betaFeedback

WorksPhysics and Astronomy

A note on Gurzadyan theorem

Abstract

The issue and proof of Gurzadyan theorem are presented concisely, avoiding tedious and unnecessary calculations that would mask what is essential. The goal is to provide a good mathematical and physical understanding of the theorem, making you want to learn more about its use in cosmology.

Newton’s shell theorem

We will start with a presentation of Newton’s famous shell theorem, of which Gurzadyan’s theorem is an extension.

We consider masses uniformly distributed on a sphere SS at an areal density of mm. The sphere is centered on point CC and has radius bb. From the superposition principle, the overall gravitational potential and gravitational field at point PP are, respectively,

U(P,C)=∫S(C,b)dU(P,Q),where dU(P,Q)=m dS(Q) u(P,Q)and G(P,C)=−∇PU(P,C)(1)U(P,C)=\int_{S(C,b)}dU(P,Q), \quad\text{where } dU(P,Q)=m\,dS(Q)\,u(P,Q) \\ \text{and } \mathbf{G}(P,C)=-\nabla_P U(P,C) \tag*{(1)}

u(P,Q)u(P,Q) being a function of both positions of point PP and point QQ, the latter being on the sphere, and given by u(P,Q)=G/(PQ)u(P,Q)=G/(PQ) from Newton’s law of gravity (GG is the gravitational constant). Using Gauss’s theorem and considering the spherical symmetry of the distribution, it is easy to find the following result, first published by Newton in 1687 [1], and referred to as Newton’s shell theorem:

  • The distribution acts on a test mass at a point PP outside the sphere as if its mass M=4πb2mM = 4\pi b^{2}m was concentrated at its center CC, that is, G(P,C)=−GMCPCP3G(P,C) = -GM \frac{\mathbf{CP}}{CP^{3}};

  • the gravitational field is zero at any point PP (strictly) inside the sphere.

As done in [2], it is relevant to ask whether this result is a direct consequence of Newton’s gravitational law (the inverse square law for the field). Without presupposing the form of u(P,Q)u(P,Q), we write the potential as:

U(P,C)=MuˉP(C),withuˉP(C)=14πb2∫S(C,b)dS(Q) u(P,Q)(2)U(P,C) = M\bar{u}_{P}(C), \quad\text{with} \quad\bar{u}_{P}(C) = \frac{1}{4\pi b^{2}}\int_{S(C,b)} dS(Q)\,u(P,Q) \tag*{(2)}

Defining uP(Q)=u(P,Q)u_{P}(Q) = u(P,Q), it becomes apparent that uˉP(C)\bar{u}_{P}(C) is the mean value of the function uP(Q)u_{P}(Q) at point CC, when this function is integrated over the sphere S(C,b)S(C,b). Under what conditions would we have uˉP(C)=uP(C)\bar{u}_{P}(C) = u_{P}(C)? This happens in two particular cases. One is trivial : when u(P,Q)u(P,Q) is constant. The other appears when PC≫bPC \gg b, because the distribution is then seen from point PP as a point mass MM located at CC. However, this does not provide any information regarding the law of forces. Is there a solution for any finite bb? The answer is given by a well-known mathematical theorem (the “Mean value theorem”, see e.g. Ref. [3]) stating that the equality uˉP(C)=uP(C)\bar{u}_{P}(C) = u_{P}(C) is an exclusive property of harmonic functions :

uˉP(C)=uP(C)if and only ifΔCuP(C)=0(3)\bar{u}_{P}(C) = u_{P}(C) \quad\text{if and only if} \quad\Delta_{C}u_{P}(C) = 0 \tag*{(3)}

ΔC\Delta_{C} being the Laplacian operator with respect to the coordinates of point CC. From the physicist’s perspective, we assume that all mathematical operations necessary for a physical description are allowed. In appendix § 4 is given a presentation of this theorem for an euclidean space of dimension n≥3n \ge3, using the appropriate Green function and Dirac distribution.

If we admit the principle that interactions are central at the corpuscular level, from which it follows that u(P,C)u(P,C) depends on the positions of PP and CC only through their relative distance a=PCa = PC, i.e. u(P,C)=u(a)u(P,C) = u(a), then

ΔCu(P,C)=ΔPu(P,C)=1ad2da2[au(a)]=0(4)\Delta_{C}u(P,C) = \Delta_{P}u(P,C) = \frac{1}{a}\frac{d^{2}}{da^{2}}[au(a)] = 0 \tag*{(4)}

Hence, the unique admissible form u(a)=K1a+K2u(a) = \frac{K_{1}}{a} + K_{2}, K1K_{1} and K2K_{2} being constants, and the inverse-square law for the gravitational field, at least for a≥ba \ge b. Note that u(a)u(a) being harmonic, the mean property stated above does not depend on the value of b≤ab \le a, more precisely, the constants K1K_{1} and K2K_{2} corresponding to the outside of the sphere do not depend on bb. The first part of Newton’s shell theorem is thus fully explained. As will be seen now, the second part of this theorem is more subtly related to Newton’s gravitational law.

Assuming u(P,Q)=u(PQ)u(P,Q)=u(PQ), the mean value in Eq. (2) is expressed as

uˉP(C)=I(a,b)=12∫−11dx u(r),withr=a2+b2+2abx(5)\bar{u}_{P}(C)=I(a,b)=\frac{1}{2}\int_{-1}^{1}dx\,u(r),\quad\text{with}\quad r=\sqrt{a^{2}+b^{2}+2abx} \tag*{(5)}

Due to the symmetry I(a,b)=I(b,a)I(a,b)=I(b,a), it is obvious that if I(a,b)I(a,b) does not depend on bb when a≥ba\ge b, it cannot depend on aa when b≥ab\ge a, whatever the form of u(r)u(r). Consequently, inside the sphere, the potential is constant and the field is zero. Only the fact that uˉP(C)\bar{u}_{P}(C) does not depend on bb when b≤ab\le a is due to the inverse-square law, as seen previously. The spherical symmetry of the distribution is the final crucial point leading to the conclusion.

At this stage, it is tempting to check the result in Eq. (4). Taking s=r2s=r^{2} as the integration variable, Eq. (5) is rewritten as

I(a,b)=u(a)=14ab∫s1s2ds u(s),wheres2=(a+b)2, s1=(a−b)2,(6)I(a,b)=u(a)=\frac{1}{4ab}\int_{s_{1}}^{s_{2}}ds\,u(\sqrt{s}),\quad\text{where}\quad s_{2}=(a+b)^{2},\ s_{1}=(a-b)^{2}, \tag*{(6)}

and for a≥ba\ge b we have

∂∂b[2abu(a)]=2au(a)=12[u(s2)∂s2∂b−u(s1)∂s1∂b]=(a+b)u(a+b)+(a−b)u(a−b),or[(a+b)u(a+b)−au(a)]+[(a−b)u(a−b)−au(a)]=0(7)\begin{aligned} \frac{\partial}{\partial b}[2abu(a)]&=2au(a)=\frac{1}{2}\left[u(\sqrt{s_{2}})\frac{\partial s_{2}}{\partial b}-u(\sqrt{s_{1}})\frac{\partial s_{1}}{\partial b}\right]\\ &=(a+b)u(a+b)+(a-b)u(a-b),\quad\text{or}\\ [(a+b)u(a+b)-au(a)]+[(a-b)u(a-b)-au(a)]&=0 \tag*{(7)} \end{aligned}

which equation must be satisfied whatever the value of bb. Expanding in powers of bb its left member which is symmetrical with respect to the change of bb into −b-b, we find

∑k=1∞b2k(2k)!d2kda2k[au(a)]=0(8)\sum_{k=1}^{\infty}\frac{b^{2k}}{(2k)!}\frac{d^{2k}}{da^{2k}}[au(a)]=0 \tag*{(8)}

and, as expected, this is realized if and only if Δu=1ad2da2[au(a)]=0\Delta u=\frac{1}{a}\frac{d^{2}}{da^{2}}[au(a)]=0.

Gurzadyan’s theorem

Gurzadyan’s theorem is an extension of the first conclusion of Newton’s shell theorem, answering the question : what is the more general law of force such that the spherical symmetric mass distribution on S(C,b)S(C,b) acts on a test mass outside the sphere exactly as a point mass MM located at its center CC?

To our knowledge, this theorem has so far only received a truncated proof by Gurzadyan himself [4], or proofs based either on power-series methods [2, 5], or on a “perturbative method”, according to the author of Ref. [6], or in a way that does not highlight the essential points [7]. In the following we propose a concise and clear proof rid of unnecessary calculations.

The novelty in the hypothesis is that the equality uˉP(C)=u(P,C)\bar{u}_{P}(C)=u(P,C) is no longer required but replaced by

G(P,C)=−M∇PuˉP(C)=−M∇Pu(P,C) outside the sphere(9)\mathbf{G}(P,C)=-M\nabla_{P}\bar{u}_{P}(C)=-M\nabla_{P}u(P,C)\ \text{outside the sphere} \tag*{(9)}

This condition is satisfied in the three following situations : u=constantu=\text{constant}, uu is harmonic, and when a≫ba\gg b. For finite b/ab/a, we will proceed as follows.

