Loading [MathJax]/jax/output/SVG/jax.js
Research article

Discrete quasiprobability distributions involving Bernoulli polynomials

  • Received: 20 December 2022 Revised: 27 February 2023 Accepted: 06 March 2023 Published: 30 March 2023
  • MSC : 60E05, 62E15

  • The aim of this short paper is to present a new family of discrete densities with two parameters based on Bernoulli numbers and polynomials. We use the properties of such numbers in order to compute the first moments and the density of a finite sum of such independent variables.

    Citation: Bander Almutairi. Discrete quasiprobability distributions involving Bernoulli polynomials[J]. AIMS Mathematics, 2023, 8(6): 12819-12829. doi: 10.3934/math.2023645

    Related Papers:

    [1] Aimin Xu . Some identities connecting Stirling numbers, central factorial numbers and higher-order Bernoulli polynomials. AIMS Mathematics, 2025, 10(2): 3197-3206. doi: 10.3934/math.2025148
    [2] Dojin Kim, Patcharee Wongsason, Jongkyum Kwon . Type 2 degenerate modified poly-Bernoulli polynomials arising from the degenerate poly-exponential functions. AIMS Mathematics, 2022, 7(6): 9716-9730. doi: 10.3934/math.2022541
    [3] Takao Komatsu, Wenpeng Zhang . Several expressions of truncated Bernoulli-Carlitz and truncated Cauchy-Carlitz numbers. AIMS Mathematics, 2020, 5(6): 5939-5954. doi: 10.3934/math.2020380
    [4] Feng Qi, Da-Wei Niu, Bai-Ni Guo . Simplifying coefficients in differential equations associated with higher order Bernoulli numbers of the second kind. AIMS Mathematics, 2019, 4(2): 170-175. doi: 10.3934/math.2019.2.170
    [5] Jizhen Yang, Yunpeng Wang . Congruences involving generalized Catalan numbers and Bernoulli numbers. AIMS Mathematics, 2023, 8(10): 24331-24344. doi: 10.3934/math.20231240
    [6] Waseem Ahmad Khan, Kottakkaran Sooppy Nisar, Dumitru Baleanu . A note on (p, q)-analogue type of Fubini numbers and polynomials. AIMS Mathematics, 2020, 5(3): 2743-2757. doi: 10.3934/math.2020177
    [7] Taekyun Kim, Dae San Kim, Hyunseok Lee, Lee-Chae Jang . A note on degenerate derangement polynomials and numbers. AIMS Mathematics, 2021, 6(6): 6469-6481. doi: 10.3934/math.2021380
    [8] Nadia N. Li, Wenchang Chu . Explicit formulae for Bernoulli numbers. AIMS Mathematics, 2024, 9(10): 28170-28194. doi: 10.3934/math.20241366
    [9] Waseem A. Khan, Abdulghani Muhyi, Rifaqat Ali, Khaled Ahmad Hassan Alzobydi, Manoj Singh, Praveen Agarwal . A new family of degenerate poly-Bernoulli polynomials of the second kind with its certain related properties. AIMS Mathematics, 2021, 6(11): 12680-12697. doi: 10.3934/math.2021731
    [10] Takao Komatsu, Ram Krishna Pandey . On hypergeometric Cauchy numbers of higher grade. AIMS Mathematics, 2021, 6(7): 6630-6646. doi: 10.3934/math.2021390
  • The aim of this short paper is to present a new family of discrete densities with two parameters based on Bernoulli numbers and polynomials. We use the properties of such numbers in order to compute the first moments and the density of a finite sum of such independent variables.



    The axioms of the probability theory in the sense of A. Kolmogorov required three conditions in order to define a rigorous notion of the probability measure on a measurable space. To this end we introduce a set Ω called the universe, and the sets of all possible events are encoded by a Borel σ-algebra B(Ω). Among these three classical conditions, is the requirement that a probability measure takes nonnegative values. If we drop-off this positivity condition, this gives rise to the notion of quasiprobability.

    Definition 1.1 (Quasiprobability). A quasiprobability measure on (Ω,B(Ω)) is a real valued measure defined on the Borel σ-algebra of Ω, in other words a map

    ˜P:B(Ω)R

    satisfying the conditions.

    (1) ˜P:(Ω)=1.

    (2) For any countable sequence A1,,An, of disjoint sets,

    ˜P(n=1An)=n1˜P(An).

    In the sequel, we will only consider discrete quasiprobabilities i.e. Ω=N. Given any discrete random variable X:ΩN which is distributed following a density fX i.e. fX(n)=˜P(X=n) then the condition (1) of the definition above just reads,

    n1fX(n)=1.

    The main issue is that fX(n) is now allowed to take negative values. This idea to relax the axioms of probabilities leading to the notion of negative probabilities has been already raised by P.A.M. Dirac and was also formulated in a more precise way by R. Feynmann. We have chosen to focus on is based on Bernoulli numbers and Bernoulli polynomials. These objects unexpectedly appear within the field of quantum statistical physics (see e.g. [14] § 2.3.1). Indeed, a central result in this theory is given by Planck's law of energy radiation of a black-body. It states that the density of energy radiation in function of the wave frequency ν at constant temperature T is given by the formula

    f(ν)=2hν3c21ehνkT1,

    where h the Planck constant, k the Boltzmann constant, and c the speed of light.

    Using Eq (4) we obtain a series expansion for f(ν) for some constant C independent of ν,

    f(ν)=Cn0Bnn!(hνkT)n+2,

    where (Bn)n0 is the sequence of Bernoulli numbers defined by the relation

    tetet1=n0Bntnn!. (1)

    The sum obtained is among all the energy microstates in the quantum formalism.

    Let us define for each n0 and a fixed ν, the function

    f(n;ν)=1f(ν)CBnn!(hνkT)n+2.

    By definition n0f(n;ν)=1 so that can interpret fn(ν) as a local density of energy radiation for a microstate at fixed frequency ν. This observation suggests that f(n,ν) defines a discrete density of probability. Unfortunately, a major obstacle is that Bernoulli numbers assume both positive and negative values, and therefore bringing us outside the field of the probability theory. Therefore this density can be seen as the distribution of a quasiprobability discrete variable in the sense of Definition 1.1 given above.

    The so-called Bernoulli polynomials and their related numbers arose in many parts in mathematics. Their definition is simply characterised by the relation:

    tetxet1=n0Bn(x)tnn!. (2)

    The series in the right hand side is entire in the open disc |t|<2π. These polynomials enjoy nice properties, an important one is the following differential equation

    Bn(x)=nBn1(x). (3)

    Thus Bn(x) defines a polynomial of degree n which can be computed by induction, the first Bernoulli polynomials are given by B0(x)=1, B1(x)=x12 and B2(x)=x2x+16. The Bernoulli polynomials have received considerable attention giving rise to a plethora of remarkable relations (see e.g. [12] Chap. 24). The Bernoulli numbers are just the constant terms of the Bernoulli polynomial i.e. Bn=Bn(0), accordingly the first terms are given by B0=1, B1=12 and B2=16. A remarkable fact is that B2n+1=0 for any integer n1. In some sense Bernoulli numbers are considered as much important as the polynomials.

    We introduce the main object which is a discrete quasiprobability distribution with two parameters.

    Definition 2.1. Given two real parameters r0 and θ(0;2π), we define a quasiprobability distribution X supported on N given by the density function,

    fX(n):=eθ1θeθrBn(r)θnn!.

    We denote by PB(r,θ) such distribution and we call it the poly-Bernoulli distribution with parameters r and θ.

    The prefix poly-simply means polynomial and it has nothing to do with the classical Bernoulli probability distributions. The fact that nNfX(n)=1 comes immediately from (4) and our definition involves two parameters θ and r. If we consider the case when r=0, then the distribution P(0,θ) has the property that only even integers contribute in our computation, in other words, fX(2k+1)=0 for any positive integer k. Moreover, using relation (8) we remark that the density fX assumes both negative and positive value and thus it defines a quasiprobability distribution in the sense of Definition 1.1.

    The use of special numbers and special functions in order to define new distribution is not new. Regarding random distribution involving special numbers one has the work of Kim et al. (see for example [5,6,7,10,11]). For another kind of random variables involving polynomials one has [8] for Dowling polynomials, [9] for Lah-Bell polynomials and [3] for derangements polynomials.

    We give the first moments of this distribution, the mean and the variance in the sense of quasiprobrobilities. We also focus on the case when r=0 which is seems already giving interesting relations regarding the distribution of the sum of such distributions.

    Bernoulli polynomials are defined by the generating series

    tetxet1=n0Bn(x)tnn!. (4)

    The corresponding Bernoulli numbers which are the constant coefficients of Bn(x), namely Bn=Bn(0) has generating series obtained by specialiazing the Eq (4) for x=0, thus

    tet1=n0Bntnn!. (5)

    Let us consider

    u(t)=tet1tet1=tet1tet1et=t(et+1)et1=n0Bn(1+(1)n)tnn!.

    Thus,

    t(et/2+et/2)et/2et/2=2n0B2nt2n(2n)!. (6)

    From (6), we can easily deduces the series expansion

    t2coth(t2)=k0B2kt2k(2k)!. (7)

    The even Bernoulli numbers coming into play in (11) are related to the Riemann-zeta function, indeed using basic Fourier analysis (see [2]), one can obtain that for every n1

    B2n(2n)!=(1)n+12ζ(2n)(2π)2n. (8)

    One has the following nice relation (e.g. 24.14.2 of [12])

    nk=0(nk)BkBnk=(1n)BnnBn1. (9)

    The following proposition provides the properties of PB-distributions.

    Proposition 3.1. Given any nonnegative integer n, and X a PB(0,θ) distribution. Then, one has the following properties,

    (1) fX(2n+1)=0.

    (2) fX(2n)=(1)n+1eθ1θ2ζ(2n)(θ2π)2n.

    (3) sgn (fX(2n))=(1)n+1.

    Proof. (1) This is an immediate consequence of the fact that B2n+1=0 for any positive integer n>0.

    (2) By definition,

    fX(2n):=eθ1θB2nθ2n(2n)!.

    Using (8), we have the required identity.

    (3) Since θ>0 and ζ(2n)>0, the previous identity in (2) proves that the sign of fX(2n) is given by (1)n+1.

    Let Y be a Poisson random variable of parameter θ>0 with density function fY (see e.g. [1] 20.7), and X with density f(n;r,θ), then we can write

    f(n;r,θ)=mn(r,θ)fY(n)wheremn(r,θ)=eθ1θeθ(r1)Bn(r).

    In other words, the densites of X and Y only differ up to a multiplicative factor depending on n,r,θ. For an adequacy with Poisson's distribution, one has to find the values of the parameters (r,θ) for which mn(r,θ)=1, n1. As we have done before we consider the case r=0, thus we get the factor mn(θ)=Bneθ(eθ1)/θ. For small values of θ we have that mn(θ)Bn so that fX(n) differs from a Poisson density P(θ), only by a factor given by Bn. To sum up, our distribution in the case where r=0 satisfies the following asymptotic property with respect to the Poisson distribution,

    limθ0+fX(n)fY(n)=Bn.

    Let us simply denote fθ(n) instead of f(n;0,θ). As we have seen above the density fθ is supported only at even nonnegative integers,

    fθ(2k)=(eθ1)B2k(2k)!θ2k1.

    Then by using Eq (8) we infer that

    fθ(2k)=(1)k+12ζ(2k)(2π)2kθ2k1(eθ1).

    In particular the cumulative distribution function FX(x)=nxfθ(n) is an alternating series. Since limkζ(2k)=1+n2limkn2k=1, we get

    |fθ(2k)|2(2k)!(2π)2kθ2k1(eθ1)ask.

    Using Striling's formula we obtain the following asymptotic estimate for |fθ(2k)|.

    Proposition 3.2. For any |θ|<2π, we have

    |fX(2n)|(nπe)2n16nθ2n1(eθ1)asn.

    The evaluation of the expectation value of X requires to use specific properties of Bernoulli polynomials which are given in section §2. We first prove the following useful lemma.

    Lemma 3.3. We have the following relation

    rf(n;r,θ)=θ(f(n1;r,θ)f(n;r,θ)).

    Proof. Using Definition 2.1 and Eq (3) we have,

    rf(n;r,θ)=(Bn(r)eθrθBn(r)erθe2θr)(eθ1)θn1n!
    =(nBn1(r)θBn(r)eθr)(eθ1)θn1n!=eθ1erθ(nBn1(r)θn1n!θBn(r)θn1n!)
    =eθ1erθ(Bn1(r)θn1(n1)!θBn(r)θn1n!)=eθ1erθθ(Bn1(r)θn2(n1)!Bn(r)θn1n!).

    Hence

    rf(n;r,θ)=θ(f(n1;r,θ)f(n;r,θ)).

    This proves the lemma.

    Theorem 3.4. Let X be a random variable with density distribution f(n;r,θ), then the mean of X is given by

    E(X)=θr+1θ1eθ.

    Proof. Let us compute the derivative of the mean of X with respect to the parameter r using the previous lemma, for convenience we write f(n) instead of f(n;r,θ),

    rE(X)=rn1nP(X=n)=n1nrf(n)=θn1n(f(n1)f(n))
    =θf(0)+θn1(n+1)f(n)nf(n)=θf(0)+θn1f(n)=θn0f(n).

    Thus we get rE(X)=θ. Now we perform an integration wrt r and we get that

    E(X)=θr+E(X)|r=0.

    Let us explicit the term E(X)|r=0 which denotes the evaluation of E(X) when r=0.

    E(X)|r=0=n1nf(n;0,θ)=(eθ1)n1Bnθn1(n1)!

    and since Bn=0 for any odd integer n>1 and B1=1/2, we obtain that

    E(X)|r=0=12(eθ1)+(eθ1)k1B2kθ2k1(2k1)!. (10)

    From (4), we have the series expansion

    θ2coth(θ2)=k0B2kθ2k(2k)!. (11)

    The differentiation of Eq (11) with respect to θ yields,

    k1B2kθ2k1(2k1)!=θ(θ2coth(θ2))=θ(θ(eθ+1)2(eθ1))=12+1eθ1θeθ(eθ1)2.

    By replacing in (10) we obtain

    E(X)|r=0=12(eθ1)+(eθ1)(12+1eθ1θeθ(eθ1)2).

    Hence we finally obtain E(X)|r=0=1θ1eθ and the proof follows.

    Proposition 3.5. Let be given sR such that s0. Then the probability generating function

    gX(s)=se(s1)θreθ1esθ1.

    Proof.

    gX(s)=n0snP(X=n)=eθ1θeθrn0snBn(r)θnn!=eθ1θeθrn0Bn(r)(sθ)nn!.

    By using (4) we obtain the required result

    gX(s)=eθ1θeθrsθesθresθ1=serθ(s1)eθ1esθ1.

    Actually we can use this formula to recover the expectation value of X, although the next proof below is in our opinion less instructive than the first one. Indeed let us consider the log-derivative of gX(s),

    gX(s)gX(s)=1s+θrθeθseθs1. (12)

    Taking the limit when s1, we recover immediately the expectation value of X,

    E(X)=1+θrθeθeθ1.

    If we proceed to the differentiation of (12) we get,

    gX(s)gX(s)gX(s)2gX(s)2=1s2dds(θ1eθs)=1s2+θ2eθs(1eθs)2. (13)

    In particular if we let s1 in Eq (13) we get

    gX(1)gX(1)2=1+θ2eθ(1eθ)2.

    Reminding that Var(X)=gX(1)gX(1)2+E(X), one obtains the following result

    Corollary 3.6. The variance of X is given by,

    Var(X)=θ(r+(θ1)eθ1(1eθ)2).

    The density of the sum of two independent and identically distributed (i.i.d.) variables in the general case, leads to a quite complicated expression. In this section we assume that r=0. In that case we are able to find a closed form for the distribution of the sum of a finte number of PB(θ)-distribution. Similary to probability theory, one defines the notion of independence of two quasiprobabilities PB(θ)-distributions X and Y, we say that X and Y are i.i.d. if X,YPB(θ) and fX+Y=fXfY as usual. This definition applies analogously to finite sums with more than two variables.

    Proposition 4.1. Let X and Y two i.i.d. random variables following the distribution PB(θ). Then the law of the sum is given by

    fX+Y(n)=(1eθθ)((n1)f(n)+θf(n1)).

    Proof. By independence,

    fX+Y(n)=fXfY(n)=nk=0˜P(X=k)˜P(Y=nk)
    =(eθ1θ)2nk=0Bkθkk!Bnkθnk(nk)!
    =(eθ1θ)2(nk=0(nk)BkBnk)θnn!.

    Now we use the well-known formula 9. Therefore we get

    fX+Y(n)=(eθ1θ)2((1n)BnnBn1)θnn!
    =(eθ1θ)2(1n)Bnθnn!(eθ1)2θBn1θn1(n1)!
    =(eθ1θ)((1n)f(n)θf(n1)).

    In order to generalize the previous result we need the following Lemma

    Lemma 4.2 (Vandiver).

    k1++kn=kk!k1!kn!Bk1Bkn=(1)n1(kn)n1i=0[nni]Bki

    where the numbers [np] are unsigned Stirling numbers of the first kind defined by the generating function x(x+1)(x+n1)=np=0[np]xp.

    Proof of the Lemma. See [13] Eq (140) or e.g. [4] Eq (1.5).

    The distribution of the the sum of n i.i.d. with density PB(θ) is given by the following formula

    Theorem 4.3. For any integer kn, we have

    fSn(k)=(1eθθ)n(kn)θkk!ik[2][nni](1)(ki)/2+1(ki)!ζ(ki)(2π)ki.

    We have

    Proof.

    fSn(k)=fX1fXn(k)=k1++kn=kfX1(k1)fXn(kn)
    =(eθ1θ)nk1++kn=kBk1θk1k1!Bknθknkn!
    =(eθ1θ)n(k1++kn=kk!k1!kn!Bk1Bkn)θkk!.

    By Lemma 4.2 we get,

    fSn(k)=(eθ1θ)n((1)n1(kn)n1i=0[nni]Bki)θkk!
    =(eθ1θ)n(1)n1(kn)n1i=0[nni]Bkiθki(ki)!θik(k1)(ki+1).

    Hence,

    fSn(k)=(1eθθ)n1(kn)n1i=0[nni]θik(k1)(ki+1)f(ki).

    As we know the densities f(n) are supported by nonnegative even integers thus the expression of the density of the sum takes the following accurate form

    fSn(k)=(1eθθ)n1(kn)ik[2][nni]θik(k1)(ki+1)f(ki).

    Using the relation f(2k)=(1)k+12ζ(2k)(2π)2kθ2k1(eθ1) we get

    fSn(k)=(1eθθ)n1(kn)ik[2][nni](1)(ki)/2+1θiθki1k(k1)(ki+1)ζ(ki)(2π)ki(eθ1)
    =(1eθθ)n(kn)θkik[2][nni](1)(ki)/2+11k(k1)(ki+1)ζ(ki)(2π)ki.

    Hence we obtain

    fSn(k)=(1eθθ)n(kn)θkk!ik[2][nni](1)(ki)/2+1(ki)!ζ(ki)(2π)ki.

    We have introduced a new class of random distributions which are have total mass one but are not necessarily nonnegative. These quasiprobability distributions are based on Bernoulli polynomials. One feature of this distribution is that it does not charge odd positive integers. Due to the numerous relations involving Bernoulli polynomials, one is able to compute the expected value and the density of the sum of independent identical PB-distributions. Also, we have obtained an asymptotic estimate of the density as n tends to infinity. That such distribution can serve as a model for discrete statistical distribution which charges only even integers.

    This research is supported by Researchers Supporting Project number (RSPD2023R650), King Saud University, Riyadh, Saudi Arabia.

    The author declares no conflict of interest.



    [1] P. Billingsley, Probability and measure, New York: John Wiley and Sons, 1995.
    [2] R. P. Brent, D. Harvey, Fast computation of Bernoulli, Tangent and Secants numbers, Comput. Anal. Math., 50 (2013), 127–142.
    [3] L. C. Jang, D. S. Kim, T. Kim, H. Lee, Some identities involving derangement polynomials and numbers and moments of Gamma random variables, J. Funct. Spaces, 2020 (2020), 6624006. https://doi.org/10.1155/2020/6624006 doi: 10.1155/2020/6624006
    [4] K. Kamano, Sums of product of hypergeometric Bernoulli numbers, J. Number Theory, 130 (2010), 2259–2271. http://dx.doi.org/10.1016/j.jnt.2010.04.005 doi: 10.1016/j.jnt.2010.04.005
    [5] T. Kim, Y. Yao, D. S. Kim, H. I. Kwon, Some identities involving special numbers and moments of random variables, Rocky Moutain J. Math., 49 (2019), 521–538. http://dx.doi.org/10.1216/RMJ-2019-49-2-521 doi: 10.1216/RMJ-2019-49-2-521
    [6] T. Kim, D. S. Kim, H. Lee, S. H. Park, Dimorphic properties of Bernoulli random variable, Filomat, 36 (2022), 1711–1717.
    [7] T. Kim, D. S. Kim, Degenerate zero-truncated Poisson random variables, Russ. J. Math. Phys., 28 (2021), 66–72. https://doi.org/10.1134/S1061920821010076 doi: 10.1134/S1061920821010076
    [8] T. Kim, D. S. Kim, H. K. Kim, Poisson degenerate central moments related to degenerate Dowling and degenerate r-Dowling polynomials, Appl. Math. Sci. Eng., 30 (2022), 583–597. https://doi.org/10.1080/27690911.2022.2118736 doi: 10.1080/27690911.2022.2118736
    [9] T. Kim, D. S. Kim, D. V. Dolgy, J. W. Park, Degenerate binomial and Poisson random variables associated with degenerate Lah-Bell polynomials, Open Math., 19 (2021), 1588–1597. https://doi.org/10.1515/math-2021-0116 doi: 10.1515/math-2021-0116
    [10] T. Kim, D. S. Kim, Some identities on Degenerate r-Striling numbers via Boson operators, Russ. J. Math. Phys., 29 (2022), 508–517.
    [11] T. Kim, D. S. Kim, L. C. Jang, H. Y. Kim, A note on discrete degenerate random variables, Proc. Jangjeon Math. Soc., 23 (2020), 125–135.
    [12] F. W. J. Olver, D. W. Lozier, R. F. Boisvert, C. W. Clark, NIST handbook of mathematical functions, Cambridge: Cambridge University Press, 2010.
    [13] H. S. Vandiver, An arithmetical theory of the Bernoulli numbers, Trans. Amer. Math. Soc., 51, (1942), 502–531.
    [14] E. Zeidler, Quantum field theory: a bridge between mathematicians and physicists, Berlin: Springer, 2009.
  • Reader Comments
  • © 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(1307) PDF downloads(60) Cited by(0)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog