequences ~v = {vn}∞n=0 and ~γ = {γn}∞n=0 such that the con- ditions (3.3) are hold it is necessary and sufficient that 1. the sequence ~v = {vn}∞n=0 satisfies (2.1) and the following conditions: v v +v v = v v +v v , (3.7) n 2 2p 1 2p 3 n 2p n 2p 3 2p 1 n 2p − − − − − − − for any n 2, 2p n; ≥ ≤ 2. the coefficients α take the following form 2m 1,n 1 − − [2m 1]!!(v )! n 1 α2m 1,n 1 = − − , (vk)! = v0v1 vk, (3.8) − − (v2m 1)!(vn 2m 1)! ··· − − − as n 1, 2m n and regarding (v )!= (v )! = 1. 1 0 In≥this case th≤e sequence ~γ = {γn}−∞n=0 is defined by the following formulas: v v 1 n 1 γn = b2(v −v ), n ≥ 1. (3.9) r 0 n− n−2 Here we will not given the proof of this theorem (see [2]) to save room. However we present some formulas arising from the proof. These expressions relate the sequence ~γ = {γn}∞n=0 and the coefficients α2m−1,n−1: 1 [2] γ = , b γ = ε +ε , (3.10) 1 0 2 1 2 b 2 0 p α 2p 1,2p [2p+1]γ2p+1 = − γ1, (3.11) [2p 1]!! − p α 2p 1,2p+1 [2p+2]γ2p+2 = − [2]γ2, (3.12) α 2p 1,2p p − p where ε and ε are defined by (2.5). 1 2 Now we shall give the following definition. Definition 3.2. The orthonormal polynomials system {ψn(x)}∞n=0 completed in Hx = L2(R;µ(dx)) is called a system of Hermite-Chihara polynomials if these polynomials are defined by (3.6),(3.5) and (3.4). GENERALIZED HERMITE POLYNOMIALS 5 Remark 3.3. It is clear that the Hermite polynomials fall in this class. Here d n ~v = {n+1}∞n=0, D~v = dx, b2n−1 = 2, [n] = n, n ≥ 1. According to (3.8), we have n! α = . (3.13) 2m−1,n−1 2mm!(n 2m)! − Substituting (3.13) into (3.5), we obtain the usual form of the Hermite polynomials (see [1], [3] ). In addition, γ = √2n, n 1 and then (3.3) is reduced to the usual rule of n ≥ derivation for the Hermite polynomials: d H (x) = 2nH (x) . (3.14) n n 1 dx − Remark 3.4. According to the theorem 3.1, by any sequence~v = {vn}∞n=0 complying with (2.1) and (3.7) we can write the the coefficients α which take the following form: 2m 1,n 1 − − v (v v ) n 1 n n 2 [1] = 1, [n]= − − − , n 2, (3.15) v ≥ 1 v (v v ) b2n−1 = b20 n−1 nv1− n−2 , n ≥ 2. (3.16) The polynomials ψ (x) satisfy the recurrent relations (3.1) and (3.2). By solving the n Hamburger moment problem of the Jacobi matrix J, we obtain the symmetric probability measure µ such that the polynomials of the system {ψn(x)}∞n=0 are orthonormal with respect to µ. If the moment problem for the Jacobi matrix J is a determined one, then the measure µ is defined uniquely. Otherwise (when the moment problem for the Jacobi matrix J is a undetermined one) there is a infinite family of such measures (see [6]). 4. Oscillator algebra for the Hermite-Chihara polynomials In this section we constructthe generalized Heisenberg algebra A correspondingto the µ system of the Hermite-Chihara polynomials (see [1]). Let ~v = {vn}∞n=0 be the positive sequence such that the conditions (2.1) and (3.7) are hold. Then the sequence {bn}∞n=0 can be found by (3.16). Furthermore, we obtain the system of the Hermite-Chihara polynomials {ψn(x)}∞n=0 by the formulas (3.5) and (3.8). These polynomials satisfy the recurrent relations (3.1) and (3.2) with above-mentioned coefficients {bn}∞n=0. Under the condition ∞ b 1 = v1 ∞ 1 = . (4.1) nX=0 −n b0 nX=1 vn(vn+1−vn−1) ∞ the moment problem for the correspondipng Jacobi matrix is a determined one (see [6]). There is the only symmetric probability measure µ such that the polynomials {ψn(x)}∞n=0 areorthonormalinthespaceH = L2(R;µ(dx)). Inaddition,theevenmomentsµ ofthe x 2n measure µ can be found from the following algebraic equations system (b = 0, n 0) 1 − ≥ ǫ(n) ǫ(n) 2 2 ( 1)m+s − α α µ = b2 +b2. (4.2) (bn 1)! 2m−1,n−1 2s−1,n−1 2+2n−2m−2s n−1 n mX=0mX=0 − Itis easy to check that thecondition (4.1) is correct for the ”classical” Hermite-Chihara polynomials to be considered in the next section. 6 V.V.BORZOV From theorem 3.1 it follows that there are the generalized derivation operator D de- ~v termined for given sequence ~v by formulas (2.4),(2.5) and the sequence ~γ = {γn}∞n=0: v n 1 γn = − , n 1, (4.3) b ≥ n 1 − such that the relations (3.3) are valid . Using the methods of ([1]), we construct the generalized Heisenberg algebra A corre- µ sponding to the ortonormal system {ψn(x)}∞n=0. By the usual formulas we define ladder operators a (the annihilation operator), a+ (the creation operator) and the number op- −µ µ erator N. It is readily seen that a = D f(N), (4.4) −µ ~v The operator-function f(N) acts on basis vectors {ψn(x)}∞n=0 by formulas: v v f(N)ψ0 = 0, f(N)ψ1 = √2b20ψ1, f(N)ψn = √2b20 n −v n−2ψn, (4.5) 1 where n 2. The position operator X is defined by the recurrence relations (3.1) and µ ≥ (3.2). Using X and a , we determine by the well-known formulas (see [1]) the operators µ −µ a+, P (the momentum operator) and H (hamiltonian): µ µ µ P = ı(√2a X ), a+ = √2X a , (4.6) µ −µ µ µ µ −µ − − H = X2+P2 = √2(a X +X a+) = a a++a+a . (4.7) µ µ µ −µ µ µ µ −µ µ µ −µ We have the following commutation relation: [a ,a+] = 2(B(N +I) B(N)). (4.8) −µ µ − The operator-function f(N) acts on basis vectors by formulas: v (v v ) B(N)ψ0 = 2b20, B(N)ψn = b2n−1ψn = b20 n−1 nv1− n−2 ψn, n ≥ 1. (4.9) Moreover, the ”energy levels” are 2b2 λ = 2b2, λ = 2(b2 +b2) = 0(v v v v ), (4.10) 0 0 n n−1 n v1 n n+1− n−1 n−2 where n 1. ≥ In what follows our prime interest is with the following question. Is it possible to get a differential equation of the second order for Hermite-Chihara polynomials from the equation H ψ = λ ψ . µ n n n 5. Classical Hermite-Chihara polynomials Now we consider a particular case of the Hermite-Chihara polynomials, namely, the well-known (see [3]) generalized Hermite polynomials which have been studied extensively in ([4]) (see also ([8])). We denote by H the Hilbert space γ 1 H = L2(R; xγ(Γ( (γ +1))) 1exp( x2)dx), γ 1. (5.1) γ − | | 2 − ≥ − Ucosminpglemteedthiondsthoef(s[p1]a)c,ewHec.onTsthreucptotlhyenocamnioanlsicψalo(xrt)hsoantoisrfmyatlhpeolryencoumrrieanltssryeslatetmion{sψ(n3(.x1))}∞n=0 γ n GENERALIZED HERMITE POLYNOMIALS 7 and (3.2). The coefficients {bn}∞n=0 are defined by formulas (3.16), where b0 = γ+21 and the sequence ~v = {vn}∞n=0 is given by the following equalities: q γ+n+1 n = 2m, v = γ+1 (5.2) n n+1 n = 2m+1. ( γ+1 It is clear that v0 = 1, v1 = γ+21 = b−02. The coefficients {bn}∞n=0 are defined by the formulas: 1 √n n = 2m, b = (5.3) n−1 2( √n+γ n = 2m+1. The formulas (3.8),(3.5),(3.4) give a explicit form of the polynomials ψ (x). Recall that n the polynomials Kγ(x) = s ψ (x), n 0, n n n ≥ as s = 1 and s = (b )!, are named the generalized Hermite polynomials in ([4])( 0 n n 1 − see also ([8])). In what follows we shall call the polynomials ψ (x) as the ”classical n Hermite-Chihara polynomials”. It is easy to prove that the family of these polynomials is a particular case of the more general class of Hermite-Chihara polynomials and that the generalized derivation operator D corresponding to the given sequence ~v is determined ~v by formulas (2.4) and (2.5), where ( 2)m 1 γ − ε =1, ε = − , m 2. (5.4) 1 m m! γ+1 ≥ In addition, the sequence ~γ = {γn}∞n=0 appearing in (3.3) is defined by equalities: √2 √n n= 2m, γ = (5.5) n γ +1 √n+γ n = 2m+1. ( Comparing (4.4), (4.5) and (3.3), we obtain γ+1 a = D . (5.6) −µ √2 ~v The following formulas are known ([4])( see, also([8])): d d n (n 1)θ ψ0 = 0, ψn = ψn 1+ − n X−1ψn 2 = (5.7) dx dx bn 1 − 2bn 1bn 2 − − − − θ n = 2b ψ ψ , n 1, (5.8) n 2 n 1 n − − − x ≥ where 1 ( 1)n θ = θ (γ) = γ − − . (5.9) n n 2 Taking into account how the annihilation operator a and the number operator N act on −µ the basis vectors {ψn(x)}∞n=0, it is easy to get from the relations (5.7)-(5.9) the following formula: d X N = (a )2. (5.10) µdx − −µ Note also that the action of the position operator Xµ on the basis vectors {ψn(x)}∞n=0 is defined by (3.1) and (5.3). Now we consider the operator Θ =2B(N) N, (5.11) N − 8 V.V.BORZOV wheretheoperator-function B(N) is definedby (4.10). Using(5.9) and(5.11), wesee that Θ ψ = θ ψ , n 1. (5.12) N n n n ≥ Taking intoaccount theequation H ψ = λ ψ , whereahamiltonian H definedby(4.7), µ n n n µ and the equality a+a = 2B(N), we have the following relation: µ −µ a a+ = 2B(N +I). (5.13) −µ µ Moreover, from (5.8) it follows that 1 d 1 a = + X 1Θ . (5.14) −µ √2dx √2 µ− N Now from (5.13), (5.14) and (4.7) we have d 1 d 1 ( +X 1Θ )(X X 1Θ )= 2B(N +I). (5.15) dx µ− N µ − 2dx − 2 µ− N Multiplying both sides of (5.15) by 2X from the left, we obtain: µ − d d2 2X (I +X )+X +X ( X 2Θ +X 1Θ ) (5.16) − µ µdx µdx2 µ − µ− N µ− N d 2Θ X +Θ +Θ X 1Θ = 2X 2B(N +I). (5.17) − N µ Ndx N µ− N − µ It is not hard to prove that : Θ X 1Θ ψ = 0, Θ X = X Θ , (5.18) N µ− N n N µ µ N+I 2X (2B(N +I) Θ ) 2X = 2X N, (5.19) µ N+I µ µ − − d d (Θ + Θ )ψ = γψ . (5.20) Ndx dx N n n′ Applyingbothsidesof(5.17)toψ andusing(5.18)-(5.20), wegetthefollowingdifferential n equation θ xψ +(γ 2x2)ψ +(2nx n)ψ = 0, n 0, (5.21) n′′ − n′ − x n ≥ which is coincident with the well-known differential equation for the classical Hermite- Chihara polynomials [4] (see also [3]). Remark 5.1. The generators a+,a of the generalized Heisenberg algebra A correspond- µ −µ µ ing to the classical Hermite-Chihara polynomials system subject to the following commu- tative relation (see [1]): [a ,a+] = (γ +1)I 2Θ . (5.22) −µ µ N − The ”energy levels” of the associated oscillator are equal to: λ = γ +1, λ = 2n+γ+1, n 1. (5.23) 0 n ≥ Finally, it follows from (4.7) and (5.14) that the momentum operator take the following form: d P = ı( +X 1Θ X ). (5.24) µ dx µ− N − µ GENERALIZED HERMITE POLYNOMIALS 9 6. Construction of the ”governing sequence” ~v of a generalized derivation operator D for the classical Hermite-Chihara polynomials ~v Let {ψn(x)}∞n=0 be a orthonormal Hermite-Chihara polynomials system. According to theorem 3.1, there is a sequence ~v such that (2.1) and (3.7) are hold. The formula (3.9)allows us to define the sequence ~γ by ~v. Furthermore, the generalized derivation operator D~v corresponding to~v is a reducing operator for the system {ψn(x)}∞n=0, i.e. the equalities (3.3) are valid. In this section we obtain the exact condition on ~v which select some family of Hermite-Chihara polynomials. This family is a natural extension of the set of classical Hermite-Chihara polynomials. The associated set of~v is a three-parameter family dependingon theparameters b ,v andv . Butit turnout thatthe parameter v is 0 1 2 1 unessential, so that the above-mentioned family is really a two-parameter one. According to (2.4) and (2.5)), the generalized derivation operator D corresponding to a sequence ~v ~v complying with (2.1) and (3.7) take the following form: d d D = X 1(B +B )= +X 1B , B = X , (6.1) ~v − 1 1 − 1 1 dx dx where ∞ dk B = ε xk . (6.2) 1 k dxk k=2 X The coefficients ε in (6.2) are defined from given sequence ~v by the recurrent relations k (2.5). Remark 6.1. For the classical Hermite-Chihara polynomials it follows from(6.2) and (5.4) that d cl X +B = δ(N), (6.3) 1 dx where δ(N) is the projection on the subspace of the polynomials of odd degree: δ(N)xn =θ (1)xn, n 0, n ≥ (see (5.9)) and hence δ(N)ψcl = θ (1)ψcl. n n n We denote by {ψn(x)}∞n=0 the orthonormal Hermite-Chihara polynomials system which is constructed according to remark 3.4 for given sequence ~v. Obviously, B ψ = B ψ = 0. 1 0 1 1 Now we shall restrict our consideration to a particular class of the Hermite-Chihara polynomials for which there are two real sequence ~δ = δn ∞n=2 and β~ = βn ∞n=0 such that: (cid:8) (cid:9) (cid:8) (cid:9) B ψ = δ Xψ , B ψ = δ Xψ +β ψ , n 3. (6.4) 1 2 2 1 1 n n n 1 n n 2 − − ≥ Replacing δ by δ and β by β , we see from (6.1) and (3.3) that the condition (6.4) is n n n n valid for B too. Then it is clear that the assumption (6.4) takes the place of the rule of 1 derivation (5.8). Substituting (6.2) into (6.4), and taking into account (3.5), we obtain 10 V.V.BORZOV the following relation: n ǫ(n−2k)( 1)m (n m)! −[n]! b20m−nα2m−1,n−1xn−2m(n 2−m k)! = k=2 m=0 − − X X pǫ(n−2) 2 ( 1)m = βn − b20m−n+2α2m 1,n 3xn−2m−2+ [n 2]! − − mX=0 − ǫ(n−p1) 2 ( 1)m +δn [−n 1]!b20m−n+1α2m−1,n−2xn−2m, n≥ 2. (6.5) mX=0 − Equating the coefficients at xn ipn the both sides of (6.5), we get δ b = A (2), n 2, (6.6) n n 1 n − ≥ where we used the notation s ε k A (m)= s! , s m, (6.7) s (s k)! ≥ k=m − X and the coefficients ε are defined by formulas (2.5). For the classical Hermite-Chihara k polynomials from (5.4), (5.5) and binomial formula it follows that (as n 2) ≥ cl √2γ √n n = 2m, δ = (6.8) n γ+1( √nn−+1γ n = 2m+1. In order that to find the quantities β in (6.4) we equal the coefficients at xt in the both n sides of (6.5) (as 0 t < n). Obviously, a coefficients at xt only distinct from zero when ≤ it is valid the following condition: n n t = 2p, 1 p ǫ( ). (6.9) − ≤ ≤ 2 We consider separately three cases t = 0,1,2. 1. Let t = 0, n =2p. We have β = 0, p 1. 2p ≥ 2. Let t = 1, n =2p+1. We have [2p]α 2p 3,2p 2 δ2p+1 = − − β2p+1, p 1. (6.10) b α ≥ p 0 2p 1,2p 1 − − 3. Let t = 0, n 2 = 2p. Regarding p 1, we have − ≥ α A (2) = β [2p+2][2p+1]α + 2p−1,2p+1 2 − 2p+2 2p−3,2p−1 p +α2p 1,2pA2p+2(2). (6.11) − Taking into account that β = 0, we have from here 2p α A (2) α A (2) = 0, p 1. (6.12) 2p 1,2p 2p+2 2p 1,2p+1 2 − − − ≥ Using (3.8), we simplify this relation: v A (2) v A (2) = 0, p 1. (6.13) 1 2p+2 2p+1 2 − ≥ From the equalities (2.5) and the designation (6.7) it follows that: A (2) = v k, k 2. (6.14) k k 1 − − ≥ Substituting (6.14) into (6.13), we get v = (p+1)v , p 1. (6.15) 2p+1 1 ≥

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.