Thanks to the spherical symmetry of the distribution, this condition is simplified into

∂∂a[uˉ(a,b)−u(a)]=0(10)\frac{\partial}{\partial a}\left[\bar{u}(a,b)-u(a)\right]=0 \tag*{(10)}

for a≥ba\ge b and whatever the value of $b. Note that here u(P,C)=u(a)u(P,C)=u(a) is supposed to be independent of bb while the mean value uˉ(a,b)\bar{u}(a,b) can depend on both aa and bb. Let us write

X=∂∂b[b∂∂a(uˉ(a,b)−u(a))]=∂∂a1a∂∂b[ab(uˉ(a,b)−u(a))]=0X=\frac{\partial}{\partial b}\left[b\frac{\partial}{\partial a}\left(\bar{u}(a,b)-u(a)\right)\right]=\frac{\partial}{\partial a}\frac{1}{a}\frac{\partial}{\partial b}\left[ab\left(\bar{u}(a,b)-u(a)\right)\right]=0

Then, repeating the calculation leading to Eq. (8), we find

X=∂∂a1a[∑k=1∞b2k(2k)!d2kda2k[au(a)]]=∑k=1∞b2k(2k)!d2k−1da2k−1Δu(a)=b22ddaΔu(a)+⋯(11)\begin{aligned} X&=\frac{\partial}{\partial a}\frac{1}{a}\left[\sum_{k=1}^{\infty}\frac{b^{2k}}{(2k)!}\frac{d^{2k}}{da^{2k}}\left[au(a)\right]\right] =\sum_{k=1}^{\infty}\frac{b^{2k}}{(2k)!}\frac{d^{2k-1}}{da^{2k-1}}\Delta u(a)\\ &=\frac{b^{2}}{2}\frac{d}{da}\Delta u(a)+\cdots \tag*{(11)} \end{aligned}

and conclude that Eq. (9) (i.e. X=0X=0) is realized if and only if

ddaΔu(a)=dda[1ad2da2[au(a)]]=0(12)\frac{d}{da}\Delta u(a)=\frac{d}{da}\left[\frac{1}{a}\frac{d^{2}}{da^{2}}\left[au(a)\right]\right]=0 \tag*{(12)}

Expressing the field as G(P,C)=−Mg(a)CPa\mathbf{G}(P,C)=-M g(a)\frac{CP}{a} with g(a)=−duda(a)g(a)=-\frac{du}{da}(a), Eq. (12) written in terms of g(a)g(a) and its derivatives with respect to aa is

a22g′′+ag′−g=0(13)\frac{a^{2}}{2}g'' + ag' - g = 0 \tag*{(13)}

which is identical to Eq. 4 of Ref. [4]. Integration of Eq. (12) gives the general expression of admissible functions u(a)u(a):

u(a)=K1a+K2+K3a22(14)u(a) = \frac{K_{1}}{a} + K_{2} + \frac{K_{3}a^{2}}{2} \tag*{(14)}

K1K_{1}, K2K_{2} and K3K_{3} being arbitrary constants. This is the final answer of Gurzadyan’s theorem: besides the Newtonian field ∝1/a2\propto1/a^{2}, the only field possessing the property required is of the Hookean type ∝a\propto a.

Note that taking u(P,Q)=r2u(P,Q)=r^{2} yields uˉ(P,C)=a2+b2\bar{u}(P,C)=a^{2}+b^{2}, depending on both aa and bb, which reflects the fact that the strict equality uˉ(P,C)=u(P,C)(=a2)\bar{u}(P,C)=u(P,C)(=a^{2}) cannot be realized if uu is not harmonic. However, the relation ∇uˉ(P,C)=∇u(P,C)\nabla\bar{u}(P,C)=\nabla u(P,C) holds for the Hookean function, but its field does not vanish inside the sphere, except at point CC.

From the physical point of view, it is important to remark that the Hookean field cannot be considered as a classical gravitational field or a Coulomb field, because in a region empty of sources the divergences of the latter are commonly considered to be zero, whereas the divergence of a Hookean field is never zero (Δr2=6\Delta r^{2}=6). In cosmology, Gurzadyan considered the Hookean field as a correction to the traditional gravitational field, that could explain in a natural way the appearance of the cosmological constant in Einstein equation of General Relativity [8, 9]. However, in Classical Physics, the divergences of fields are usually considered as cause and effect relationships connecting fields to their sources. According to this, the sources of the Hookean field would not only be located on the surface SS, but would actually fill the entire space uniformly. Then arises the question of the meaning of this strange Hookean field itself, or, if the previous interpretation is incorrect, that of the true relationship between the fields and their sources.

In the general case, the mean value uˉ(P,C)\bar{u}(P,C) does depend on both aa and bb and it seems difficult to achieve an extension of Gurzadyan’s theorem. However, various authors have considered the interesting case where this mean value appears in the form of the product u(a)w(b)u(a)w(b) [10, 11, 12]. The result is given in appendix A.

An extension of Gurzadyan’s theorem

Let’s look for what types of functions u(PQ)u(PQ) and w(b)w(b) could satisfy the relation

uˉ(P,C)=u(a)w(b)=14ab∫s1s2ds u(s)(15)\bar{u}(P,C)=u(a)w(b)=\frac{1}{4ab}\int_{s_1}^{s_2} ds\,u(\sqrt{s}) \tag*{(15)}

We have

∂∂b[2abu(a)w(b)]=2au(a)ddb[bw(b)]=12[u(s2)∂s2∂b−u(s1)∂s1∂b]=[(a+b)u(a+b)−au(a)]+[(a−b)u(a−b)−au(a)]+2au(a)(16)\begin{aligned} \frac{\partial}{\partial b}[2abu(a)w(b)] &= 2au(a)\frac{d}{db}[bw(b)] = \frac{1}{2}\left[u(\sqrt{s_2})\frac{\partial s_2}{\partial b}-u(\sqrt{s_1})\frac{\partial s_1}{\partial b}\right] \\ &= [(a+b)u(a+b)-au(a)]+[(a-b)u(a-b)-au(a)]+2au(a) \tag*{(16)} \end{aligned}

Hence,

au(a)[ddb[bw(b)]−1]=∑k=1∞b2k(2k)!d2kda2k[au(a)](17)au(a)\left[\frac{d}{db}[bw(b)]-1\right]=\sum_{k=1}^{\infty}\frac{b^{2k}}{(2k)!}\frac{d^{2k}}{da^{2k}}[au(a)] \tag*{(17)}

It is obvious that this equation is satisfied if and only if d2kda2k[au(a)]=ck[au(a)]\frac{d^{2k}}{da^{2k}}[au(a)]=c_k[au(a)] where ckc_k is constant. Thus, we must have

d2da2[au(a)]=c2[au(a)],\frac{d^2}{da^2}[au(a)]=c_2[au(a)],
d4da4[au(a)]=c4[au(a)]=d2da2[d2da2[au(a)]]=c22[au(a)],\frac{d^4}{da^4}[au(a)]=c_4[au(a)]=\frac{d^2}{da^2}\left[\frac{d^2}{da^2}[au(a)]\right]=c_2^2[au(a)],
d6da6[au(a)]=c23[au(a)],etc.\frac{d^6}{da^6}[au(a)]=c_2^3[au(a)],\quad\text{etc.}

and finally, Eq. (A.3) is realized only by functions uu and ww verifying

d2da2[au(a)]=c2[au(a)]andddb[bw(b)]−1=∑k=1∞c2kb2k(2k)!(18)\frac{d^2}{da^2}[au(a)]=c_2[au(a)]\quad\text{and}\quad\frac{d}{db}[bw(b)]-1=\sum_{k=1}^{\infty}\frac{c_2^k b^{2k}}{(2k)!} \tag*{(18)}

Integrating these equations gives the following forms.

  • If c2=0c_2=0,

u(a)=K1a+K2,w(b)=1+K3b(19)u(a)=\frac{K_1}{a}+K_2,\quad w(b)=1+\frac{K_3}{b} \tag*{(19)}
  • If c2>0c_{2}>0, we set c2=λ2c_{2}=\lambda^{2}. Then,

ddb[bw(b)]=1+∑k=1∞(λb)2k(2k)!=cosh⁡λb,andw(b)=sinh⁡λbλb+K3b,u(a)=K1exp⁡(−λa)a+K2exp⁡(λa)a(20)\begin{aligned} \frac{d}{db}[bw(b)] &= 1+\sum_{k=1}^{\infty}\frac{(\lambda b)^{2k}}{(2k)!}=\cosh\lambda b,\quad\text{and}\quad w(b)=\frac{\sinh\lambda b}{\lambda b}+\frac{K_{3}}{b},\\ u(a) &= K_{1}\frac{\exp(-\lambda a)}{a}+K_{2}\frac{\exp(\lambda a)}{a} \tag*{(20)} \end{aligned}
  • If c2<0c_{2}<0, we set c2=−ω2c_{2}=-\omega^{2}. Then,

ddb[bw(b)]=1+∑k=1∞(−1)k(ωb)2k(2k)!=cos⁡ωb,andw(b)=sin⁡ωbωb+K3b,u(a)=K1cos⁡(ωa)a+K2sin⁡(ωa)a(21)\begin{aligned} \frac{d}{db}[bw(b)] &= 1+\sum_{k=1}^{\infty}(-1)^{k}\frac{(\omega b)^{2k}}{(2k)!}=\cos\omega b,\quad\text{and}\quad w(b)=\frac{\sin\omega b}{\omega b}+\frac{K_{3}}{b},\\ u(a) &= K_{1}\frac{\cos(\omega a)}{a}+K_{2}\frac{\sin(\omega a)}{a} \tag*{(21)} \end{aligned}

In each case, K1K_{1}, K2K_{2} and K3K_{3} are constants with appropriate dimensions.

The physical content of this extension is that the field at a point PP outside the sphere is the same as that of a point source located at the center CC, the mass of which is no longer the total mass MM of the distribution, but the product of the latter by a dimensionless factor w(b)w(b) depending on the radius of the sphere: G(P,C)=−GM′∇Pu(P,C)\mathbf{G}(P,C)=-GM'\nabla_{P}u(P,C) with M′=Mw(b)M'=Mw(b). Among non-harmonic potentials satisfying this property, the Yukawa potential u(r)=e−λr/ru(r)=e^{-\lambda r}/r (λ>0\lambda>0) shows a good behavior for physical modelisations. In fact, it has a long history in Physics, particularly in Particle Physics [13]. Nowadays, it often appears in this form or modified forms in research works on gravitation theory [14, 15, 16], and the phenomenology of dark universe [17, 18].

Extension to nn dimensions

In an euclidean space of dimension n≥3n \ge3, the equivalent of Newton’s potential 1/r1/r in three dimensions is UN(r)=1/rn−2U_N(r)=1/r^{n-2}, where r=x12+x22+⋯+xn2r=\sqrt{x_1^2+x_2^2+\cdots+x_n^2}. It satisfies the equation

ΔUN=−knδ(r),where(22)\Delta U_N=-k_n\delta(\mathbf{r}), \qquad\text{where} \tag*{(22)}
  • Δ\Delta is the Laplace operator Δ=∑i=1n∂2∂xi2\Delta=\sum_{i=1}^{n}\frac{\partial^2}{\partial x_i^2}, which, when applied to a spherical symmetric function, reduces to

Δ=d2dr2+n−1rddr\Delta=\frac{d^2}{dr^2}+\frac{n-1}{r}\frac{d}{dr}
  • δ(r)=∏i=1nδ(xi)\delta(\mathbf{r})=\prod_{i=1}^{n}\delta(x_i) is the Dirac distribution for dimension nn, such that, for any volume VV, any point CC, and any sufficiently regular function FF

∫Vδ(rQ−rC)F(Q) dV(Q)=θV(C)F(C)(23)\int_V\delta(r_Q-r_C)F(Q)\,dV(Q)=\theta_V(C)F(C) \tag*{(23)}

where θV(C)=1\theta_V(C)=1 if CC is inside VV and θV(C)=0\theta_V(C)=0 if CC is outside VV;

  • kn=(n−2)Ωnk_n=(n-2)\Omega_n where Ωn=2πn/2Γ(n/2)\Omega_n=2\frac{\pi^{n/2}}{\Gamma(n/2)} is the total solid angle, Γ\Gamma being the Euler’s Gamma function.

Considering a ball BB of volume VV delimited by the sphere SS of radius bb and centered at CC, Green’s theorem states that for any functions (or even distributions) uu and vv, we have

∫S[v∇u−u∇v]⋅dS=∫B[vΔu−uΔv] dV(24)\int_S [v\nabla u-u\nabla v]\cdot d\mathbf{S}=\int_B[v\Delta u-u\Delta v]\,dV \tag*{(24)}

Let’s take v=(CQ)2−n−b2−nv=(CQ)^{2-n}-b^{2-n} for which v=0v=0 when QQ is on the surface SS and ΔQv=−knδ(CQ)\Delta_Qv=-k_n\delta(CQ). We have

∫S[v∇u−u∇v]⋅dS=b1−n(n−2)∫Su dS=∫B[r2−n−b2−n]Δu dV+knu(P,C)\int_S[v\nabla u-u\nabla v]\cdot d\mathbf{S}=b^{1-n}(n-2)\int_Su\,dS \\ =\int_B\left[r^{2-n}-b^{2-n}\right]\Delta u\,dV+k_nu(P,C)

Since dS=bn−1dΩndS=b^{n-1}d\Omega_n and uˉ(P,C)=1Ωn∫dΩn u(P,Q)\bar{u}(P,C)=\frac{1}{\Omega_n}\int d\Omega_n\,u(P,Q) being the mean value of uu at point CC, we obtain

uˉ(P,C′)−u(P,C)=1kn∫B[r2−n−b2−n]ΔQu(P,Q) dV(Q)(25)\bar{u}(P,C')-u(P,C)=\frac{1}{k_n}\int_B\left[r^{2-n}-b^{2-n}\right]\Delta_Q u(P,Q)\,dV(Q) \tag*{(25)}

with dV=rn−1drdΩndV=r^{n-1}drd\Omega_n. It is then obvious that uˉ(P,C)=u(P,C)\bar{u}(P,C)=u(P,C) if uu is harmonic, i.e. if ΔQu(P,Q)=0\Delta_Q u(P,Q)=0 everywhere in the volume VV (note that the factor r2−n−b2−nr^{2-n}-b^{2-n} prevent singularities when r=br=b).

Conversely, suppose that uˉ(P,C)=u(P,C)\bar{u}(P,C)=u(P,C) independently of bb. Then, taking bb as small enough, we can write

0=1kn∫B[r2−n−b2−n]Δu dV≈1knΔCu(P,C) Ωn∫0b[r−rn−1b2−n] dr=b22nΔCu(P,C)\begin{aligned} 0&=\frac{1}{k_n}\int_B\left[r^{2-n}-b^{2-n}\right]\Delta u\,dV\approx\frac{1}{k_n}\Delta_Cu(P,C)\,\Omega_n\int_0^b\left[r-r^{n-1}b^{2-n}\right]\,dr\\ &=\frac{b^2}{2n}\Delta_Cu(P,C) \end{aligned}

and therefore ΔCu(P,C)=0\Delta_Cu(P,C)=0, which means that uu must be harmonic. Considering here central functions, i.e. functions such that u(P,Q)=u(PQ)u(P,Q)=u(PQ), we have as well ΔPu(P,C)=ΔCu(P,C)=0\Delta_Pu(P,C)=\Delta_Cu(P,C)=0. The approximate relation

uˉ(C)−u(C)≈αb2ΔCu(C)(26)\bar{u}(C)-u(C)\approx\alpha b^2\Delta_Cu(C) \tag*{(26)}

for any function u(C)u(C) provides an interpretation of its Laplacian as a measure of the local anomaly of that function i.e. the gap between the value of a function at a given point and the average of its values in the vicinity of this point, see e.g. [19, 20, 21]. The numerical constant α\alpha is a priori specific to the closed surface surrounding CC and bb is a characteristic dimension of this surface.

More generally, the second member of Eq. (B.4) can be expressed as a series in powers of bb, [22, 23]. First, for central potential u(PQ)u(PQ), let’s rewrite Eq. (B.4) in the form

uˉ(P,C)−u(PC)=1knΔP∫dΩn∫0bφ0(r)u(PQ) drwith\bar{u}(P,C)-u(PC)=\frac{1}{k_n}\Delta_P\int d\Omega_n\int_0^b\varphi_0(r)u(PQ)\,dr\quad\text{with}
φ0(r)=r−rn−1b2−n(27)\varphi_0(r)=r-r^{n-1}b^{2-n} \tag*{(27)}

Then, let’s expand u(PQ)u(PQ) in the Taylor series

u(PQ)=u(PC)+∑k=1∞rkk!∑ip=1nx^i1⋯x^ik∂i1⋯∂iku(PC)u(PQ)=u(PC)+\sum_{k=1}^{\infty}\frac{r^k}{k!}\sum_{i_p=1}^{n}\hat{x}_{i_1}\cdots\hat{x}_{i_k}\partial_{i_1}\cdots\partial_{i_k}u(PC)

ends_mid=1 where xix_i are the cartesian coordinates of QQ relative to CC, r=CQr=CQ, x^i=xi/r\hat{x}_i=x_i/r and ∂i=∂/∂xi\partial_i=\partial/\partial x_i. Thus,

uˉ(P,C)−u(PC)=ΔP∑k=0∞Ik∑ip=1nTi1…ik∂i1⋯∂iku(PC)\bar{u}(P,C)-u(PC)=\Delta_P\sum_{k=0}^{\infty}I_k\sum_{i_p=1}^{n}T_{i_1\ldots i_k}\partial_{i_1}\cdots\partial_{i_k}u(PC)

where

Ik=1n−21k!∫0brkφ0(r) drI_k=\frac{1}{n-2}\frac{1}{k!}\int_0^b r^k\varphi_0(r)\,dr

and

Ti1…ik=⟨x^i1⋯x^ik⟩=1Ωn∫dΩn x^i1⋯x^ikT_{i_1\ldots i_k}=\langle\hat{x}_{i_1}\cdots\hat{x}_{i_k}\rangle=\frac{1}{\Omega_n}\int d\Omega_n\,\hat{x}_{i_1}\cdots\hat{x}_{i_k}

Note that, mathematically speaking, φ0(r)\varphi_0(r) is identically zero if n=2n=2. Consequently, the integral in IkI_k must be proportional to n−2n-2, thus cancelling the factor 1/(n−2)1/(n-2); it is also proportional to bk+2b^{k+2}. Due to the spherical symmetry, the tensor Ti1…ikT_{i_1\ldots i_k} is non-zero only if kk is an even number. In this case, its components whose indices are all different are zero. For a component to be non-zero, its indices must be associated in pairs, which results in the presence of Kronecker symbols like δipiq\delta_{i_p i_q}. Moreover, the tensor is completely symmetric. For example, we must have (implicit summation on repeated indices is assumed)

Tij=A2δij,withA2=1nδijTij=1n,sinceδijTij=⟨∑i=1nx^i2⟩=⟨1⟩=1T_{ij}=A_2\delta_{ij},\quad\text{with}\quad A_2=\frac{1}{n}\delta_{ij}T_{ij}=\frac{1}{n},\quad\text{since}\quad\delta_{ij}T_{ij}=\left\langle\sum_{i=1}^{n}\hat{x}_i^2\right\rangle=\langle1\rangle=1

and

Tijkl=A4[δijδkl+δikδjl+δilδjk]withA4=1n(n+2)T_{ijkl}=A_4\left[\delta_{ij}\delta_{kl}+\delta_{ik}\delta_{jl}+\delta_{il}\delta_{jk}\right]\quad\text{with}\quad A_4=\frac{1}{n(n+2)}

It is then obvious that the contraction Ti1…ik∂i1⋯∂ikT_{i_1\ldots i_k}\partial_{i_1}\cdots\partial_{i_k} leads to repeated action of the Laplace operator Δ\Delta:

Tij∂i∂j=A2Δ,Tijkl∂i∂j∂k∂l=3A4ΔΔ,…,T_{ij}\partial_i\partial_j=A_2\Delta,\quad T_{ijkl}\partial_i\partial_j\partial_k\partial_l=3A_4\Delta\Delta,\ldots,
Ti1,…,i2k∂i1⋯∂i2k=(2k−1)!!∏q=0k−1(n+2q)Δk(28)T_{i_1,\ldots,i_{2k}}\partial_{i_1}\cdots\partial_{i_{2k}}=\frac{(2k-1)!!}{\prod_{q=0}^{k-1}(n+2q)}\Delta^k \tag*{(28)}

Hence, said expansion is of the form

uˉ(P,C)−u(PC)=∑k=0∞C2kb2(k+1)Δ(k+1)u(PC),with(29)\bar{u}(P,C)-u(PC)=\sum_{k=0}^{\infty}C_{2k}b^{2(k+1)}\Delta^{(k+1)}u(PC),\quad\text{with} \tag*{(29)}
C2k=12k+1(k+1)!∏q=0k(n+2q),i.e.C_{2k} = \frac{1}{2^{k+1}(k+1)! \prod_{q=0}^{k}(n+2q)}, \qquad\text{i.e.}
C0=12n,C2=18n(n+2),C4=148n(n+2)(n+4),⋯C_0=\frac{1}{2n},\quad C_2=\frac{1}{8n(n+2)},\quad C_4=\frac{1}{48n(n+2)(n+4)},\cdots

Both Newton’s shell theorem and Gurzadyan’s theorem are transposable to nn dimensions, provided that the Newton’s potential be considered as 1/rn−21/r^{n-2}:

  • the Laplace equation for a spherical symmetric function (and for r≠0r\ne0) has, up to a multiplicative constant, the unique non-constant solution u(r)=UN(r)=1/rn−2u(r)=U_N(r)=1/r^{n-2} giving uˉ(C)=u(C)\bar{u}(C)=u(C); note that this last relation is independent of the closed surface surrounding CC because the field ∇UN\nabla U_N satisfies Gauss’s theorem adapted to the nn dimensions;

  • from Eq. (B.8) for central potentials, the condition ∇[uˉ−u]=0\nabla[\bar{u}-u]=0 (independently of bb) of the Gurzadyan’s theorem is equivalent to ∇Δu=0\nabla\Delta u=0 and this last equation has the unique non-constant solution

u(r)=K1rn−2+K2r2u(r)=\frac{K_1}{r^{n-2}}+K_2r^2

The extension of both previous theorems, that requires uˉ−w(b)u=0\bar{u}-w(b)u=0 independently of bb, is also transposable to nn dimensions, see [24]. From Eq. (B.8), the condition is equivalent to

u(PC)[w(b)−1]=∑k=0∞C2kb2(k+1)Δ(k+1)u(PC)u(PC)\left[w(b)-1\right]=\sum_{k=0}^{\infty}C_{2k}b^{2(k+1)}\Delta^{(k+1)}u(PC)

and, PC=aPC=a and bb being independent, this equation can be satisfied if and only if Δu\Delta u and uu are proportional, i.e. when uu is a solution of the equation

Δu(a)=u′′+n−1au′=c2u(a)(30)\Delta u(a)=u''+\frac{n-1}{a}u'=c_2u(a) \tag*{(30)}

c2c_2 being a real constant (i.e. independent of aa and bb). By setting u=a12−nBu=a^{\frac{1}{2}-n}B and considering BB as a function of the new variable zz such that z2=−c2a2z^2=-c_2a^2, we find that B(z)B(z) must verify the equation

z2d2Bdz2+zdBdz+B[z2−ν2]=0,withν=n2−1(31)z^2\frac{d^2B}{dz^2}+z\frac{dB}{dz}+B\left[z^2-\nu^2\right]=0,\quad\text{with}\quad\nu=\frac{n}{2}-1 \tag*{(31)}

which is that of Bessel functions of order ν\nu, see [25]. Note that zz is real or complex according to the sign of c2c_2. Since ν\nu is non-integer, the general solution of Eq. (B.10) is a linear combination of the two Bessel functions Jν(z)J_\nu(z) and J−ν(z)J_{-\nu}(z).

As for w(b)w(b), it is expressed as

w(b)=1+∑k=0∞C2kb2(k+1)C2k+1w(b) = 1 + \sum_{k=0}^{\infty} C_{2k} b^{2(k+1)} C_{2}^{k+1}

It is easy to check that for n=3n = 3, the expressions of uu and ww are those already obtained in Appendix § 3.

References

  1. [1]I. Newton, Philosophiae Naturalis Principia Mathematica, Sec. XII (1687).
  2. [2]B. Cameron Reed, “A note on Newton’s shell-point equivalency theorem”, Am. J. Phys. 90, 394-396 (2022)DOI
  3. [3]S. Axler, P. Bourdon, W. Ramey, Harmonic function Theory, 2nd Ed., GTM Vol. 137, Springer (2001)
  4. [4]V. G. Gurzadyan, “The cosmological constant in the McCrea-Milne cosmological scheme”, Observatory 105, 42-43 (1985).
  5. [5]C. Carimalo, “Comment on the article “A note on Newton’s shell-point equivalency theorem” [Am. J. Phys. 90, 394 (2022)]”, Am. J. Phys. 91, 1031-1032 (2023)
  6. [6]Trung V Phan, “On the derivation of the cosmological Gurzadyan’s theorem”, Phys. Educ. 61, 015003-015006 (2026)DOI
  7. [7]M. Markoutsakis, Geometry, Symmetries, and Classical Physics, A Mosaic, CRC Press, Taylor & Francis (2022)DOI
  8. [8]V. G. Gurzadyan, A. Stepanyan, “Two fundamental constants of gravity unifying dark matter and dark energy”, Eur. Phys. J. Plus 78, 632-636 (2018)DOI
  9. [9]V. G. Gurzadyan, “On the common nature of dark matter and dark energy : galaxy groups”, Eur. Phys. J. Plus 134, 14-17 (2019)DOI
  10. [10]C. J. Chapman, “A problem of unusual gravity”, The Mathematical gazette, 67, 14-17 (1983)DOI
  11. [11]A. Barnes, C. K. Keogh, “Unusual gravity”, The Mathematical Gazette, 68, 138-139 (1984)
  12. [12]Mugur B. Răuț, “A Newtonian equivalent for the cosmological constant”, viXra :1610.0224, Relativity and Cosmology (2016)
  13. [13]J. S. Rowlinson, “The Yukawa potential”, Physica A 156, 15-34 (1989)DOI
  14. [14]A. S. Goldhaber, M. M. Nieto, “Mass of the graviton”, Phys. Rev. D 9, 1119-1121 (1974)DOI
  15. [15]M. Wisser, “Mass for the Graviton”, General Relativity and Gravitation 30, 1717-1728 (1998)
  16. [16]A. Beckwith, “Can a Massive Graviton be a Stable Particle?”, J. Mod. Phys. 2, 350-353 (2011)DOI
  17. [17]A. A. Araújo Filho, K. Jusufi, B. Cuadros-Melgar, G. Leon, “Dark matter signature of black holes with Yukawa potential”, Phys. Dark Univ. 44, 101500-101523 (2024)DOI
  18. [18]K. Jusufi, G. Leon, A. D. Millano, “Dark Universe Phenomenology from Yukawa Potential?”, Phys. Dark Univ. 42, 101318-101330 (2023)DOI
  19. [19]J. E. McDonald, “Maxwellian Interpretation of the Laplacian”, Am. J. Phys. 33, 706-711 (1965)DOI
  20. [20]J. Sivardière, “Une introduction géométrique et physique du laplacien”, BUP 86, n⁰ 744, 771-788 (1992)
  21. [21]F. Styer, “The geometrical significance of the Laplacian”, Am. J. Phys. 83, 992-997 (2015)DOI
  22. [22]David. J. Siminovitch, “Comment on “The geometrical significance of the Laplacian” [Am. J. Phys. 83, 992-997 (2015)]”, Am. J. Phys., 64, 646 (2016)DOI
  23. [23]M. N. Olevskii, “An explicit expression of the remainder in the Pizzetti formula”, Funct. Anal. Appl., 23, 331-333 (1989)DOI
  24. [24]A. K. Tse, O. M. Markowich, T. V. Phan, “Generalizing Shell Theorem to Constant Curvature Spaces in All Dimensions and Topologies”, arXiv :2511.20852v2[physics.class-ph]
  25. [25]E. Jamke, F. Emde, F. Lösch, Tables of Higher Functions, 6th Ed., B. G. Teubner Verlagsgesellschaft mbH, Stuttgart, McGraw-Hill ed. (1960)

Paper details

vlnc:W_pfcb4gr3
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents