Research article

Spatial twisted central configuration for Newtonian (2N+1)-body problem

  • Received: 04 November 2023 Revised: 11 March 2024 Accepted: 10 April 2024 Published: 20 May 2024
  • 70F10, 70F15

  • For a spatial twisted central configuration of the Newtonian (2N+1)-body problem where 2N masses are at the vertices of two paralleled regular N-polygons with distance h>0, and the twist angle between the two regular N-polygons is 0θ<2π, we study the sufficient and necessary conditions for the existence of the spatial twisted central configuration. Additionally, we obtain the uniqueness of the spatial twisted central configuration.

    Citation: Liang Ding, Jinrong Wang, Jinlong Wei. Spatial twisted central configuration for Newtonian (2N+1)-body problem[J]. Communications in Analysis and Mechanics, 2024, 16(2): 388-415. doi: 10.3934/cam.2024018

    Related Papers:

    [1] Xueqi Sun, Yongqiang Fu, Sihua Liang . Normalized solutions for pseudo-relativistic Schrödinger equations. Communications in Analysis and Mechanics, 2024, 16(1): 217-236. doi: 10.3934/cam.2024010
    [2] Wang Xiao, Xuehua Yang, Ziyi Zhou . Pointwise-in-time $ \alpha $-robust error estimate of the ADI difference scheme for three-dimensional fractional subdiffusion equations with variable coefficients. Communications in Analysis and Mechanics, 2024, 16(1): 53-70. doi: 10.3934/cam.2024003
    [3] Yining Yang, Cao Wen, Yang Liu, Hong Li, Jinfeng Wang . Optimal time two-mesh mixed finite element method for a nonlinear fractional hyperbolic wave model. Communications in Analysis and Mechanics, 2024, 16(1): 24-52. doi: 10.3934/cam.2024002
    [4] Shengbing Deng, Qiaoran Wu . Existence of normalized solutions for the Schrödinger equation. Communications in Analysis and Mechanics, 2023, 15(3): 575-585. doi: 10.3934/cam.2023028
    [5] Enzo Vitillaro . Nontrivial solutions for the Laplace equation with a nonlinear Goldstein-Wentzell boundary condition. Communications in Analysis and Mechanics, 2023, 15(4): 811-830. doi: 10.3934/cam.2023039
    [6] Zhen Wang, Luhan Sun . The Allen-Cahn equation with a time Caputo-Hadamard derivative: Mathematical and Numerical Analysis. Communications in Analysis and Mechanics, 2023, 15(4): 611-637. doi: 10.3934/cam.2023031
    [7] Ho-Sik Lee, Youchan Kim . Boundary Riesz potential estimates for parabolic equations with measurable nonlinearities. Communications in Analysis and Mechanics, 2025, 17(1): 61-99. doi: 10.3934/cam.2025004
    [8] Mohamed Karim Hamdani, Lamine Mbarki, Mostafa Allaoui . A new class of multiple nonlocal problems with two parameters and variable-order fractional $ p(\cdot) $-Laplacian. Communications in Analysis and Mechanics, 2023, 15(3): 551-574. doi: 10.3934/cam.2023027
    [9] Cheng Yang . On the Hamiltonian and geometric structure of Langmuir circulation. Communications in Analysis and Mechanics, 2023, 15(2): 58-69. doi: 10.3934/cam.2023004
    [10] Siegfried Carl . Quasilinear parabolic variational-hemivariational inequalities in $ \mathbb{R}^N\times (0, \tau) $ under bilateral constraints. Communications in Analysis and Mechanics, 2025, 17(1): 41-60. doi: 10.3934/cam.2025003
  • For a spatial twisted central configuration of the Newtonian (2N+1)-body problem where 2N masses are at the vertices of two paralleled regular N-polygons with distance h>0, and the twist angle between the two regular N-polygons is 0θ<2π, we study the sufficient and necessary conditions for the existence of the spatial twisted central configuration. Additionally, we obtain the uniqueness of the spatial twisted central configuration.



    For the spatial Newtonian n-body problem, the equations of motion for the n masses mk>0 and positions xkR3 with k{1,,n} can be described by Newton's second law and Newton's universal gravitation law:

    mk¨xk=(1s<jnmjms|xjxs|)xk.

    Define

    Ω={x: x=(x1,x2,,xn)(R3)n},

    and let

    =1jsn{x=(x1,x2,,xn)|xj=xs,1jsn}

    be the collision set. As usual, the set Ω is called the configuration space. First, we introduce the definition of central configuration for the Newtonian n-body problem (see [1]).

    Definition 1.1. Given n masses mk>0 with positions qkR3, k=1,,n, we say a configuration qΩ is a central configuration at some moment if there exists a constant λR such that

    {jk1jnmjmk(qjqk)|qjqk|3=λmk(qkx0),  k=1,2,,n,x0=1knmkqk1knmk. (1.1)

    Central configurations play a very important role in the study of the Newtonian n-body problem, and especially, central configurations can lead to rigid-motion solutions and homothetically collapsing solutions [1]. Central configurations of the Newtonian three-body (n=3) problem with any given three masses have long been known, and there are always exactly two kinds of central configurations: Euler collinear central configuration and Lagrange equilateral-triangle central configuration [2,3]. For a planar Newtonian N-body problem with n=N4, Perko and Walter [4] proved that if N masses are located at the vertices of a regular N-polygon (see Figure 1), then they can form a regular polygonal central configuration if and only if all the values of N masses are equal to each other. For more results of planar central configuration with one regular N-polygon, one can refer to [5,6,7,8].

    Figure 1.  Planar N-body problem.

    For a planar central configuration with n=2N and N2 such that two regular N-polygons are concentric and that 2N equal masses are placed at the vertices of the two regular N-polygons (see Figure 2), Zhang and Zhou [9] proved that the values of masses in each separate regular N-polygon were equal. We say that p regular N-polygons with p2 are nested if they are coplanar and have the same number of vertices N and the same center, and the positions of the vertices of the innermost regular N-polygon R(1)j and those of the remaining p-1 regular N-polygons R(k)j with any k{2,,p} satisfy the relation that R(p)j=s1R(p1)j=s2R(p2)j==sp1R(1)j for some scale factors sp1>sp2>>s1>1 and for all j=1,2,,N. For the central configuration such that two regular N-polygons are nested, masses on different regular N-polygons may be different, and Moeckel and Simó [10] proved that for every mass ratio b between the two masses, there were exactly two planar central configurations. Also, for the case of n=2N such that N equal masses are placed at the vertices of one regular N-polygon and the remaining N equal masses are placed at the vertices of the other regular N-polygon, which is rotated exactly at an angle θ=π/N with respect to the former regular N-polygon, Barrabés and Cors [11] proved the existence of the planar central configuration with any value of the mass ratio. For the case of n=pN and p2, Corbera, Delgado, and Llibre [12] proved the existence of the nested central configuration such that pN masses were at the vertices of the p nested regular N-polygon with a common center. Moreover, all the masses on the same regular N-polygon were equal, but masses on a different regular N-polygon could be different. For the case of n=pN+gN with p1 and g1, Zhao and Chen [13] proved the existence of planar central configurations such that p regular N-polygons were nested, and g regular N-polygons were rotated exactly at an angle π/N with respect to the other ones. For more details in this direction, we refer to [14,15,16,17,18,19,20,21] and the references therein.

    Figure 2.  Planar 2N-body problem.

    Note that for a planar central configuration with n=N+1, Chen and Luo [22] proved that if N masses are located at the vertices of one regular N-polygon and the position of the (N+1)-th mass is on the plane containing the regular N-polygon (see Figure 3), then all the values of N masses located at the vertices of the regular N-polygon are equal to each other. For a spatial central configuration with n=N+1 and the (N+1)-th mass off the plane containing the regular N-polygon (see Figure 4), Ouyang, Xie, and Zhang [23] showed that the distance between the (N+1)-th mass and the regular N-polygon was unique.

    Figure 3.  Planar (N+1)-body problem.
    Figure 4.  Spatial (N+1)-body problem.

    In this paper, we consider the spatial central configuration of a Newtonian (2N+1)-body problem in R3 formed by 2N masses placed at the vertices of two paralleled regular N-polygons with distance h>0. It is assumed that the lower layer regular N-polygon lies in a horizontal plane, and the upper regular N-polygon parallels the lower one in R3 with distance h, and the z-axis passes through both centers of the two regular N-polygons (see Figure 5). For convenience, when choosing the coordinates, we treat R3 as the direct product of the complex plane and real axis. For the positions of the 2N+1 masses q=(q1,q2,,q2N,q2N+1)Ω, we have

    {qk=(ρk, 0),   k=1,,N,ql=(aρleiθ, h),   0θ<2π,  l=N+1,,2N,   a>0,   h>0, (1.2)
    Figure 5.  Spatial (2N+1)-body problem.

    where a is the ratio of the sizes of the two regular N-polygons, ρd is the d(modN)-th complex root of unity, i.e., ρk=eiθk with k=1,2,,N, and ρl=eiθl with l=N+1,N+2,,2N and θd=2dπ/N with dZ. Here, we define θ as the twist angle between the two paralleled regular N-polygons with distance h>0. Moreover, for the position of the (2N+1)-th mass and the barycenter x0=(c0,h0), we define

    {q2N+1=(a1eiα, h2N+1),  a10,  0α<2π,  <h2N+1<+,c0=1kNmkρk+N+1l2Namlρleiθ+a1m2N+1eiαm2N+1+1kNmk+N+1l2Nml,h0=N+1l2Nmlh+m2N+1h2N+1m2N+1+1kNmk+N+1l2Nml. (1.3)

    Then, for the spatial twisted configuration with n=2N+1 and the notations (1.2)–(1.3), we have the following results.

    For the existence, we have the following:

    Theorem 1.1. Suppose the values of N masses with N2 located at the vertices of one regular N-polygon are equal to each other, and all of the sides in the two regular N-polygons have the same size. Define the position of q2N+1 as (1.3). Then, the 2N+1 masses form a central configuration if and only if all the values of the 2N masses located at the vertices of the two regular N-polygons are equal to each other, a1=0 and h2N+1=h/2, and the twist angle is θ=sπ/N with s{0,1,,2N1}.

    Remark 1.1. For the spatial twisted central configuration of the Newtonian 2N-body problem, under the assumption that the values of masses in each separate regular N-polygon were equal, Yu and Zhang [24] proved that the twist angle must be θ=0 or θ=π/N (see Figures 6 and 7). Meanwhile, in Theorem 1.1, we consider the spatial twisted central configuration of the Newtonian (2N+1)-body problem. Under the assumptions that the values of the N masses located at the vertices of one regular N-polygon are equal and that all of the sides in the two regular N-polygons have the same size, we not only obtain the values of the twist angle; but also prove that all 2N masses must be equal. Moreover, we know the position of the (2N+1)-th mass is (0,0,h/2).

    Figure 6.  Spatial 2N-body problem with θ=0.
    Figure 7.  Spatial 2N-body problem with θ=π/N.

    For the uniqueness, we have the following:

    Theorem 1.2. For the spatial twisted central configuration, if the values of the N masses located at the vertices of one regular N-polygon are equal to each other and all of the sides in the two regular N-polygons have the same size, then for any N2, both the distance between the two regular N-polygons and the position of the (2N+1)-th mass are unique.

    Lemma 2.1. [24, Lemma 2.9] For any a>0, any γ(,+), any h>0, and any N2, let

    f(γ)=1jNasin(θj+γ)[1+a22acos(θj+γ)+h2]32. (2.1)

    Then,

    f(πN)=0,  f(γ)=f(γ)  and  f(γ+2πN)=f(γ).

    Remark 2.1. In Lemma 2.1, if we choose a=1 and γ=π/N where N2, then

    1jNsin(θjπN)[22cos(θjπN)+h2]32=1jNsin(θj+πN)[22cos(θj+πN)+h2]32=0,  where  h>0.

    Lemma 2.2. [24, Lemma 2.10] If γ(0,π/N) with N2, then for any a>0 and any h>0, we have f(γ)>0.

    Lemma 2.3. If θ=sπ/N with s{0,1,,2N1} and N2, then for any k{1,2,,N}, any l{N+1,N+2,,2N}, and any h>0, we have

    {1kNei(θklθ)1[|ei(θklθ)1|2+h2]32=N+1l2Nei(θlk+θ)1[|ei(θlk+θ)1|2+h2]32=1jNei(θj+θ)1[|ei(θj+θ)1|2+h2]32R,1kNh[|ei(θklθ)1|2+h2]32=N+1l2Nh[|ei(θlk+θ)1|2+h2]32=1jNh[|ei(θj+θ)1|2+h2]32. (2.2)

    Proof. Let ˆμ{1,2,,N}, ˜μ{0,1,,N1}, ˜α(2π,2π), and κ{0,1}. We define a mapping σ by

    {{κN+1,κN+2,,κN+N}σ{˜α,2πN+˜α,...,2(N1)πN+˜α},σ(μ)=[(μˆμ+˜μ)(modN)]2πN+˜α,μ{κN+1,κN+2,,κN+N}. (2.3)

    Notice that both {κN+1,κN+2,,κN+N} and {˜α,(2π)/N+˜α,,2(N1)π/N+˜α} are finite sets; the mapping σ is a surjection. Let us show σ is an injective mapping. The proof for κ=0 is similar to κ=1; we only check for κ=1. Let μ1μ2 and μ1,μ2{N+1,N+2,,2N}. If σ(μ1)=σ(μ2), then there exist s1,s2Z such that

    (μ1ˆμ+˜μ)+s1N=(μ2ˆμ+˜μ)+s2N. (2.4)

    Hence, μ1μ2=(s2s1)N. By the facts that N<μ1μ2<N and s2s1Z, s2=s1, and thus μ1=μ2, which is a contradiction. Therefore, σ is injective, which implies that σ is a bijection.

    Similarly, for l{N+1,N+2,,2N}, ˜s{0,1,,N1}, and θ[0,2π), if we define another mapping σ1 by

    {{1,2,,N}σ1{θ,2πN+θ,,2(N1)πN+θ},σ1(k)=[(lk+˜s)(modN)]2πN+θ,k{1,2,,N}, (2.5)

    then σ1 is a bijection as well. Together with the fact that θ=sπ/N with s{0,1,,2N1} is equivalent to θ=2˜sπ/N or θ=2˜sπ/N+π/N with ˜s{0,1,,N1}, let us show (2.2) holds by considering the following two cases.

    Case 1. θ=2˜sπ/N with ˜s{0,1,,N1}.

    Observing that θj=2jπ/N with j{1,2,,N},

    1jNsinθj[22cosθj+h2]32=1jNsin2πNjN[22cos2πNjN+h2]32=1jNsinθj[22cosθj+h2]32,

    which implies that

    1jNsinθj[22cosθj+h2]32=0. (2.6)

    Let k{1,2,,N} and ˜s{0,1,,N1}. In (2.3), if we choose μ=l, ˆμ=k, ˜μ=˜s, κ=1, and ˜α=0, then the mapping

    {{N+1,N+2,,2N}σ{0,2πN,,2(N1)πN},σ(l)=[(lk+˜s)(modN)]2πN,  l{N+1,N+2,,2N}

    is a bijection. Combining (2.6), and θd=2πd/N with dZ and

    {cos(θlk+2πN˜s)=cos(2πN[(lk+˜s)(modN)]),sin(θlk+2πN˜s)=sin(2πN[(lk+˜s)(modN)])

    for any k{1,2,,N} and any ˜s{0,1,,N1}, we have

    {N+1l2Nsin(θlk+2πN˜s)[22cos(θlk+2πN˜s)+h2]32=N+1l2Nsinθlk+˜s[22cosθlk+˜s+h2]32=1jNsinθj[22cosθj+h2]32=0,N+1l2Ncos(θlk+2πN˜s)1[22cos(θlk+2πN˜s)+h2]32=N+1l2Ncos(θlk+˜s)1[22cos(θlk+˜s)+h2]32=1jNcosθj1[22cosθj+h2]32,N+1l2Nh[22cos(θlk+2πN˜s)+h2]32=N+1l2Nh[22cos(θlk+˜s)+h2]32=1jNh[22cosθj+h2]32. (2.7)

    On the other hand, in (2.3), for any l{N+1,N+2,,2N} and any ˜s{0,1,,N1}, if we let μ=k, ˆμ=l, ˜μ=N˜s, κ=0, and ˜α=0, then we know that the mapping

    {{1,2,,N}σ{0,2πN,,2(N1)πN},σ(k)=[(kl+(N˜s))(modN)]2πN=[(kl˜s)(modN)]2πN,  k{1,2,,N}

    is a bijection as well. Thus, similar to the procedure of obtaining (2.7), for any l{N+1,N+2,,2N}, any ˜s{0,1,,N1}, and any h>0, we have

    {1kNsin(θkl2πN˜s)[22cos(θkl2πN˜s)+h2]32=1jNsinθj[22cosθj+h2]32=0,1kNcos(θkl2πN˜s)1[22cos(θkl2πN˜s)+h2]32=1jNcosθj1[22cosθj+h2]32,1kNh[22cos(θkl2πN˜s)+h2]32=1jNh[22cosθj+h2]32. (2.8)

    Employing (2.7), (2.8), and θd=2πd/N with dZ, we have

    N+1l2Nei(θlk+2πN˜s)1[|ei(θlk+2πN˜s)1|2+h2]32=N+1l2N[cos(θlk+2πN˜s)1]+isin(θlk+2πN˜s)[22cos(θlk+2πN˜s)+h2]32=1jN(cosθj1)+isinθj[22cosθj+h2]32=1jNcosθj1[22cosθj+h2]32=1kN[cos(θkl2πN˜s)1]+isin(θkl2πN˜s)[22cos(θkl2πN˜s)+h2]32=1kNei(θkl2πN˜s)1[|ei(θkl2πN˜s)1|2+h2]32R

    and

    N+1l2Nh[|ei(θlk+2πN˜s)1|2+h2]32=N+1l2Nh[22cos(θlk+2πN˜s)+h2]32=1jNh[22cosθj+h2]32=1jNh[22cosθj+h2]32=1kNh[22cos(θkl2πN˜s)+h2]32=1kNh[|ei(θkl2πN˜s)1|2+h2]32,

    where ˜s{0,1,,N1}, k{1,2,,N}, and l{N+1,N+2,,2N}. Thus, (2.2) holds for the case of θ=2˜sπ/N with ˜s{0,1,2,,N1}.

    Case 2. θ=2˜sπ/N+π/N with ˜s{0,1,2,,N1}.

    For any l{N+1,N+2,,2N} and any ˜s{0,1,,N1}, in (2.3), if we choose μ=k, ˆμ=l, ˜μ=N˜s, κ=0, and ˜α=π/N, then the mapping

    {{1,2,,N}σ{πN,2πNπN,,2(N1)πNπN},σ(k)=[(kl+(N˜s))(modN)]2πNπN=[(kl˜s)(modN)]2πNπN, k{1,2,,N}

    is a bijection. Then, we have

    {1kNsin(θkl2πN˜sπN)[22cos(θkl2πN˜sπN)+h2]32=1kNsin(θkl˜sπN)[22cos(θkl˜sπN)+h2]32=1jNsin(θjπN)[22cos(θjπN)+h2]32,1kNcos(θkl2πN˜sπN)1[22cos(θkl2πN˜sπN)+h2]32=1kNcos(θkl˜sπN)1[22cos(θkl˜sπN)+h2]32=1jNcos(θjπN)1[22cos(θjπN)+h2]32, (2.9)

    where ˜s{0,1,,N1}, l{N+1,N+2,,2N}, and h>0.

    By the first equation of (2.9) and Remark 2.1, for any ˜s{0,1,,N1}, any l{N+1,N+2,,2N}, and any h>0, we see that

    1kNsin(θkl2πN˜sπN)[22cos(θkl+2πN˜sπN)+h2]32=1jNsin(θjπN)[22cos(θjπN)+h2]32=1jNsin(θj+πN)[22cos(θj+πN)+h2]32=0. (2.10)

    Moreover, in (2.5), for any l{N+1,N+2,,2N} and any ˜s{0,1,,N1}, if we choose θ=π/N, then the mapping

    {{1,2,,N}σ1{πN,2πN+πN,,2(N1)πN+πN},σ1(k)=[(lk+˜s)(modN)]2πN+πN, k{1,2,,N}

    is a bijection. Hence, combining (2.10), this leads to

    {1kNsin(θlk+2πN˜s+πN)[22cos(θlk+2πN˜s+πN)+h2]32=1kNsin(θlk+˜s+πN)[22cos(θlk+˜s+πN)+h2]32=1jNsin(θj+πN)[22cos(θj+πN)+h2]32=0,1kNcos(θlk+2πN˜s+πN)1[22cos(θlk+2πN˜s+πN)+h2]32=1kNcos(θlk+˜s+πN)1[22cos(θlk+˜s+πN)+h2]32=1jNcos(θj+πN)1[22cos(θj+πN)+h2]32. (2.11)

    Furthermore, one can verify that

    1jNcos(θj+πN)1[22cos(θj+πN)+h2]321jNcos(θjπN)1[22cos(θjπN)+h2]32=2jN+1cos(θjπN)1[22cos(θjπN)+h2]321jNcos(θjπN)1[22cos(θjπN)+h2]32=1jNcos(θjπN)1[22cos(θjπN)+h2]321jNcos(θjπN)1[22cos(θjπN)+h2]32=0,

    and this implies that

    1jNcos(θj+πN)1[22cos(θj+πN)+h2]32=1jNcos(θjπN)1[22cos(θjπN)+h2]32. (2.12)

    Employing (2.9), (2.10), (2.11), and (2.12), for any l,l{N+1,N+2,,2N} and any ˜s{0,1,,N1}, we have

    {1kNcos(θkl2πN˜sπN)1[22cosθ(kl2πN˜sπN)+h2]32=1kNcos(θlk+2πN˜s+πN)1[22cosθ(lk+2πN˜s+πN)+h2]32=1jNcos(θjπN)1[22cos(θjπN)+h2]32,1kNsin(θkl2πN˜sπN)[22cos(θkl2πN˜sπN)+h2]32=1kNsin(θlk+2πN˜s+πN)[22cos(θlk+2πN˜s+πN)+h2]32=1jNsin(θjπN)[22cos(θjπN)+h2]32=0.

    Thus, (2.2) holds for the case of θ=2˜sπ/N+π/N, where ˜s{0,1,2,,N1}.

    By Cases 1–2, we arrive at the conclusion that (2.2) holds for θ=sπ/N with s{0,1,,2N1}.

    Remark 2.2. Similar to dealing with the mappings σ and σ1, for any k{1,2,,N} and any l{N+1,N+2,,2N} with N2, if one defines σ2 and σ3 by

    {{1,2,,N}{k}σ2{2πN,4πN,,2(N1)πN},σ2(k)=[(kk)(modN)]2πN, k{1,2,,N}{k},

    and

    {{N+1,N+2,,2N}{l}σ3{2πN,4πN,,2(N1)πN},σ3(l)=[(ll)(modN)]2πN, l{N+1,N+2,,2N}{l},

    then σ2 and σ3 are bijections.

    Lemma 2.4. For any θR, N2, a>0, h>0, and m>0, we have

    1kNhm[|eiθkaei(θl+θ)|2+h2]32constant,  l{N+1,N+2,,2N}. (2.13)

    Proof. In fact, (2.13) is equivalent to

    1kNh[|ei(θklθ)a|2+h2]32constant,  l{N+1,N+2,,2N}.

    It is easy to see that klZ. Moreover, in (2.3), for any l{N+1,N+2,,2N} and any θ[0,2π), if we let μ=k, ˆμ=lN, ˜μ=0, κ=0, and α=θ, then the mapping

    {{1,2,,N}σ{θ,2πN+θ,,2(N1)πN+θ},σ(k)=[(k(lN))(modN)]2πN+θ=[(kl)(modN)]2πN+θ,   k{1,2,,N}

    is a bijection, which implies that

    1kNh[|ei(θklθ)a|2+h2]32=1jNh[|ei(θjθ)a|2+h2]32,  l{N+1,N+2,,2N}.

    Thus, Lemma 2.4 is true.

    Lemma 2.5. [4, Page 304] For any N2, 1jN1(1eiθj)/|1eiθj|3=[1jN1csc(jπ/N)]/4.

    Lemma 2.6. [20, Pages 1431 and 1437] For any N2, the inequality

    1jN1+cos(θj+πN)[22cos(θj+πN)]321jN11eiθj|1eiθj|32>0

    holds.

    Now, we introduce the definition of circulant matrix and state its properties.

    Definition 2.1. [25, Pages 65–66] A matrix ˜C=(˜ckj)N×N is circulant if ˜cˆk,ˆj=˜cˆk1,ˆj1 where 1k,j,ˆk,ˆjN and N2.

    In Definition 2.1, we take the circulant matrix ˜C as the following:

    ˜C=:C=(ck,j),    where   ck,j={1ρkj|1ρkj|3,kj,0,k=j. (2.14)

    We have some properties for the special circulant matrix C.

    Lemma 2.7. [4, Page 303] The circulant matrix C has the same forms of the eigenvalues λj(C) and the corresponding eigenvectors ξj; more precisely,

    λj(C)=1kNc1,kρk1j1,ξj=(ρj1,ρ2j1,,ρNj1)T,j=1,2,,N,

    where N2 and ρj1=eiθj1=e2(j1)πi/N.

    Lemma 2.8. [4, Corollary and Lemma 12] For the eigenvalues of C with jN and N4, λj0 except that λ(N+1)/2=0 for odd N.

    Lemma 2.9. [22, Proposition 2.2] The eigenvectors ξj (j=1,2,,N and N3) of circulant matrix C form a basis of CN.

    Lemma 2.10. [25, Page 65] Denote the conjugate transpose of νk by (ˉνk)T. Then,

    (ˉξk)Tξj={N, k=j,0,  kj, (ρ1,ρ2,,ρN)(ˉξN)T=N.

    Let k{1,2,,N} and l{N+1,N+2,,2N}. By Definition 1.1, it suffices to study the following system:

    {(q2N+1qk)m2N+1mk|q2N+1qk|3+N+1l2N(qlqk)mlmk|qlqk|3+1kkN(qkqk)mkmk|qkqk|3=λmk(qkx0),(q2N+1ql)m2N+1ml|q2N+1ql|3+1kN(qkql)mkml|qkql|3+N+1ll2N(qlql)mlml|qkql|3=λml(qlx0),1kN(qkq2N+1)mkm2N+1|qkq2N+1|3+N+1l2N(qlq2N+1)mlm2N+1|qlq2N+1|3=λm2N+1(q2N+1x0). (3.1)

    Thanks to (1.2), (1.3), and (3.1), the 2N+1 masses form a central configuration if and only if

    {(a1eiαeiθk,h2N+1)m2N+1mk[|a1eiαeiθk|2+h22N+1]32+N+1l2N(aei(θl+θ)eiθk,h)mlmk[|aei(θl+θ)eiθk|2+h2]32+1kkN(eiθkeiθk,0)mkmk|eiθkeiθk|3=λmk(eiθkc0,h0),(a1eiαaei(θl+θ),h2N+1h)m2N+1ml[|a1eiαaei(θl+θ)|2+(h2N+1h)2]32+1kN(eiθkaei(θl+θ),h)mkml[|eiθkaei(θl+θ)|2+h2]32+N+1ll2N(aei(θl+θ)aei(θl+θ),0)mlml|aei(θl+θ)aei(θl+θ)|3                    =λml(aei(θl+θ)c0,hh0),1kN(eiθka1eiα,h2N+1)mkm2N+1[|eiθka1eiα|2+h22N+1]32+N+1l2N(aei(θl+θ)a1eiα,hh2N+1)mlm2N+1[|aei(θl+θ)a1eiα|2+(hh2N+1)2]32=λm2N+1(a1eiαc0,h2N+1h0). (3.2)

    By the assumption that the values of N masses located at the vertices of one regular N-polygon are equal to each other, without loss of generality, we suppose that m1=m2==mN:=m>0, and we divide the proof of the necessity into four steps.

    Step 1. We prove that a1=0.

    Employing m1=m2==mN=m>0 and the second equation of (3.2), we have

    (h2N+1h)m2N+1[|a1eiαaei(θl+θ)|2+(h2N+1h)2]32+1kNhm[|eiθkaei(θl+θ)|2+h2]32=λ(hh0), (3.3)

    where l{N+1,N+2,,2N}. Combining Lemma 2.4, (3.3),

    h0=N+1l2Nmlh+m2N+1h2N+1m2N+1+1kNmk+N+1l2Nml,

    and that λ is independent of the choice of l, we deduce that

    (h2N+1h)m2N+1[|a1eiαaei(θl+θ)|2+(h2N+1h)2]32constant,  l{N+1,N+2,,2N}.

    Thus, for any l{N+1,N+2,,2N}, we have |a1eiαaei(θl+θ)|2constant, i.e.,

    |[a1cosαacos(θl+θ)]+i[a1sinαasin(θl+θ)]|2constant,  l{N+1,N+2,,2N}.

    Then, one computes that

    a1acos(θl+θα)constant,  l{N+1,N+2,,2N}.

    Since a represents the ratio of the sizes of the two regular N-polygons, a>0. Hence, if a10, then

    cos(θl+θα)constant,  l{N+1,N+2,,2N}. (3.4)

    In what follows, we assume that a10, and we divide the proof of impossibility of a10 into two cases: N=2 and N3.

    (i) N=2:

    In this case, l{3,4}, and a10. Then, by (3.4), we have cos(3π+θα)=cos(4π+θα), which implies that cos(θα)=0.

    Under the assumption that m1=m2=m, we convert (1.2) and (1.3) into

    {q1=(1, 0),   q2=(1, 0),q3=(aρ3eiθ, h)=(aeiθ, h),  q4=(aρ4eiθ, h)=(aeiθ, h), 0θ<2π, a>0, h0,q5=(a1eiα, h5),  a10,  0α2π,  <h5<+,c0=aeiθ(m4m3)+a1m5eiαm1+m2+m3+m4+m5=aeiθ(m4m3)+a1m5eiα2m+m3+m4+m5,h0=(m3+m4)h+m5h5m1+m2+m3+m4+m5=(m3+m4)h+m5h52m+m3+m4+m5. (3.5)

    First, in (2.3), for any θ[0,2π) and any l{N+1,N+2,,2N} with N2, if we let μ=k, ˆμ=lN, ˜μ=0, κ=0, and ˜α=θ(2π,0], then the mapping

    {{1,2,,N}σ{θ,2πNθ,,2(N1)πNθ},σ(k)=[(k(lN))(modN)]2πNθ=[(kl)(modN)]2πNθ,   k{1,2,,N}

    is a bijection. Then, by the second equation of (3.2) and m1=m2==mN=m>0, we have

    (1a1aei(αθθl))m2N+1[|aa1ei(αθθl)|2+(h2N+1h)2]32+1kN(1ei(θklθ)a)m[|aei(θklθ)|+h2]32+N+1ll2N(1eiθll)mla3|1eiθll|3=(1a1aei(αθθl))m2N+1[|aa1ei(αθθl)|2+(h2N+1h)2]32+1kN(1ei(θkθ)a)m[|aei(θkθ)|2+h2]32+N+1ll2N(1eiθll)mla3|1eiθll|3=λλac0ei(θl+θ),  where  l{N+1,N+2,,2N}. (3.6)

    Note that all of the sides in the two regular N-polygons have the same size, and a represents the ratio of the sizes of the two regular N-polygons, so a=1. Choosing θl=4π with l=4, and then employing (3.6) with a=1 and N=2, we have

    (1a1ei(αθ))m5[|1a1ei(αθ)|2+(h5h)2]32+1k2(1ei(θkθ))m[|1ei(θkθ)|+h2]32+3ll4(1eiθll)ml|1eiθll|3=λλc0eiθ,  where  l{3,4}. (3.7)

    Combining N=2, (3.7) and the definition of circulant matrix C in (2.14),

    CM=˜b1˜ξ1˜b2˜ξ2,

    where

    ˜b1=λ(1a1ei(αθ))m5[|1a1ei(αθ)|2+(h5h)2]321k2(1ei(θkθ))m[|1ei(θkθ)|+h2]32,
    ˜b2=λeiθc0=λeiθaeiθ(m4m3)+a1m2N+1eiαm1+m2+m3+m4+m5,

    M=(m3,m4)T, ˜ξ1=(1,1)T, and ˜ξ2=(ρ1,ρ21)T=(1,1)T. Thus, we have

    (˜b1˜b2ρ1˜b1˜b2ρ2)=(01ρ1|1ρ1|31ρ1|1ρ1|30)(m3m4).

    Then, by ρ1=ρ1+2=ρ1=1, one computes that ˜b2=(m4m3)/8.

    On the other hand, by (3.5), we have

    m4m38=˜b2=λeiθc0=λa(m4m3)+a1m2N+1ei(αθ)2m+m3+m4+m5R.

    In addition, according to lines 1-8 of page 109 of [7], we have λ>0 for Definition 1.1. Combining aR and a10, one computes that Im(ei(αθ))=0, i.e., sin(αθ)=0, which contradicts with cos(θα)=0. Thus, cos(θα)=0 is impossible, which implies that for the spatial twisted central configuration with N=2, we deduce the conclusion that a1=0.

    (ii) N3:

    For (3.4), if we let β=θα and choose l=N+1,N+2, and N+3, then

    {cos(4πN+β)cos(2πN+β)=0,cos(6πN+β)cos(2πN+β)=0,

    and this is equivalent to

    {sinπNsin(β+3πN)=0,sin2πNsin(β+4πN)=0. (3.8)

    Observing that N3, sin(π/N)0, and sin(2π/N)0. Combining with (3.8), we have

    {β=k1π3πN,  k1Z,β=k2π4πN,  k2Z,

    which implies that k1π3π/N=k2π4π/N. So, (k2k1)π=π/N where positive integer N3, and this is impossible. Therefore, (3.4) does not hold. Then, for the spatial twisted central configuration with N3, we arrive at the conclusion that a1=0, too.

    Step 2. We prove that θ=sπ/N with s{0,1,,2N1}, and we divide the proof into two sub-steps.

    Step 2.1. We show that mN+1=mN+2==m2N.

    In fact, inserting a=1 and a1=0 into (3.6), we have

    m2N+1[1+(h2N+1h)2]32+1kN(1ei(θkθ))m[|1ei(θkθ)|2+h2]32+N+1ll2N(1eiθll)ml|1eiθll|3=λλc0ei(θl+θ),  where  l{N+1,N+2,,2N}. (3.9)

    Combining N2, m1=m2==mN=m, (3.9), ρd=eiθd with θd=2dπ/N and dZ, along with the definition of circulant matrix C in (2.14),

    CM=b1ξ1b2ξN, (3.10)

    where

    b1=λm2N+1[1+(h2N+1h)2]321kN(1ei(θkθ))m[|1ei(θkθ)|+h2]32,
    b2=λeiθc0=λeiθ1kNmkρk+N+1l2Nmlρleiθm2N+1+1kNmk+N+1l2Nml=λN+1l2Nmlρlm2N+1+1kNmk+N+1l2Nml, (3.11)

    M=(mN+1,mN+2,,m2N)T, ξ1=(1,1,,1)T and ξN=(ρN1,ρ2N1,,ρNN1)T.

    In the following, we divide the proof of mN+1=mN+2==m2N into three cases: N=2, N=3, and N4.

    (i) N=2:

    By (3.10) with N=2,

    (b1b2ρ1b1b2ρ2)=(01ρ1|1ρ1|31ρ1|1ρ1|30)(m3m4).

    Moreover, when N=2, it is easy to see that ρ1+ρ2=0 and ρ1=ρ1=1. Thus,

    2b2=1ρ1|1ρ1|3m41ρ1|1ρ1|3m3=1ρ1|1ρ1|3(m4m3),

    which implies that b2=(m4m3)/8. Thus, when N=2, then inserting ρ3=1, and ρ4=1 into (3.11) and combining with b2=(m4m3)/8, we have

    b2=λ(m4m3)1k5mk=(m4m3)8. (3.12)

    In what follows, we prove that for the spatial twisted central configuration with N=2, m3=m4, and we prove it by contradiction. We assume m3m4.

    In fact, on the one hand, by m3m4, m1=m2=m, and (3.12), we have

    λ=(2m+m3+m4+m5)8. (3.13)

    Moreover, thanks to N=2, a=1, a1=0, and the fourth equation of (3.5), one computes that

    c0=aeiθ(m4m3)+a1m5eiα2m+m3+m4+m5=eiθ(m4m3)2m+m3+m4+m5. (3.14)

    Summing the equations of the first part of (3.2) over k=1 and k=2, by N=2, a1=0, m1=m2=m, (3.13), and (3.14), we have

    1k2eiθkm5[1+h25]32+1k23l4(aei(θl+θ)eiθk)ml[|aei(θl+θ)eiθk|2+h2]32+1k21kk2(eiθkeiθk)mk|eiθkeiθk|3=λ1k2eiθk+2λc0=eiθ(m4m3)4.

    Then, combining eiθ1=eiθ1=1 (N=2), we have

    0+1k2eiθk3l4(aei(θlk+θ)1)ml[|aei(θlk+θ)1|2+h2]32+1k21kk2(eiθkeiθk)m|eiθkeiθk|3=1k2eiθk3l4(aei(θl+θ)1)ml[|aei(θl+θ)1|2+h2]32+1k2eiθk1kk2(eiθkk1)m|eiθkk1|3=(eiθ11)m|eiθ11|31k2eiθk=0=eiθ(m4m3)4,

    i.e., m3=m4, and this contradicts with the assumption that m3m4. Hence, for the spatial twisted central configuration with N=2, we deduce that m3=m4.

    (ii) N=3:

    By (3.10),

    (b1b2ρ2b1b2ρ1b1b2ρ3)=(01ρ1|1ρ1|31ρ2|1ρ2|31ρ1|1ρ1|301ρ2|1ρ2|31ρ2|1ρ2|31ρ1|1ρ1|30)(m4m5m6). (3.15)

    From N=3, we have ρ1=ρ2 and ρ2=ρ1; thus, ρ1+ρ2+ρ3=0. Together with (3.15) and

    {Re(1ρ1|1ρ1|3)=Re(1ρ2|1ρ2|3),Im(1ρ1|1ρ1|3)=Im(1ρ2|1ρ2|3),

    there is

    3b1=3b1b2(ρ1+ρ2+ρ3)=1ρ1|1ρ1|3m5+1ρ2|1ρ2|3m6+1ρ1|1ρ1|3m4+1ρ2|1ρ2|3m6+1ρ2|1ρ2|3m4+1ρ1|1ρ1|3m5=(1ρ1|1ρ1|3+1ρ2|1ρ2|3)m4+(1ρ1|1ρ1|3+1ρ2|1ρ2|3)m5+(1ρ1|1ρ1|3+1ρ2|1ρ2|3)m6R,

    which implies that b1R.

    On the other hand, for N=3, Lemma 2.9 gives us information that there exist constants c1, c2, and c3 such that M=c1ξ1+c2ξ2+c3ξ3 where M=(m4,m5,m6)T. Thus, combining (3.10), we obtain

    c1λ1(C)ξ1+c2λ2(C)ξ2+c3λ3(C)ξ3=b1ξ1b2ξ3. (3.16)

    Then, it follows from (3.16) and Lemma 2.9 that c1λ1(C)ξ1=b1ξ1 and c3λ3(C)ξ3=b2ξ3.

    Employing Lemma 2.5, Lemma 2.7, ρ3=1, ρ4=ρ1, ρ1+ρ2=1, and |1ρ1|=|1ρ2|, we have

    {λ1(C)=1ρ1|1ρ1|3+1ρ2|1ρ2|3=1j21eiθj|1eiθj|3R,λ3(C)=(1ρ1)ρ2|1ρ1|3+(1ρ2)ρ1|1ρ2|3=ρ2ρ3+ρ1ρ3|1ρ1|3=3|1ρ1|3R. (3.17)

    Then, thanks to b1R, ξ1=(1,1,,1)T, and c1λ1(C)ξ1=b1ξ1, one computes that c1R. In what follows, we will prove that c2R and c3R.

    By ξ1=(ρ0,ρ0,ρ0), ξ2=(ρ1,ρ2,ρ3), ξ3=(ρ2,ρ1,ρ3), and M=c1ξ1+c2ξ2+c3ξ3 with N=3, we have

    {Im(c1ρ0+c2ρ1+c3ρ2)=0,Im(c1ρ0+c2ρ2+c3ρ1)=0,Im(c1ρ0+c2ρ3+c3ρ3)=0. (3.18)

    Based upon ρ0=ρ3=1, c1R, and the third equation of (3.18), we have c2+c3R.

    Employing ρ0=1, c1R, and the first equation of (3.18), we see that c2ρ1+c3ρ2R. Note that

    c2ρ1+c3ρ2=(c2cos2π3+ic2sin2π3)+(c3cos4π3+ic3sin4π3)=(c2+c3)cos2π3+(c2c3)isin2π3,

    and then c2=c3. Therefore, with the help of c2+c3R, we have c3R.

    In virtue of (3.17), λ3(C)R. Moreover, combining c3R and c3λ3(C)ξ3=b2ξ3 where ξ3=(ρ2,ρ4,ρ6)T=(ρ2,ρ1,ρ3)T, one computes that b2R.

    By now, for (3.15), by the accumulated facts b1R, b2R, |1ρ1|=|1ρ2|, ρ1=ρ2, ρ2=ρ1, and Im(ρ1)=Im(ρ2), we have m4=m5=m6.

    (iii) N4:

    Lemma 2.9 gives us information that there exist constants ˜c1,˜c2,,˜cN such that M=˜c1ξ1+˜c2ξ2++˜cNξN where M=(mN+1,mN+2,,m2N)T. We can regard (3.10) as (3.11) of [26]; moreover, we regard C, b1, and b2 of this paper as Aα, Nk=1mk, and Nk=1mkqk of [26], respectively. Then, combining N4 and Lemmas 2.8–2.10, similar to the procedure of Case 2.1 on pages 6–7 of [26], we obtain mN+1=mN+2==m2N.

    Step 2.2. Based on Step 1 and Step 2.1, we prove that θ=sπ/N with s{0,1,,2N1}.

    Inserting m1=m2==mN, a1=0, and mN+1=mN+2==m2N into the second equality of (1.3), we have c0=0. Then, with the help of the first equation of (3.2), for any k{1,2,,N}, we obtain

    m2N+1[1+h22N+1]32+N+1l2N(aei(θlk+θ)1)ml[|aei(θlk+θ)1|2+h2]32+1kkN(eiθkk1)mk|eiθkk1|3=λR. (3.19)

    For any k{1,2,,N}, it follows from Remark 2.2 that the mapping

    {1,2,,N}{k}σ2{2πN,4πN,,2(N1)πN},

    where

    σ2(k)=[(kk)(modN)]2πN, k{1,2,,N}{k},

    is a bijection. Thus, by the procedure of obtaining (2.6), and θd=2πd/N with dZ, we have

    {kk1kNsinθkk|22cosθkk|3=1jN1sinθj|22cosθj|3=0, k{1,2,,N},kk1kNcosθkk1|22cosθkk|3=1jN1cosθj1|22cosθj|3, k{1,2,,N},kk1kN1|22cosθkk|3=1jN11|22cosθj|3, k{1,2,,N}, (3.20)

    and then

    1kkNeiθkk1|eiθkk1|3R. (3.21)

    Combining (3.19) with (3.21), we have

    Im(N+1l2Naei(θlk+θ)1[|aei(θlk+θ)1|2+h2]32)=0,  where  h>0  and  k{1,2,,N}. (3.22)

    For any l{N+1,N+2,,2N}, in (2.5), if we let ˜s=0, then the mapping

    {{1,2,,N}σ1{θ,2πN+θ,,2(N1)πN+θ},σ1(k)=[(lk)(modN)]2πN+θ, k{1,2,,N}

    is a bijection, too. Hence, we have

    {1kNacos(θlk+θ)1[1+a22acosθ(lk+πN)+h2]32=1jNacos(θj+θ)1[1+a22acos(θj+θ)+h2]32,1kNasin(θlk+θ)[1+a22acos(θlk+θ)+h2]32=1jNasin(θj+θ)[1+a22acos(θj+θ)+h2]32. (3.23)

    Using the definition of f(θ) in (2.1), (3.22), and the second equation of (3.23), we can see that

    f(θ)=1jNasin(θj+θ)[|1+a22acos(θj+θ)|2+h2]32=0. (3.24)

    On the one hand, if θ(2˜sπ/N,2˜sπ/N+π/N) where ˜s{0,1,,N1}, then by Lemmas 2.1-2.2, there is f(θ)>0, which contradicts (3.24).

    On the other hand, if θ(2˜sπ/N+π/N,2˜sπ/N+2π/N) where ˜s{0,1,,N1}, then by Lemma 2.1, we have f(θ)=f(θ)=f(θ+2π/N) and θ+2π/N(2˜sπ/N,2˜sπ/N+π/N). Therefore, it follows from Lemma 2.2 that f(θ)<0, which also contradicts (3.24).

    Thus, combining θ[0,2π), we conclude that the twist angle must be θ=2˜sπ/N or θ=2˜sπ/N+π/N with ˜s{0,1,,N1}, and then θ=sπ/N with s{0,1,,2N1}.

    Step 3. We show that m1=m2==mN=mN+1=mN+2==m2N.

    Based on the first part of Step 2, we can assume that mN+1=mN+2==m2N:=bm where constant b>0. By the assumption that m1=m2==mN:=m, it suffices to show that the value of b can only take b=1, and we prove it by contradiction. We assume that b1.

    Thanks to a=1, m1=m2==mN:=m, a1=0, and mN+1=mN+2==m2N:=bm, for (3.2) we have

    {Nm2N+1h2N+1[1+h22N+1]32+N+1l2Nbmh[|ei(θlk+θ)1|2+h2]32=λh0,Nm2N+1(h2N+1h)[1+(h2N+1h)2]321kNmh[|ei(θklθ)1|2+h2]32=λ(h0h),Nmh2N+1[1+h22N+1]32+Nbm(h2N+1h)[1+(hh2N+1)2]32=λ(h2N+1h0), (3.25)

    where k{1,2,,N} and l{N+1,N+2,,2N}. Combining the first and second equations of (3.25) with (2.7),

    {Nm2N+1h2N+1[1+h22N+1]32+1jNbmh[|ei(θj+θ)1|2+h2]32=λh0,Nm2N+1(h2N+1h)[1+(h2N+1h)2]321jNmh[|ei(θj+θ)1|2+h2]32=λ(h0h),Nmh2N+1[1+h22N+1]32+Nbm(h2N+1h)[1+(h2N+1h)2]32=λ(h2N+1h0). (3.26)

    Let

    {ˆx=Nh2N+1[1+h22N+1]32,y=N(h2N+1h)[1+(h2N+1h)2]32,z=1jNmh[|ei(θj+θ)1|2+h2]32. (3.27)

    Thus, (3.26) can be simplified into

    {m2N+1ˆx+bz=λh0,m2N+1yz=λ(h0h),mˆx+bmy=λ(h2N+1h0). (3.28)

    On the one hand, by (3.28), we see that

    {m2N+1ˆxm2N+1y+(b+1)z=λh,mˆx+bmy=λ(h2N+1h0), (3.29)

    and

    {m2N+1ˆx+m2N+1y+(b1)z=2λh0λh,mˆx+bmy=λ(h2N+1h0). (3.30)

    Then, it follows from (3.29), (3.30), and b1 that

    λh0m2N+1+λm2N+1h2N+1λmh+m(b+1)zmm2N+1(b+1)=y=λh0m2N+12λmh0+λm2N+1h2N+1+λmh+m(b1)zmm2N+1(b1).

    Thus,

    h0=bmh+m2N+1h2N+1m2N+1+m+bm.

    Combining with

    h0=N+1l2Nmlh+m2N+1h2N+1m2N+1+1kNmk+N+1l2Nml=bNmh+m2N+1h2N+1m2N+1+Nm+bNm, (3.31)

    we have

    bmh+m2N+1h2N+1m2N+1+m+bm=bNmh+m2N+1h2N+1m2N+1+Nm+bNm,

    which implies that (1+b)h2N+1=bh. Moreover, with the help of (3.31), one computes that h0=h2N+1. Then, it follows from the second equation of (3.29) that x=by. Hence, with the aid of (3.27) and (1+b)h2N+1=bh, we have

    bh1+b[1+b2h2(1+b)2]32=bh1+b[1+h2(1+b)2]32,

    that is, b=±1, which contradicts b>0 and the assumption b1. So, b=1, and we arrive at the conclusion that m1=m2==mN=mN+1=mN+2==m2N.

    Step 4. We prove that h2N+1=h/2.

    In virtue of (3.30), we have

    m(2λh0λh)=m2N+1λ(h2N+1h0).

    Note that lines 1–8 of page 109 of [7] show us that in Definition 1.1, λ must be a positive number. Then, combining the last equality of (1.3) and m1=m2==mN=mN+1=mN+2==m2N=m,

    m(2Nmh+2h2N+1m2N+12Nm+m2N+1h)=m2N+1(Nmh+h2N+1m2N+12Nm+m2N+1+h2N+1),

    which implies that N=1 or h=2h2N+1. Combining with N2, for the spatial twisted central configuration, we have h2N+1=h/2.

    Remark 3.1. The proof of Step 1 is independent of the condition that a=1. That is, for the twisted central configuration of the (2N+1)-body problem with the assumption that m1=m2==mN, and without the assumption that a=1, we have a1=0. That is, the (2N+1)-th mass must be in the vertical line of the two paralleled planes containing the two regular N-polygons, respectively, and the vertical line segment passes through the geometric centers of the two regular N-polygons.

    We divide the proof into two steps.

    Step 1. Based on the assumptions that θ=sπ/N with s{0,1,,2N1}, a=1, h2N+1=h/2, a1=0, and m1=m2==mN=mN+1=mN+2==m2N=m, we show that if there exists a constant λR such that

    {m2N+1[1+h24]32+1jN(1ei(θj+θ))m[|ei(θj+θ)1|2+h2]32+1jN(1eiθj)m|eiθj1|3=λ,m2N+1[1+h24]32+1jN2m[|ei(θj+θ)1|2+h2]32=λ, (3.32)

    then the 2N+1 masses can form a central configuration.

    In fact, by (1.3), in this situation we get

    {c0=1kNmρk+meiθN+1l2Nρl2Nm+m2N+1=0,h0=N+1l2Nmh+m2N+1h2N+12Nm+m2N+1=Nmh+m2N+1h22Nm+m2N+1=h2. (3.33)

    Employing a=1, h2N+1=h/2, a1=0, m1=m2==mN=mN+1=mN+2==m2N=m, (3.33) and

    1kNeiθk=N+1l2Nei(θl+θ)=0,

    we see that (3.2) holds if and only if

    {(eiθk,h2)mm2N+1[1+h24]32+N+1l2N(ei(θl+θ)eiθk,h)m2[|ei(θl+θ)eiθk|2+h2]32+1kkN(eiθkeiθk,0)m2|eiθkeiθk|3=λm(eiθk,h2),(ei(θl+θ),h2)mm2N+1[1+h24]32+1kN(eiθkei(θl+θ),h)m2[|eiθkei(θl+θ)|2+h2]32+N+1ll2N(ei(θl+θ)ei(θl+θ),0)m2|ei(θl+θ)ei(θl+θ)|3=λm(ei(θl+θ),h2). (3.34)

    Therefore, it suffices to verify (3.34) holds. Clearly, (3.34) is equivalent to

    {(1,h2)m2N+1[1+h24]32+N+1l2N(ei(θlk+θ)1,h)m[|ei(θlk+θ)1|2+h2]32+1kkN(eiθkk1,0)m|eiθkk1|3=λ(1,h2),(1,h2)m2N+1[1+h24]32+1kN(ei(θklθ)1,h)m[|ei(θklθ)1|2+h2]32+N+1ll2N(eiθll1,0)m|eiθll1|3=λ(1,h2). (3.35)

    On the other hand, for any l{N+1,N+2,,2N}, it follows from Remark 2.2 that the mapping

    {N+1,N+2,,2N}σ3{2πN,4πN,,2(N1)πN},

    where

    σ3(l)=[(ll)(modN)]2πN, l{N+1,N+2,,2N}{l},

    is a bijection. Thus, by θd=2πd/N with dZ, we have

    {llN+1l2Nsinθll|22cosθll|3=1jN1sinθj|22cosθj|3, l{N+1,N+2,,2N},llN+1l2Ncosθll1[|22cosθll|3=1jN1cosθj1|22cosθj|3, l{N+1,N+2,,2N},llN+1l2N1|22cosθll|3=1jN11|22cosθj|3, l{N+1,N+2,,2N}.

    Together with (3.20), for any k{1,2,,N} and any l{N+1,N+2,,2N}, it follows that

    kk1kN(eiθkk1)m|eiθkk1|3=1jN1(1eiθj)m|eiθj1|3=llN+1l2N(eiθll1)m|eiθll1|3. (3.36)

    Employing Lemma 2.3, (3.35), and (3.36), we conclude that if there exists a constant λR such that (3.32) holds, then by Definition 1.1, the 2N+1 masses form a central configuration.

    Step 2. We prove the existence of the spatial twisted central configuration, i.e., we prove the existence of λ of Step 1.

    Define the function g as follows:

    g(h)=121jN(1+ei(θj+θ))m[|ei(θj+θ)1|2+h2]32+121jN1(1eiθj)m|eiθj1|3, (3.37)

    where h>0 and θ=sπ/N with s{0,1,,2N1}. Thanks to Lemmas 2.3, 2.5, and (3.37), we see that g(h)R, which implies that

    g(h)=121jN(1+cos(θj+θ))m[2+h22cos(θj+θ)]32+121jN1(1cosθj)m|22cosθj|32. (3.38)

    In what follows, we prove that there exists h=ˉh(N) such that g(ˉh(N))=0. Note that θ=sπ/N with s{0,1,,2N1} is equivalent to θ=2˜sπ/N or θ=2˜sπ/N+π/N with ˜s{0,1,,N1}. We divide the following proof into two cases.

    Case 1. θ=2˜sπ/N with ˜s{0,1,,N1}.

    On the one hand, since m>0, θ=2˜sπ/N, and 1+cosθj0 with jZ, we have

    121jN(1+cos(θj+θ))m[2+h22cos(θj+θ)]32=121+˜sj+˜sN+s(1+cosθj+˜s)m[2+h22cosθj+˜s]32=121jN(1+cosθj)m[2+h22cosθj]32<0. (3.39)

    Moreover, if h0+, then

    121jN(1+cosθj)m[2+h22cosθj]32,

    which implies that when the twist angle is θ=2˜sπ/N with ˜s{0,1,,N1}, there exists h=h1(N) such that g(h1(N))<0.

    On the other hand, notice that if h+, then

    121jN(1+cosθj)m[2+h22cosθj]320.

    Then, by (3.38)–(3.39), we see that when the twist angle is θ=2˜sπ/N with ˜s{0,1,,N1}, there exists h=h2(N) such that g(h2(N))>0.

    Hence for the case of θ=2˜sπ/N with ˜s{0,1,,N1}, employing the fact that g is a continuous function, there exists h=ˉh(N) such that g(ˉh(N))=0.

    Case 2. θ=2˜sπ/N+π/N with ˜s{0,1,,N1}.

    In this case, by (3.38), ˜s{0,1,,N1}, and θd=2πd/N with dZ, we have

    g(h)=m2[1j+˜sN+˜s1+cos(θj+˜s+πN)[2+h22cos(θj+˜s+πN)]32+1jN11cosθj|22cosθj|32]=m2[1jN1+cos(θj+πN)[2+h22cos(θj+πN)]32+1jN11cosθj|22cosθj|32]. (3.40)

    Then, it follows from (3.40) and Lemma 2.5 that

    g(h)=m2[1jN1+cos(θj+πN)[2+h22cos(θj+πN)]32+1jN11eiθj|1eiθj|32]. (3.41)

    If h0, then with the help of (3.41) and Lemma 2.6, limh0g(h)<0, which implies that when the twist angle is θ=2˜sπ/N+π/N with ˜s{0,1,,N1}, there exists h=h3(N) such that g(h3(N))<0.

    On the other hand, if h+, then

    1jN1+cos(θj+πN)[2+h22cos(θj+πN)]320,

    which implies that when the twist angle is θ=2˜sπ/N+π/N with ˜s{0,1,,N1}, there exists h=h4(N) such that g(h4(N))>0.

    Therefore, for the case of θ=2˜sπ/N+π/N with ˜s{0,1,,N1}, combining with the continuity of function g, there exists h=ˉh(N) such that g(ˉh(N))=0.

    By now, Cases 1–2 show us that when θ=sπ/N with s{0,1,,2N1}, there exists h=ˉh(N) such that g(ˉh(N))=0, which implies that

    121jN(1ei(θj+θ))m[|ei(θj+θ)1|2+(ˉh(N))2]32+121jN1(1eiθj)m|eiθj1|3=1jNm[|ei(θj+θ)1|2+(ˉh(N))2]32.

    Moreover, note that

    1jNm[|ei(θj+θ)1|2+(ˉh(N))2]32R.

    Then, there is a constant λR such that

    121jN(1ei(θj+θ))m[|ei(θj+θ)1|2+(ˉh(N))2]32+121jN(1eiθj)m|eiθj1|3=12λ12m2N+1[1+(ˉh(N))24]32=1jNm[|ei(θj+θ)1|2+(ˉh(N))2]32>0,

    i.e.,

    {12m2N+1[1+(ˉh(N))24]32+121jN(1ei(θj+θ))m[|ei(θj+θ)1|2+(ˉh(N))2]32+121jN(1eiθj)m|eiθj1|3=12λ,12m2N+1[1+(ˉh(N))24]32+1jNm[|ei(θj+θ)1|2+(ˉh(N))2]32=12λ,

    which implies that there exists a constant λR such that (3.32) holds. Then, by Step 1, the 2N+1 masses form a central configuration.

    For the spatial twisted central configuration, (3.2) holds. Moreover, note that m1=m2==mN=m and a=1. Then, all the assumptions of Theorem 1.1 are satisfied, so we have a1=0, h2N+1=h/2, and m1=m2==mN=mN+1==m2N=m. Thus, c0=0, h0=h/2, and q2N+1=(0+0i,h/2). Thus, in the following, it suffices to prove the uniqueness of h.

    In fact, by (3.36) and the first equation of (3.2), for any k{1,2,,N} and any l{N+1,N+2,,2N}, one computes that

    {m2N+1[1+h24]32h2+N+1l2N[(1ei(θlk+θ))m][|1ei(θlk+θ)|2+h2]32h2+1jN1[(1eiθj)m]|1eiθj|3h2=λh2,h2m2N+1[1+h24]32+1jNmh[|1ei(θj+θ)|2+h2]32=λh0=λh2,

    where h>0 and θ[0,2π). Hence, it follows from Lemma 2.3 that

    {m2N+1h2[1+h24]32+1jN(1ei(θj+θ))mh2[|1ei(θj+θ)|2+h2]32+1jN1(1eiθj)mh2|1eiθj|3=λh2,h2m2N+1[1+h24]32+1jNmh[|1ei(θj+θ)|2+h2]32=λh2. (4.1)

    Let

    {ˉx=1jN11eiθj|1eiθj|3,ˉy(h)=1jNcos(θj+θ)[22cos(θj+θ)+h2]32,ˉz(h)=1jN1[22cos(θj+θ)+h2]32. (4.2)

    Combining λR, hR, (4.1), and (4.2), we have

    {h2m2N+1[1+h24]32+mh2[ˉz(h)ˉy(h)]+mh2ˉx=λh2,h2m2N+1[1+h24]32+mhˉz(h)=λh2,

    which implies that

    mh2[ˉz(h)ˉy(h)]mhˉz(h)+mh2ˉx=λh2,

    and then we obtain ˉx=ˉy(h)+ˉz(h). Hence, if the 2N+1 masses form a central configuration, then the distance h must satisfy that ˉx=ˉy(h)+ˉz(h). Next, we prove the uniqueness of the distance h.

    We take G(h)=ˉy(h)+ˉz(h)ˉx for h>0, and by (4.2), it is easy to verify that G(h)<0. Moreover, note that m1=m2==mN=m and a=1. Then, employing Theorem 1.1, we obtain that the existence of the central configuration implies that θ=sπ/N with s{0,1,,2N1} holds. Then, due to the rotational symmetry of the central configuration, in order to prove Theorem 1.2, it suffices to consider the following two cases: θ=0 and θ=π/N.

    Case 1. θ=0.

    Lemma 2.5 shows that

    ˉx=1jN11eiθj|1eiθj|3R,

    and then by (4.2),

    G(h)=ˉy(h)+ˉz(h)ˉx=1jN1+cosθj[22cosθj+h2]321jN11cosθj[22cosθj]32=1jN11+cosθj[22cosθj+h2]32+2h31jN11cosθj[22cosθj]32. (4.3)

    Employing (4.3), one verifies that there exist a small enough constant h=ˉh>0 and a big enough constant h=˜h>0 such that G(ˉh)>0 and G(˜h)<0. Then, combining the continuity and monotonicity of function G, for the case of θ=0, there is a unique h=ˆh>0, such that G(ˆh)=0, i.e., ˉx=ˉy(ˆh)+ˉz(ˆh).

    Case 2. θ=π/N.

    Employing (4.2), we have

    ˉy(h)+ˉz(h)=1jN1+cos(θj+θ)[22cos(θj+θ)+h2]32=1jN1+cos(θj+πN)[22cos(θj+πN)+h2]32=1jN1+cos(θj+πN)[22cos(θj+πN)+h2]32.

    By Lemma 2.6, limh0G(h)>0. Furthermore, by Lemma 2.5, we see that limh+G(h)<0. Thus, for the case of θ=π/N, due to the continuity and monotonicity of function G, there is a unique h=˘h>0, such that G(˘h)=0, i.e., ˉx=ˉy(˘h)+ˉz(˘h).

    Based upon Cases 1–2, there exists only one h>0 such that ˉx=ˉy(h)+ˉz(h), i.e., there is only one h>0 such that the 2N+1 masses form a central configuration. Moreover, combining the other two conclusions that a1=0 and h2N+1=h/2, we obtain that q2N+1=(a1eiα,h2N+1) is unique, i.e., there are only one positive distance h between the two paralleled regular N-polygons and only one position q2N+1 of the (2N+1)-th mass such that the 2N+1 masses form a central configuration.

    The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

    The authors sincerely thank the referees and the editors for their valuable comments and suggestions, and the authors also express their sincere gratitude to professor Shiqing Zhang for his helpful discussions and suggestions. This work was supported by the National Natural Science Foundation of China (Grant Nos. 12361022 and 12371163), the China Postdoctoral Science Foundation (Grant No. 2021M700968), the Guizhou Provincial Basic Research Program (Natural Science, Grant Nos. ZK[2021]003 and ZK[2023]138), and the Natural Science Research Project of the Department of Education of Guizhou Province (Grant No. QJJ2022047).

    The authors declare there is no conflict of interest.



    [1] A. Wintner, The analytical foundations of celestial mechanics, Princeton University Press, New Jersey, 1941.
    [2] L. Euler, De moto rectilineo trium corporum se mutuo attrahentium, Novi. Comm. Acad. Sci. Imp. Petrop., 11 (1767), 144–151.
    [3] J. Lagrange, Essai sur le problème des trois crops, Œuvres, 6 (1772), 229–324.
    [4] L. Perko, E. Walter, Regular polygon solutions of the N-body problem, Proc. Amer. Math. Soc., 94 (1985), 301–309. https://doi.org/10.1090/S0002-9939-1985-0784183-1 doi: 10.1090/S0002-9939-1985-0784183-1
    [5] M. Corbera, C. Valls, On centered co-circular central configurations of the n-body problem, J. Dyn. Differ. Equ., 31 (2019), 2053–2060. https://doi.org/10.1007/s10884-018-9699-2 doi: 10.1007/s10884-018-9699-2
    [6] W. Li, Z. Q. Wang, The relationships between regular polygon central configurations and masses for Newtonian N-body problems, Phys. Lett. A, 377 (2013), 1875–1880. https://doi.org/10.1016/j.physleta.2013.05.044 doi: 10.1016/j.physleta.2013.05.044
    [7] J. Llibre, R. Moeckel, C. Simó, Central configurations, periodic orbits, and Hamiltonian systems, Birkhäuser, New York, 2015. https://doi.org/10.1007/978-3-0348-0933-7
    [8] D. G. Saari, On the role and the properties of n body central configurations, Celest. Mech., 21 (1980), 9–20. https://doi.org/10.1007/bf01230241 doi: 10.1007/bf01230241
    [9] S. Q. Zhang, Q. Zhou, Periodic solutions for planar 2N-body problems, Proc. Amer. Math. Soc., 131 (2003), 2161–2170. https://doi.org/10.1090/S0002-9939-02-06795-3 doi: 10.1090/S0002-9939-02-06795-3
    [10] R. Moeckel, C. Simó, Bifurcation of spatial central configurations from planar ones, SIAM. J. Math. Anal., 26 (1995), 978–998. https://doi.org/10.1137/S0036141093248414 doi: 10.1137/S0036141093248414
    [11] E. Barrabés, J. M. Cors, On central configurations of the κn-body problem, J. Math. Anal. Appl., 476 (2) (2019), 720–736. https://doi.org/10.1016/j.jmaa.2019.04.010
    [12] M. Corbera, J. Delgado, J. Llibre, On the existence of central configurations of p nested n-gons, Qual. Theory Dyn. Syst., 8 (2009), 255–265. https://doi.org/10.1007/s12346-010-0004-y doi: 10.1007/s12346-010-0004-y
    [13] F. Zhao, J. Chen, Central configurations for (pN+gN)-body problems, Celest. Mech. Dyn. Astr., 121 (1) (2015), 101–106. https://10.1007/s10569-014-9593-0
    [14] J. Lei, M. Santoprete, Rosette central configurations, degenerate central configurations and bifurcations, Celest. Mech. Dyn. Astr., 94 (2006), 271–287. https://doi.org/10.1007/s10569-005-5534-2 doi: 10.1007/s10569-005-5534-2
    [15] J. Llibre, L. F. Mello, Triple and quadruple nested central configurations for the planar n-body problem, Physica D, 238 (2009), 563–571. https://doi.org/10.1016/j.physd.2008.12.014 doi: 10.1016/j.physd.2008.12.014
    [16] M. Marchesin, A family of three nested regular polygon central configurations, Astrophys. Space Sci., 364 (2019), 1–12. https://doi.org/10.1007/s10509-019-3648-3 doi: 10.1007/s10509-019-3648-3
    [17] A. Siluszyk, On a class of central configurations in the planar 3n-body problem, Math. Comput. Sci., 11 (2017), 457–467. http://dx.doi.org/10.1007/s11786-017-0309-1 doi: 10.1007/s11786-017-0309-1
    [18] Z. Q. Wang, F. Y. Li, A note on the two nested regular polygonal central configurations, Proc. Amer. Math. Soc., 143 (2015), 4817–4822. https://doi.org/10.1090/S0002-9939-2015-12618-4 doi: 10.1090/S0002-9939-2015-12618-4
    [19] Z. F. Xie, G. Bhusal, H. Tahir, Central configurations in the planar 6-body problem forming two equilateral triangles, J. Geom. Phys., 153 (2020), 1–11. https://doi.org/10.1016/j.geomphys.2020.103645 doi: 10.1016/j.geomphys.2020.103645
    [20] S. Q. Zhang, C. R. Zhu, Central configurations consist of two layer twisted regular polygons, Sci. China Ser. A, 45 (2002), 1428–1438. https://doi.org/10.1007/BF02880037 doi: 10.1007/BF02880037
    [21] X. Yu, S. Q. Zhang, Central configurations formed by two twisted regular polygons, J. Math. Anal. Appl., 425 (2015), 372–380. https://doi.org/10.1016/j.jmaa.2014.12.023 doi: 10.1016/j.jmaa.2014.12.023
    [22] J. Chen, J. B. Luo, Solutions of regular polygon with an inner particle for Newtonian N+1-body problem, J. Differ. Equ., 265 (2018), 1248–1258. https://doi.org/10.1016/j.jde.2018.04.004 doi: 10.1016/j.jde.2018.04.004
    [23] T. Ouyang, Z. F. Xie, S. Q. Zhang, Pyramidal central configurations and perverse solutions, Electron. J. Differ. Equ., 1 (2004), 1–9.
    [24] X. Yu, S. Q. Zhang, Twisted angles for central configurations formed by two twisted regular polygons, J. Differ. Equ., 253 (2012), 2106–2122. https://doi.org/10.1016/j.jde.2012.06.017 doi: 10.1016/j.jde.2012.06.017
    [25] M. Marcus, H. Minc, A survey of matrix theory and matrix inequalities, Allyn & Bacon, Boston, 1964.
    [26] L. Ding, J. L. Wei, S. Q. Zhang, The central configuration of the planar (N+1)-body problem with a regular N-polygon for homogeneous force laws, Astrophys. Space Sci., 367 (2022), 1–9. https://doi.org/10.1007/s10509-022-04095-w doi: 10.1007/s10509-022-04095-w
  • Reader Comments
  • © 2024 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(998) PDF downloads(44) Cited by(0)

Figures and Tables

Figures(7)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog