Loading [MathJax]/jax/element/mml/optable/BasicLatin.js
Research article

An infinite semipositone problem with a reversed S-shaped bifurcation curve

  • We study positive solutions to the two point boundary value problem:

    Lu=u=λ{Auγ+M[uα+uδ]};(0,1)u(0)=0=u(1)

    where A<0, α(0,1),δ>1,γ(0,1) are constants and λ>0,M>0 are parameters. We prove that the bifurcation diagram (λ vs u) for positive solutions is at least a reversed S-shaped curve when M1. Recent results in the literature imply that for M1 there exists a range of λ where there exist at least two positive solutions. Here, when M1, we prove the existence of a range of λ for which there exist at least three positive solutions and that the bifurcation diagram is at least a reversed S-shaped curve. Further, via a quadrature method and Python computations, for M1, we show that the bifurcation diagram is exactly a reversed S-shaped curve. Also, when the operator L is replaced by a p-Laplacian operator with p>1, as well as p-q Laplacian operator with p=4 and q=2, we show that the bifurcation diagram is again an exactly reversed S-shaped curve when M1.

    Citation: Amila Muthunayake, Cac Phan, Ratnasingham Shivaji. An infinite semipositone problem with a reversed S-shaped bifurcation curve[J]. Electronic Research Archive, 2023, 31(2): 1147-1156. doi: 10.3934/era.2023058

    Related Papers:

    [1] Jiaye Lin . Evaluations of some Euler-type series via powers of the arcsin function. AIMS Mathematics, 2025, 10(4): 8116-8130. doi: 10.3934/math.2025372
    [2] 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
    [3] Waleed Mohamed Abd-Elhameed, Omar Mazen Alqubori, Ahmed Gamal Atta . A collocation procedure for the numerical treatment of FitzHugh–Nagumo equation using a kind of Chebyshev polynomials. AIMS Mathematics, 2025, 10(1): 1201-1223. doi: 10.3934/math.2025057
    [4] Taekyun Kim, Dae San Kim, Dmitry V. Dolgy, Hye Kyung Kim, Hyunseok Lee . A new approach to Bell and poly-Bell numbers and polynomials. AIMS Mathematics, 2022, 7(3): 4004-4016. doi: 10.3934/math.2022221
    [5] Jongkyum Kwon, Patcharee Wongsason, Yunjae Kim, Dojin Kim . Representations of modified type 2 degenerate poly-Bernoulli polynomials. AIMS Mathematics, 2022, 7(6): 11443-11463. doi: 10.3934/math.2022638
    [6] Nusrat Raza, Mohammed Fadel, Kottakkaran Sooppy Nisar, M. Zakarya . On 2-variable q-Hermite polynomials. AIMS Mathematics, 2021, 6(8): 8705-8727. doi: 10.3934/math.2021506
    [7] Qiuxia Hu, Bilal Khan, Serkan Araci, Mehmet Acikgoz . New double-sum expansions for certain Mock theta functions. AIMS Mathematics, 2022, 7(9): 17225-17235. doi: 10.3934/math.2022948
    [8] Gamal Hassan, Mohra Zayed . Expansions of generalized bases constructed via Hasse derivative operator in Clifford analysis. AIMS Mathematics, 2023, 8(11): 26115-26133. doi: 10.3934/math.20231331
    [9] Mohra Zayed, Taghreed Alqurashi, Shahid Ahmad Wani, Cheon Seoung Ryoo, William Ramírez . Several characterizations of bivariate quantum-Hermite-Appell Polynomials and the structure of their zeros. AIMS Mathematics, 2025, 10(5): 11184-11207. doi: 10.3934/math.2025507
    [10] Chunli Li, Wenchang Chu . Remarkable series concerning (3nn) and harmonic numbers in numerators. AIMS Mathematics, 2024, 9(7): 17234-17258. doi: 10.3934/math.2024837
  • We study positive solutions to the two point boundary value problem:

    Lu=u=λ{Auγ+M[uα+uδ]};(0,1)u(0)=0=u(1)

    where A<0, α(0,1),δ>1,γ(0,1) are constants and λ>0,M>0 are parameters. We prove that the bifurcation diagram (λ vs u) for positive solutions is at least a reversed S-shaped curve when M1. Recent results in the literature imply that for M1 there exists a range of λ where there exist at least two positive solutions. Here, when M1, we prove the existence of a range of λ for which there exist at least three positive solutions and that the bifurcation diagram is at least a reversed S-shaped curve. Further, via a quadrature method and Python computations, for M1, we show that the bifurcation diagram is exactly a reversed S-shaped curve. Also, when the operator L is replaced by a p-Laplacian operator with p>1, as well as p-q Laplacian operator with p=4 and q=2, we show that the bifurcation diagram is again an exactly reversed S-shaped curve when M1.



    In [12,Definition 11.2] and [18,p. 134,Theorem A], the second kind Bell polynomials Bn,k for nk0 are defined by

    Bn,k(x1,x2,,xnk+1)=Nnk+10n!nk+1i=1i!nk+1i=1(xii!)i,

    where N0={0}N, the sum is taken over =(1,2,,nk+1) with iN0 satisfying nk+1i=1i=k and nk+1i=1ii=n. This kind of polynomials are very important in combinatorics, analysis, and the like. See the review and survey article [53] and closely related references therein.

    In [36,pp. 13–15], when studying Grothendieck's inequality and completely correlation-preserving functions, Oertel obtained the interesting identity

    2nk=1(1)k(2n+k)!k!B2n,k(0,16,0,340,0,5112,,1+(1)k+12[(2nk)!!]2(2nk+2)!)=(1)n

    for nN, where

    Bn,k(x1,x2,,xnk+1)=k!n!Bn,k(1!x1,2!x2,,(nk+1)!xnk+1). (1.1)

    In [36,p. 15], Oertel wrote that "However, already in this case we don't know a closed form expression for the numbers

    B2n,k(0,16,0,340,0,5112,,1+(1)k+12[(2nk)!!]2(2nk+2)!). (1.2)

    An even stronger problem appears in the complex case, since already a closed-form formula for the coefficients of the Taylor series of the inverse of the Haagerup function is still unknown''.

    By virtue of the relation (1.1), we see that, to find a closed-form formula for the sequence (1.2), it suffices to discover a closed-form formula for

    B2n,k(0,13,0,95,0,2257,,1+(1)k+12[(2nk)!!]22nk+2). (1.3)

    In this paper, one of our aims is to derive closed-form formulas for the sequence (1.3). The first main result can be stated as the following theorem.

    Theorem 1.1. For k,n0, mN, and xmC, we have

    B2n+1,k(0,x2,0,x4,,1+(1)k2x2nk+2)=0. (1.4)

    For k,nN, we have

    B2n,2k1(0,13,0,95,0,2257,,0,[(2n2k+1)!!]22n2k+3)=22n(2k1)![kp=1(4)p1(2k12p1)(2n+2p12p1)2p2q=0T(n+p1;q,2p2;12)k1p=1(1)p1(2k12p)(2n+2p2p)2p2q=0T(n+p1;q,2p2;1)]

    and

    B2n,2k(0,13,0,95,0,2257,,[(2n2k1)!!]22n2k+1,0)=22n(2k)![kp=1(1)p1(2k2p)(2n+2p2p)2p2q=0T(n+p1;q,2p2;1)kp=1(4)p1(2k2p1)(2n+2p12p1)2p2q=0T(n+p1;q,2p2;12)],

    where s(n,k), which can be generated by

    xn=nm=0s(n,m)xm, (1.5)

    denote the first kind Stirling numbers and

    T(r;q,j;ρ)=(1)q[rm=q(ρ)ms(r,m)(mq)][rm=jq(ρ)ms(r,m)(mjq)]. (1.6)

    In Section 2, for proving Theorem 1.1, we will establish two general expressions for power series expansions of (arcsinx)21 and (arcsinx)2 respectively.

    In Section 3, with the aid of general expressions for power series expansions of the functions (arcsinx)21 and (arcsinx)2 established in Section 2, we will prove Theorem 1.1 in details.

    In Section 4, basing on arguments in [20,p. 308] and [28,Section 2.4] and utilizing general expressions for power series expansions of (arcsinx)21 and (arcsinx)2 established in Section 2, we will derive series representations of generalized logsine functions which were originally introduced in [34] and have been investigating actively, deeply, and systematically by mathematicians [9,10,14,15,16,17,29,30,31,37,38,57] and physicists [3,19,20,28].

    Finally, in Section 5, we will list several remarks on our main results and related stuffs.

    To prove Theorem 1.1, we need to establish the following general expressions of the power series expansions of (arcsinx) for N.

    Theorem 2.1. For N and |x|<1, the functions (arcsinx) can be expanded into power series

    (arcsinx)21=(4)1n=04n(2n)![22q=0T(n+1;q,22;12)]x2n+21(2n+2121) (2.1)

    or

    (arcsinx)2=(1)1n=04n(2n)![22q=0T(n+1;q,22;1)]x2n+2(2n+22), (2.2)

    where s(n,k) denotes the first kind Stirling numbers generated in (1.5) and T(r;q,j;ρ) is defined by (1.6).

    Proof. In [4,pp. 262–263,Proposition 15], [7,p. 3], [20,p. 308], and [28,pp. 49–50], it was stated that the generating expression for the series expansion of (arcsinx)n with nN is

    exp(tarcsinx)==0b(t)x!,

    where b0(t)=1, b1(t)=t, and

    b2(t)=1k=0[t2+(2k)2],b2+1(t)=tk=1[t2+(2k1)2]

    for N. This means that, when writing

    b(t)=k=0β,ktk,0,

    where β0,0=1, β2,0=0, β2,2k+1=0, and β21,2k=0 for k0 and 1, we have

    =0(arcsinx)t!==0x!k=0β,ktk=k=0=kx!β,ktk==0[m=βm,xmm!]t.

    Equating coefficients of t gives

    (arcsinx)=!m=βm,xmm!=!n=0βn+,xn+(n+)!,N. (2.3)

    It is not difficult to see that

    b2(t)=41t2(1it2)1(1+it2)1andb2+1(t)=4t(12it2)(12+it2),

    where i=1 is the imaginary unit and

    (z)n=n1=0(z+)={z(z+1)(z+n1),n11,n=0

    is called the rising factorial of zC, while

    zn=n1=0(z)={z(z1)(zn+1),n11,n=0 (2.4)

    is called the falling factorial of zC. Making use of the relation

    (z)n=(1)nznorzn=(1)n(z)n

    in [52,p. 167], we acquire

    b2(t)=41t2it211it211andb2+1(t)=4tit212it212.

    Utilizing the relation (1.5) in [59,p. 19,(1.26)], we obtain

    b2(t)=41t21m=0s(1,m)2m(it2)m1m=0(1)ms(1,m)2m(it+2)m=41t21m=0s(1,m)2mmk=0(mk)iktk(2)mk1m=0(1)ms(1,m)2mmk=0(mk)iktk2mk=41t21m=0(1)ms(1,m)mk=0(1)k2k(mk)iktk1m=0(1)ms(1,m)mk=012k(mk)iktk=41t21k=0[1m=k(1)m+ks(1,m)2k(mk)]iktk1k=0[1m=k(1)ms(1,m)2k(mk)]iktk=41t22(1)k=0kq=0[1m=q(1)m+qs(1,m)2q(mq)1m=kq(1)ms(1,m)2kq(mkq)]iktk=41t22(1)k=012kkq=0[1m=q(1)m+qs(1,m)(mq)1m=kq(1)ms(1,m)(mkq)]iktk=412(1)k=0ik2k[kq=0(1m=q(1)ms(1,m)(mq))1m=kq(1)ms(1,m)(mkq)]tk+2=412(1)k=0ik2k[kq=0T(1;q,k;1)]tk+2

    and

    b2+1(t)=4tm=0s(,m)2m(it1)mm=0(1)ms(,m)2m(it+1)m=4tm=0s(,m)2mmk=0(1)mk(mk)iktkm=0(1)ms(,m)2mmk=0(mk)iktk=4tk=0[m=k(1)ms(,m)2m(mk)](i)ktkk=0[m=k(1)ms(,m)2m(mk)]iktk=42k=0ik[kq=0(1)q(m=q(1)ms(,m)2m(mq))m=kq(1)ms(,m)2m(mkq)]tk+1=42k=0ik[kq=0T(;q,k;12)]tk+1.

    This means that

    2k=0β2,ktk=2(1)k=2β2,k+2tk+2=2(1)k=0β2,k+2tk+2=412(1)k=0ik2k[kq=0T(1;q,k;1)]tk+2

    and

    2+1k=0β2+1,ktk=2k=1β2+1,k+1tk+1=2k=0β2+1,k+1tk+1=42k=0ik[kq=0T(;q,k;12)]tk+1.

    Further equating coefficients of tk+2 and tk+1 respectively arrives at

    β2,k+2=41ik2kkq=0T(1;q,k;1)andβ2+1,k+1=4ikkq=0T(;q,k;12)

    for k0.

    Replacing by 21 for N in (2.3) leads to

    (arcsinx)21=(21)!n=0βn+21,21xn+21(n+21)!=(21)!n=0β2n+21,21x2n+21(2n+21)!=(21)!n=0[4n+1i2(1)2(1)q=0T(n+1;q,22;12)]x2n+21(2n+21)!=(1)141(21)!n=0[4n2(1)q=0T(n+1;q,22;12)]x2n+21(2n+21)!=(4)1n=04n(2n)![22q=0T(n+1;q,22;12)]x2n+21(2n+2121).

    Replacing by 2 for N in (2.3) leads to

    (arcsinx)2=(2)!n=0βn+2,2xn+2(n+2)!=(2)!n=0β2n+2,2x2n+2(2n+2)!=(1)1(2)!n=0[4n2(1)q=0T(n+1;q,22;1)]x2n+2(2n+2)!=(1)1n=04n(2n)![22q=0T(n+1;q,22;1)]x2n+2(2n+22).

    The proof of Theorem 2.1 is complete.

    We now start out to prove Theorem 1.1.

    In the last line of [18,p. 133], there exists the formula

    1k!(m=1xmtmm!)k=n=kBn,k(x1,x2,,xnk+1)tnn! (3.1)

    for k0. When taking x2m1=0 for mN, the left hand side of the formula (3.1) is even in t(,) for all k0. Therefore, the formula (1.4) is valid.

    Ones know that the power series expansion

    arcsint==0[(21)!!]2(2+1)!t2+1,|t|<1 (3.2)

    is valid, where (1)!!=1. This implies that

    B2n,k(0,13,0,95,0,2257,,1+(1)k+12[(2nk)!!]22nk+2)=B2n,k((arcsint)|t=02,(arcsint)|t=03,(arcsint)(4)|t=04,,(arcsint)(2nk+2)|t=02nk+2).

    Employing the formula

    Bn,k(x22,x33,,xnk+2nk+2)=n!(n+k)!Bn+k,k(0,x2,x3,,xn+1)

    in [18,p. 136], we derive

    B2n,k(0,13,0,95,0,2257,,1+(1)k+12[(2nk)!!]22nk+2)=(2n)!(2n+k)!B2n+k,k(0,(arcsint)|t=0,(arcsint)|t=0,,(arcsint)(2n+1)|t=0).

    Making use of the formula (3.1) yields

    n=0Bn+k,k(x1,x2,,xn+1)k!n!(n+k)!tn+kn!=(m=1xmtmm!)k,n=0Bn+k,k(x1,x2,,xn+1)(n+kk)tn+kn!=(m=1xmtmm!)k,Bn+k,k(x1,x2,,xn+1)=(n+kk)limt0dndtn[m=0xm+1tm(m+1)!]k,B2n+k,k(x1,x2,,x2n+1)=(2n+kk)limt0d2ndt2n[m=0xm+1tm(m+1)!]k.

    Setting x1=0 and xm=(arcsint)(m)|t=0 for m2 gives

    d2ndt2n[m=0xm+1tm(m+1)!]k=d2ndt2n[1tm=2(arcsint)(m)|t=0tmm!]k=d2ndt2n(arcsinttt)k=d2ndt2nkp=0(1)kp(kp)(arcsintt)p=kp=1(1)kp(kp)d2ndt2n(arcsintt)p.

    Accordingly, we obtain

    limt0d2ndt2n[1tm=2(arcsint)(m)|t=0tmm!]2k1=2k1p=1(1)2kp1(2k1p)limt0d2ndt2n(arcsintt)p=kp=1(2k12p1)limt0d2ndt2n(arcsintt)2p1k1p=1(2k12p)limt0d2ndt2n(arcsintt)2p

    and

    limt0d2ndt2n[1tm=2(arcsint)(m)|t=0tmm!]2k=2kp=1(1)2kp(2kp)limt0d2ndt2n(arcsintt)p=kp=1(2k2p)limt0d2ndt2n(arcsintt)2pkp=1(2k2p1)limt0d2ndt2n(arcsintt)2p1.

    From the power series expansions (2.1) and (2.2) in Theorem 2.1, it follows that

    limt0d2ndt2n(arcsintt)2p1=(1)p14p1(2p1)!×limt0d2ndt2nj=0[4j2p2q=0T(j+p1;q,2p2;12)]t2j(2j+2p1)!=(1)p14n+p1(2n+2p12n)2p2q=0T(n+p1;q,2p2;12)

    and

    limt0d2ndt2n(arcsintt)2p=(1)p1(2p)!limt0d2ndt2nj=0[4j2p2q=0T(j+p1;q,2p2;1)]t2j(2j+2p)!=(1)p14n(2n+2p2n)2p2q=0T(n+p1;q,2p2;1).

    Therefore, we arrive at

    limt0d2ndt2n[1tm=2(arcsint)(m)|t=0tmm!]2k1=4nkp=1(4)p1(2k12p1)(2n+2p12p1)2p2q=0T(n+p1;q,2p2;12)4nk1p=1(1)p1(2k12p)(2n+2p2p)2p2q=0T(n+p1;q,2p2;1)

    and

    limt0d2ndt2n[1tm=2(arcsint)(m)|t=0tmm!]2k=4nkp=1(1)p1(2k2p)(2n+2p2p)2p2q=0T(n+p1;q,2p2;1)4nkp=1(4)p1(2k2p1)(2n+2p12p1)2p2q=0T(n+p1;q,2p2;12).

    Consequently, we acquire

    B2n,2k1(0,13,0,95,0,2257,,0,[(2n2k+1)!!]22n2k+3)=(2n)!(2n+2k1)!B2n+2k1,2k1(0,(arcsint)|t=0,(arcsint)|t=0,,(arcsint)(2n+1)|t=0)=(2n)!(2n+2k1)!(2n+2k12k1)limt0d2ndt2n(1tm=2(arcsint)(m)|t=0tmm!)2k1=1(2k1)![4nkp=1(4)p1(2k12p1)(2n+2p12p1)2p2q=0T(n+p1;q,2p2;12)4nk1p=0(1)p1(2k12p)(2n+2p2p)2p2q=0T(n+p1;q,2p2;1)]

    and

    B2n,2k(0,13,0,95,0,2257,,[(2n2k1)!!]22n2k+1,0)=(2n)!(2n+2k)!B2n+2k,2k(0,(arcsint)|t=0,(arcsint)|t=0,,(arcsint)(2n+1)|t=0)=(2n)!(2n+2k)!(2n+2k2k)limt0d2ndt2n(1tm=2(arcsint)(m)|t=0tmm!)2k=1(2k)![4nkp=1(1)p1(2k2p)(2n+2p2p)2p2q=0T(n+p1;q,2p2;1)4nkp=1(4)p1(2k2p1)(2n+2p12p1)2p2q=0T(n+p1;q,2p2;12)].

    The proof of Theorem 1.1 is complete.

    The logsine function

    Lsj(θ)=θ0(ln|2sinx2|)j1dx

    and generalized logsine function

    Ls()j(θ)=θ0x(ln|2sinx2|)j1dx

    were introduced originally in [34,pp. 191–192], where ,j are integers, j+11, and θ is an arbitrary real number. There have been many papers such as [3,9,10,14,15,16,17,19,20,28,29,30,31,37,38,57] devoted to investigation and applications of the (generalized) logsine functions in mathematics, physics, engineering, and other mathematical sciences.

    Theorem 4.1. Let zn for zC and n{0}N denote the falling factorial defined by (2.4) and let T(r;q,j;ρ) be defined by (1.6). In the region 0<θπ and for j,N, generalized logsine functions Ls()j(θ) have the following series representations:

    1. for j2+13,

    Ls(21)j(θ)=θ22[ln(2sinθ2)]j2(1)(j2)(21)!(ln2)j1(2sinθ2ln2)2×n=0(2sinθ2)2n(2n+2)![22q=0T(n+1;q,22;1)]×[j21α=0(lnsinθ2ln2)α(j21α)αk=0(1)kαk(2n+2)k+1(lnsinθ2)k]; (4.1)

    2. for j2+24,

    Ls(2)j(θ)=θ2+12+1[ln(2sinθ2)]j21+(1)(j21)(2)!(ln2)j12(4sinθ2ln2)2+1×n=0[(2sinθ2)2n(2n+2+1)!2q=0T(n+;q,2;12)]×[j22α=0(j22α)(lnsinθ2ln2)ααk=0(1)kαk(2n+2+1)k+1(lnsinθ2)k]; (4.2)

    3. for j211,

    Ls(22)j(θ)=(1)243(22)!(ln2)j(sinθ2ln2)21×n=0[(2sinθ2)2n(2n+22)!22q=0T(n+1;q,22;12)]×j2+1α=0(j2+1α)(lnsinθ2ln2)ααk=0(1)kαk(2n+21)k+1(lnsinθ2)k; (4.3)

    4. for j211,

    Ls(21)j(θ)=(1)(21)!(ln2)j(2sinθ2ln2)2×n=0[(2sinθ2)2n(2n+21)!22q=0T(n+1;q,22;1)]×j2α=0(j2α)(lnsinθ2ln2)ααk=0(1)kαk(2n+2)k+1(lnsinθ2)k. (4.4)

    Proof. In [28,p. 49,Section 2.4], it was obtained that

    Ls(k)j(θ)=θk+1k+1[ln(2sinθ2)]jk1+2k+1(jk1)k+1sin(θ/2)0(arcsinx)k+1lnjk2(2x)xdx (4.5)

    for 0<θπ and jk20. Making use of Theorem 2.1 and the formula

    xnlnmxdx=xn+1mk=0(1)kmklnmkx(n+1)k+1,m,n0 (4.6)

    in [22,p. 238,2.722], we acquire

    sin(θ/2)0(arcsinx)2lnj21(2x)xdx=(1)1(2)!n=04n(2n+2)![22q=0T(n+1;q,22;1)]sin(θ/2)0x2n+21lnj21(2x)dx=(1)1(2)!n=04n(2n+2)![22q=0T(n+1;q,22;1)]×[sin(θ/2)0x2n+21(ln2+lnx)j21dx]=(1)1(2)!n=04n(2n+2)![22q=0T(n+1;q,22;1)]×[j21α=0(j21α)(ln2)j2α1sin(θ/2)0x2n+21(lnx)αdx]=(1)1(2)!n=04n(2n+2)![22q=0T(n+1;q,22;1)]×[j21α=0(j21α)(ln2)j2α1(sinθ2)2n+2αk=0(1)kαk(2n+2)k+1(lnsinθ2)αk]=(1)1(2)!(ln2)j21(sinθ2)2n=04n(2n+2)!(sinθ2)2n[22q=0T(n+1;q,22;1)]×[j21α=0(lnsinθ2ln2)α(j21α)αk=0(1)kαk(2n+2)k+1(lnsinθ2)k]

    for j2+13. Substituting this result into (4.5) for k=21 yields (4.1).

    Similarly, by virtue of Theorem 2.1 and the formula (4.6), we also have

    sin(θ/2)0(arcsinx)2+1lnj22(2x)xdx=(1)4(2+1)!n=0[4n(2n+2+1)!2q=0T(n+;q,2;12)]sin(θ/2)0x2n+2lnj22(2x)dx=(1)4(2+1)!n=0[4n(2n+2+1)!2q=0T(n+;q,2;12)]×j22α=0(j22α)(ln2)j2α2sin(θ/2)0x2n+2(lnx)αdx=(1)4(2+1)!n=0[4n(2n+2+1)!2q=0T(n+;q,2;12)]×j22α=0(j22α)(ln2)j2α2(sinθ2)2n+2+1αk=0(1)kαk(lnsinθ2)αk(2n+2+1)k+1=(1)4(2+1)!(sinθ2)2+1(ln2)j22n=0[4n(2n+2+1)!(sinθ2)2n2q=0T(n+;q,2;12)]×[j22α=0(j22α)(lnsinθ2ln2)ααk=0(1)kαk(2n+2+1)k+1(lnsinθ2)k]

    for N and j2(+1)4. Substituting this result into (4.5) for k=2 yields (4.2).

    In [20,p. 308], it was derived that

    Ls(k)j(θ)=2k+1sin(θ/2)0(arcsinx)k1x2lnjk1(2x)dx (4.7)

    for 0<θπ and jk+11. Differentiating with respect to x on both sides of the formulas (2.1) and (2.2) in Theorem 2.1 results in

    (arcsinx)221x2=(1)141(22)!n=0[4n22q=0T(n+1;q,22;12)]x2n+22(2n+22)! (4.8)

    and

    (arcsinx)211x2=(1)1(21)!n=0[4n22q=0T(n+;q,2;1)]x2n+21(2n+21)! (4.9)

    for N. Substituting the power series expansions (4.8) and (4.9) into (4.7) and employing the indefinite integral (4.6) respectively reveal

    Ls(22)j(θ)=221sin(θ/2)0(arcsinx)221x2lnj2+1(2x)dx=(1)243(22)!n=0[4n(2n+22)!22q=0T(n+1;q,22;12)]×sin(θ/2)0x2n+22(ln2+lnx)j2+1dx=(1)243(22)!n=0[4n(2n+22)!22q=0T(n+1;q,22;12)]×j2+1α=0(j2+1α)(ln2)j2α+1sin(θ/2)0x2n+22(lnx)αdx=(1)243(22)!(ln2)j(sinθ2ln2)21n=0[4n(2n+22)!(sinθ2)2n×22q=0T(n+1;q,22;12)]×j2+1α=0(j2+1α)(lnsinθ2ln2)ααk=0(1)kαk(2n+21)k+1(lnsinθ2)k

    for j211 and

    Ls(21)j(θ)=22sin(θ/2)0(arcsinx)211x2lnj2(2x)dx=(1)22(21)!n=0[4n(2n+21)!22q=0T(n+1;q,22;1)]×sin(θ/2)0x2n+21(ln2+lnx)j2dx=(1)22(21)!n=0[4n(2n+21)!22q=0T(n+1;q,22;1)]×j2α=0(j2α)(ln2)j2αsin(θ/2)0x2n+21(lnx)αdx=(1)(21)!(ln2)j(2sinθ2ln2)2n=0[(2sinθ2)2n(2n+21)!22q=0T(n+1;q,22;1)]×j2α=0(j2α)(lnsinθ2ln2)ααk=0(1)kαk(2n+2)k+1(lnsinθ2)k

    for j21. The series representations (4.3) and (4.4) are thus proved. The proof of Theorem 4.1 is complete.

    Finally, we list several remarks on our main results and related stuffs.

    Remark 5.1. For nk1, the first kind Stirling numbers s(n,k) can be explicitly computed by

    |s(n+1,k+1)|=n!n1=k11112=k112k21k1=21k1k11k=11k. (5.1)

    The formula (5.1) was derived in [41,Corollary 2.3] and can be reformulated as

    |s(n+1,k+1)|n!=nm=k|s(m,k)|m!

    for nk1. From the equation (1.5), by convention, we assume s(n,k)=0 for n<k and k,n<0. In recent years, the first kind Stirling numbers s(n,k) have been investigated in [39,40,41,42,45] and closely related references therein.

    Remark 5.2. For |x|<1, we have the following series expansions of arcsinx and its powers.

    1. The series expansion (3.2) of arcsinx can be rewritten as

    arcsinxx=1!n=0[(2n1)!!]2x2n(2n+1)!, (5.2)

    where (1)!!=1. Various forms of (5.2) can be found in [1,4.4.40] and [2,p. 121,6.41.1].

    2. The series expansion of (arcsinx)2 can be rearranged as

    (arcsinxx)2=2!n=0[(2n)!!]2x2n(2n+2)!. (5.3)

    The variants of (5.3) can be found in [2,p. 122,6.42.1], [4,pp. 262–263,Proposition 15], [5,pp. 50–51 and p. 287], [6,p. 384], [7,p. 2,(2.1)], [13,Lemma 2], [20,p. 308], [21,pp. 88-90], [22,p. 61,1.645], [32,p. 1011], [33,p. 453], [47,Section 6.3], [58], [60,p. 59,(2.56)], or [62,p. 676,(2.2)]. It is clear that the series expansion (5.3) and its equivalent forms have been rediscovered repeatedly. For more information on the history, dated back to 1899 or earlier, of the series expansion (5.3) and its equivalent forms, see [7,p. 2] and [32,p. 1011].

    3. The series expansion of (arcsinx)3 can be reformulated as

    (arcsinxx)3=3!n=0[(2n+1)!!]2[nk=01(2k+1)2]x2n(2n+3)!. (5.4)

    Different variants of (5.4) can be found in [2,p. 122,6.42.2], [4,pp. 262–263,Proposition 15], [11,p. 188,Example 1], [20,p. 308], [21,pp. 88–90], [22,p. 61,1.645], or [27,pp. 154–155,(832)].

    4. The series expansion of (arcsinx)4 can be restated as

    (arcsinxx)4=4!n=0[(2n+2)!!]2[nk=01(2k+2)2]x2n(2n+4)!. (5.5)

    There exist three variants of (5.5) in [4,pp. 262–263,Proposition 15], [7,p. 3,(2.2)], and [20,p. 309].

    5. Basing on the formula (2.21) in [28,p. 50], we concretely obtain

    (arcsinxx)5=5!2n=0[(2n+3)!!]2[(n+1k=01(2k+1)2)2n+1k=01(2k+1)4]x2n(2n+5)!. (5.6)

    6. In [7], the special series expansions

    (arcsinx2)2=12n=1x2n(2nn)n2,(arcsinx2)4=32n=1(n1m=11m2)x2n(2nn)n2,(arcsinx2)6=454n=1(n1m=11m2m1=112)x2n(2nn)n2,(arcsinx2)8=3152n=1(n1m=11m2m1=1121p=11p2)x2n(2nn)n2

    were listed. In general, it was obtained in [7,pp. 1–2] that

    (arcsinx2)2=(2)!n=1H(n)x2n(2nn)n2,N (5.7)

    and

    (arcsinx2)2+1=(2+1)!n=1G(n)(2nn)24n+1x2n+12n+1,{0}N, (5.8)

    where H1(n)=14, G0(n)=1,

    H+1(n)=14n1m1=11(2m1)2m11m2=11(2m2)2m11m=11(2m)2,

    and

    G(n)=n1m1=01(2m1+1)2m11m2=01(22+1)2m11m=01(2m+1)2.

    The convention is that the sum is zero if the starting index exceeds the finishing index.

    7. In [7,(2.9) and (4.3)], [25,p. 480,(88.2.2)], and [56,p. 124], there exist the formulas

    (5.9)

    and

    (5.10)

    All the power series expansions from (5.2) to (5.6) can also be deduced from Theorem 2.1.

    By the way, we notice that the quantity in the pair of bigger brackets, the coefficient of , in the formula (5.9) has no explicit relation with . This means that there must be some misprints and typos somewhere in the formula (5.9). On 30 January 2021, Christophe Vignat (Tulane University) pointed out that is the missing information in the formula (5.9).

    In [28,pp. 49–50,Section 2.4], the power series expansions of for were concretely and explicitly written down in alternative forms. The main idea in the study of the power series expansions of for was related with series representations for generalized logsine functions in [28,p. 50,(2.24) and (2.25)]. The special interest is special values of generalized logsine functions defined by [28,p. 50,(2.26) and (2.27)].

    In [54,Theorem 1.4] and [55,Theorem 2.1], the th derivative of was explicitly computed.

    In [43,44], three series expansions (5.2), (5.3), (5.4) and their first derivatives were used to derive known and new combinatorial identities and others.

    Because coefficients of and in (2.1) and (2.2) contain three times sums, coefficients of and in (5.7) and (5.8) contain times sums, coefficients of in (5.9) contain times sums, and coefficients of in (5.10) contain times sums, we conclude that the series expansions (2.1) and (2.2) are more elegant, more operable, more computable, and more applicable.

    Remark 5.3. Two expressions (2.1) and (2.2) in Theorem 2.1 for series expansions of and are very close and similar to, but different from, each other. Is there a unified expression for series expansions of and ? If yes, two closed-form formulas for in Theorem 1.1 would also be unified. We believe that the formula

    (5.11)

    mentioned in [7,p. 3,(2.7)] and collected in [25,p. 210,(10.49.33)] would be useful for unifying two expressions (2.1) and (2.2) in Theorem 2.1, where extended Pochhammer symbols

    (5.12)

    were defined in [25,p. 5,Section 2.2.3], and the Euler gamma function is defined [59,Chapter 3] by

    What are closed forms and why do we care closed forms? Please read the paper [8].

    Remark 5.4. In [2,p. 122,6.42], [27,pp. 154–155,(834)], [33,p. 452,Theorem], and [47,Section 6.3,Theorem 21,Sections 8 and 9], it was proved or collected that

    (5.13)

    In [6,p. 385], [47,Theorem 24], and [61,p. 174,(10)], it was proved that

    (5.14)

    These series expansions (5.13) and (5.14) can be derived directly from the series expansion for and are a special case of (4.9) for .

    Remark 5.5. The series expansion of the function was listed in [2,p. 122,6.42.4] which can be corrected and reformulated as

    (5.15)

    Basing on the relation

    and utilizing series expansions of and , after simple operations, we can readily derive

    (5.16)

    and

    (5.17)

    From (4.8) and (4.9), we can generalize the series expansions (5.15), (5.16), and (5.17) as

    (5.18)

    and

    (5.19)

    for , where

    and is defined by (1.6). Considering both coefficients of and in the power series expansions (5.18) and (5.19) must be , we acquire two combinatorial identities

    for , where is defined by (1.6).

    Remark 5.6. Making use of Theorem 1.1, we readily obtain the first several values of the sequence (1.3) in Tables 1 and 2.

    Table 1.  The sequence in (1.3) for .

     | Show Table
    DownLoad: CSV
    Table 2.  The sequence in (1.3) for .

     | Show Table
    DownLoad: CSV

    In the papers [46,48,49,50,51,52,53,54,55] and closely related references therein, the authors and their coauthors discovered and applied closed form expressions for many special values of the second kind Bell polynomials for .

    Remark 5.7. Taking in (4.3) and (4.4) give

    and

    for , where for and denotes the falling factorial defined by (2.4) and is defined by (1.6). In [28,p. 50], it was stated that the values have been related to special interest in the calculation of the multiloop Feynman diagrams [19,20].

    Similarly, we can also deduce series representations for special values of the logsine function at , , and . These special values were originally derived in [30,31,34] and also considered in [3,9,10,14,15,16,17,19,20,28,29,37,38,57] and closely related references therein.

    Remark 5.8. This paper is a revised version of electronic arXiv preprints [23,24].

    The authors thank

    1. Frank Oertel (Philosophy, Logic & Scientific Method Centre for Philosophy of Natural and Social Sciences, London School of Economics and Political Science, UK; f.oertel@email.de) for his citing the paper [53] in his electronic preprint [35]. On 10 October 2020, this citation and the Google Scholar Alerts leaded the authors to notice the numbers (1.2) in [35]. On 26 January 2021, he sent the important paper [7] to the authors and others. We communicated and discussed with each other many times.

    2. Chao-Ping Chen (Henan Polytechnic University, China; chenchaoping@sohu.com) for his asking the combinatorial identity in [43,Theorem 2.2], or the one in [44,Theorem 2.1], via Tencent QQ on 18 December 2020. Since then, we communicated and discussed with each other many times.

    3. Mikhail Yu. Kalmykov (Bogoliubov Laboratory of Theoretical Physics, Joint Institute for Nuclear Research, Russia; kalmykov.mikhail@googlemail.com) for his noticing [43,Remark 4.2] and providing the references [19,20,28,30,31,34] on 9 and 27 January 2021. We communicated and discussed with each other many times.

    4. Li Yin (Binzhou University, China; yinli7979@163.com) for his frequent communications and helpful discussions with the authors via Tencent QQ online.

    5. Christophe Vignat (Department of Physics, Universite d'Orsay, France; Department of Mathematics, Tulane University, USA; cvignat@tulane.edu) for his sending electronic version of those pages containing the formulas (5.9), (5.11), and (5.12) in [25,56] on 30 January 2021 and for his sending electronic version of the monograph [27] on 8 February 2021.

    6. Frédéric Ouimet (California Institute of Technology, USA; ouimetfr@caltech.edu) for his photocopying by Caltech Library Services and transferring via ResearchGate those two pages containing the formulas (5.9) and (5.11) on 2 February 2021.

    7. anonymous referees for their careful corrections to and valuable comments on the original version of this paper.

    The author Dongkyu Lim was partially supported by the National Research Foundation of Korea under Grant NRF-2021R1C1C1010902, Republic of Korea.

    All authors contributed equally to the manuscript and read and approved the final manuscript.

    The authors declare that they have no conflict of interest.



    [1] D. D. Hai, R. Shivaji, Existence and multiplicity of positive radial solutions for singular superlinear elliptic systems in the exterior of a ball, J. Differ. Equ., 266 (2019), 2232–2243. https://doi.org/10.1016/j.jde.2018.08.027 doi: 10.1016/j.jde.2018.08.027
    [2] K. J. Brown, M. M. A. Ibrahim, R. Shivaji, S-Shaped bifurcation curves, Nonlinear Anl., 5 (1981), 475–486.
    [3] T. Laetsch, The number of solutions of a nonlinear two point boundary value problem, Indiana Univ. Math. J., 20 (1970), 1–13.
    [4] E. Ko, E. K. Lee, R. Shivaji, On S-shaped and reversed S-shaped bifurcation curves for singular problems, Electron. J. Qual. Theo., 31 (2011), 1–12. https://doi.org/10.14232/ejqtde.2011.1.31 doi: 10.14232/ejqtde.2011.1.31
    [5] U. Das, A. Muthunayake, R. Shivaji, Existence results for a class of p-q Laplacian semipositone boundary value problems, Electron. J. Qual. Theo., 88 (2020), 1–7. https://doi.org/10.14232/ejqtde.2020.1.88 doi: 10.14232/ejqtde.2020.1.88
    [6] A. Castro, R. Shivaji, Non-negative solutions for a class of non-positone problems, P. Roy. Soc. Edinb. A, 108 (1988), 291–302. https://doi.org/10.1017/S0308210500014670 doi: 10.1017/S0308210500014670
    [7] A. C. Lazer, P. J. McKenna, On a singular nonlinear elliptic boundary value problem, P. Am. Math. Soc., 111 (1991), 720–730. https://doi.org/10.2307/2048410 doi: 10.2307/2048410
    [8] Z. Zhang, On a Dirichlet problem with a singular nonlinearity, J. Math. Anal. Appl., 194 (1995), 103–113. https://doi.org/10.1006/jmaa.1995.1288 doi: 10.1006/jmaa.1995.1288
    [9] Y. Miaoxin, J. Shi, On a singular nonlinear semilinear elliptic problem, P. Roy. Soc. Edinb. A, 128 (1998), 1389–1401. https://doi.org/10.1017/S0308210500027384 doi: 10.1017/S0308210500027384
  • This article has been cited by:

    1. Can Kizilateş, Explicit, determinantal, recursive formulas, and generating functions of generalized Humbert–Hermite polynomials via generalized Fibonacci polynomials, 2023, 0170-4214, 10.1002/mma.9048
    2. Sumedh B. Thool, Yogesh J. Bagul, Ramkrishna M. Dhaigude, Christophe Chesneau, Bounds for Quotients of Inverse Trigonometric and Inverse Hyperbolic Functions, 2022, 11, 2075-1680, 262, 10.3390/axioms11060262
    3. Feng Qi, Muhammet Cihat Dağlı, Dongkyu Lim, Several explicit formulas for (degenerate) Narumi and Cauchy polynomials and numbers, 2021, 19, 2391-5455, 833, 10.1515/math-2021-0079
    4. Muhammet Cihat Dağlı, Explicit, determinantal, recursive formulas and relations of the Peters polynomials and numbers, 2022, 45, 0170-4214, 2582, 10.1002/mma.7941
    5. Bai-Ni Guo, Dongkyu Lim, Feng Qi, MacLaurin’s series expansions for positive integer powers of inverse (hyperbolic) sine and tangent functions, closed-form formula of specific partial Bell polynomials, and series representation of generalized logsine function, 2022, 16, 1452-8630, 427, 10.2298/AADM210401017G
    6. Yan Hong, Bai-Ni Guo, Feng Qi, Determinantal Expressions and Recursive Relations for the Bessel Zeta Function and for a Sequence Originating from a Series Expansion of the Power of Modified Bessel Function of the First Kind, 2021, 129, 1526-1506, 409, 10.32604/cmes.2021.016431
    7. Jian Cao, Feng Qi, Wei-Shih Du, Closed-Form Formulas for the nth Derivative of the Power-Exponential Function xx, 2023, 15, 2073-8994, 323, 10.3390/sym15020323
    8. Wen-Hui Li, Qi-Xia Shen, Bai-Ni Guo, Several Double Inequalities for Integer Powers of the Sinc and Sinhc Functions with Applications to the Neuman–Sándor Mean and the First Seiffert Mean, 2022, 11, 2075-1680, 304, 10.3390/axioms11070304
    9. Ravi Prakash Agarwal, Erdal Karapinar, Marko Kostić, Jian Cao, Wei-Shih Du, A Brief Overview and Survey of the Scientific Work by Feng Qi, 2022, 11, 2075-1680, 385, 10.3390/axioms11080385
    10. Feng Qi, Taylor’s series expansions for real powers of two functions containing squares of inverse cosine function, closed-form formula for specific partial Bell polynomials, and series representations for real powers of Pi , 2022, 55, 2391-4661, 710, 10.1515/dema-2022-0157
    11. Lan Wu, Xue-Yan Chen, Muhammet Cihat Dağli, Feng Qi, On Degenerate Array Type Polynomials, 2022, 131, 1526-1506, 295, 10.32604/cmes.2022.018778
    12. Wei-shih DU, Dongkyu LIM, Feng Qİ, Several recursive and closed-form formulas for some specific values of partial Bell polynomials, 2022, 6, 2587-2648, 528, 10.31197/atnaa.1170948
    13. Feng Qi, Peter Taylor, Series expansions for powers of sinc function and closed-form expressions for specific partial bell polynomials, 2024, 18, 1452-8630, 92, 10.2298/AADM230902020Q
    14. Feng Qi, Power Series Expansions of Real Powers of Inverse Cosine and Sine Functions, Closed-Form Formulas of Partial Bell Polynomials at Specific Arguments, and Series Representations of Real Powers of Circular Constant, 2024, 16, 2073-8994, 1145, 10.3390/sym16091145
    15. A. V. Grishaev, V. F. Sazonov, New formulas for Maclaurin's series expansions of powers of arctangent and other related topics, 2024, 1065-2469, 1, 10.1080/10652469.2024.2382810
    16. Feng Qi, Da-Wei Niu, Dongkyu Lim, Some combinatorial identities containing central binomial coefficients or Catalan numbers*, 2023, 31, 2769-0911, 10.1080/27690911.2023.2204233
    17. Feng Qi, Gradimir Milovanovic, Dongkyu Lim, Specific values of partial bell polynomials and series expansions for real powers of functions and for composite functions, 2023, 37, 0354-5180, 9469, 10.2298/FIL2328469Q
    18. Feng Qi, Series and Connections Among Central Factorial Numbers, Stirling Numbers, Inverse of Vandermonde Matrix, and Normalized Remainders of Maclaurin Series Expansions †, 2025, 13, 2227-7390, 223, 10.3390/math13020223
  • 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(1887) PDF downloads(63) Cited by(1)

Figures and Tables

Figures(7)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog