betaFeedback

WorksPhysics and Astronomy

Improved numerical plotting of elliptical orbits using radial action coordinates – has the symmetry of Leibniz radial theory based on inertia versus gravity been ignored?

Abstract

We that show two body gravitational orbits may be plotted using a radial reference frame rather than the customary Newtonian rectilinear inertial frame. Infinitesimal calculus cofounder and continental contemporary of Newton, Leibniz claimed that the second radial derivative could be found by taking the difference between an inertial force varying inversely, with the cubed radius and the gravitational force varying with an inversed squared radius. His radial method was severely criticised by Newton and supporters, who preferred the rectilinear inertial frame, also claiming Leibniz failed to satisfy Newton's third law of gravity. We show that these two approaches are equivalent if the central body is much larger than that in orbit. Furthermore we justify the Leibniz least action approach using the semi-latus rectum of the orbit, characteristic of the eccentricity to calculate the Newtonian centripetal acceleration opposed by the inertial acceleration, with the net radial acceleration controlling radius also allowing estimation of angular displacement and the action for each interval. Numerically, our Leibniz model can be more accurate since it requires no assumption that linear vectors adequately simulate a curvilinear trajectory. Our review of the Newton gravitational equation using centre of mass coordinates concludes that the third law is satisfied for symmetry if the gravitational and inertial masses are taken separately as equating force expressions for gravity and radial inertia, directed to the centre of mass for radial inertia. This symmetry validates centre of mass radial coordinates for more accurate plotting of orbits including dual stars, possibly applicable for better understanding of Einstein relativity. An advantage of radial coordinates is that orbiting couples can also accept perturbations from other nearby bodies acting centripetally, but not inertially.

Short title: Plotting elliptical orbits using radial coordinates.

Keywords: Newton, Leibniz, radial action, numerical plots, gravitational orbits, centripetal acceleration, inertial acceleration, semi-latus rectum

Introduction

Since Kepler and Newton established the laws of planetary motion in the 17th century, the mathematical theory enabling the plotting of orbits based on classical gravitation has been prolific [1, 2, 3]. Can anything of value still be added to this theory?

The main thesis of this paper is that extra clarity regarding the plotting of orbits in classical gravitation can still be obtained using symmetry. This may aid our understanding of the nature of gravitation, still a perplexing problem not adequately explained by Einstein’s tensor model of general relativity. While orbits can be plotted mathematically using an exact analytical solution for the two-body problem as ellipses, this is not possible when these orbits are perturbed by the presence of other massive bodies. Depending on the coordinates employed, the orbits of satellites can be plotted by a range of numerical methods. The most common method employs Newton’s inertial reference system with Cartesian vectors to plot the orbit, by numerically recalculating the centripetal acceleration for each time interval [3], now in simulations such as MATLAB.

A new numerical method, described here for the first time, relies on the fact that in a rotational reference system the net radial acceleration can be estimated for each interval in the orbit by simply subtracting the centripetal gravitational acceleration inwards, defined partly by the orbital semi-latus rectum, from the inertial acceleration, exerted outwards along the radius. This provides an easily understood means of plotting the orbit as a function of the current radius and the constancy of its angular motion (h=mr2ωh = mr^{2}\omega). It is suggested that this novel approach may be extended to more than two-body calculations. The radial action method used here also reveals features of gravity not normally seen.

We anticipate that the ability to consider inertial and gravitational effects as independent may have other practical applications in physical and environmental processes. This is the main reason for our development of this communication, as a basis for future studies in the global environment.

Methods

Estimating velocity and acceleration for central forces

The position of a point mm in a plane may be determined by Cartesian coordinates (x,y)(x, y) relative to fixed rectangular axes; or, using trigonometry, by polar coordinates (r,θ)(r, \theta) indicating the distance rr from the origin MM, with θ\theta the angle that the radius vector from the origin to the point makes with the semi-major axis of elliptic motion. This angle in the plane of rotation can also be considered as showing orientation with respect to the distant stars.

A very important physical property of dynamic symmetry (r2ω=hr^{2}\omega= h) has been referred to as the specific orbital action [7], being equivalent to the angular momentum per unit mass. Although action has the same physical dimensions as angular momentum (ML2T−1\mathrm{ML}^{2}\mathrm{T}^{-1}), it corresponds to the actual motion per unit mass through a sector of its trajectory or angle (δθ\delta\theta) – effectively a slice of action (r2ωδθr^{2}\omega\delta\theta or rω.rδθ=rω.δsr\omega.r\delta\theta= r\omega.\delta s). Angular motion is considered as a ratio between the radius and an arc of the circumference and is therefore dimensionless, but r2ωr^{2}\omega will only have the same value as r2ωδθr^{2}\omega\delta\theta for one radian of angular motion.

For a central force such as gravitation with constant energy r2ω=hr^{2}\omega= h is constant, as required by Kepler’s observations and his second law of the radius vector sweeping equal areas in equal times. For elliptical orbits, this requires that the infinitesimal changes in rr and θ\theta exactly compensate for each other. That is, the Coriolis acceleration 2dr/dt.dθ/dt2dr/dt.d\theta/dt is equal and opposite in sign to rd2θ/dt2rd^{2}\theta/dt^{2} so that the net normal component of acceleration is zero. This symmetry depends on r2ωr^{2}\omega remaining constant, consistent with a path of least variation in action.

For circular motion where dr/dtdr/dt and d2r/dt2d^{2}r/dt^{2} are both zero, just as in the case of a stone rotated on a sling, we have −r(dθ/dt)2-r(d\theta/dt)^{2} or −rω2-r\omega^{2} as the well-known centripetal acceleration, causing the curved motion towards the centre pivot. For an inertial frame of reference, the centripetal and central accelerations on the stone can be regarded as equal and opposite, so there is no acceleration along the radius d2r/dt2d^{2}r/dt^{2}. In this case the source of the centripetal force giving the circular motion is the tension generated by the musculature in the arm holding the sling.

However, for gravitation in elliptical orbits of varying radius, we will conclude that the term rω2r\omega^{2} cannot be the actual centripetal acceleration because that should be synonymous with the gravitational radial acceleration equal to d2r/dt2−rω2d^{2}r/dt^{2} - r\omega^{2}. For a rotating frame of reference, we can introduce an inertial force acting centrifugally outwards, regarded as fictitious by Newton’s first law for recto-linear inertial coordinates. It can then be considered that the relative imbalance between the attractive centripetal and the repulsive inertial effects leads to appropriate changes in d2r/dt2d^{2}r/dt^{2} during the orbit, producing an elliptical orbit. The value of d2r/dt2d^{2}r/dt^{2} is an observation or a result, not a cause.

Symmetry in gravitational acceleration

We draw attention to the characteristic function for a gravitational couple in an elliptical orbit (see Figure 1 for geometry), the celerity (C=h2/lC = h^2/l) or quickness of action [7], the product resulting from the combined mass of a couple (M+mM+m) and the universal gravitational constant GG defined by Newton. From Kepler’s second law of equal areas h/2h/2 being swept by the radius vector in equal times, we can conclude (1). Here aa is the semi-major axis of the ellipse, bb the semi-minor axis and ll a characteristic parameter known as the semilatus rectum (Figure 1), the normal to the axis aa passing through the focus (F or F’) near the location of the major body.

Figure 1. Geometrical features of an elliptical orbit (reproduced from [7]. Major semiaxis: AO = gravitational OA’ = FB = F’B = FB’ = F’B’ = aa; Minor semiaxis: OB = OB’ = bb

r2ω=b2ωb=a2ωa=l2ωl=h(1)r^2\omega= b^2\omega_b = a^2\omega_a = l^2\omega_l = h \tag*{(1)}

The period TT of the elliptical orbit is equal to the area of the ellipse πab\pi ab divided by the area swept out per unit time h/2h/2, as shown in (2).

T=2πabh(2)T = \frac{2\pi ab}{h} \tag*{(2)}

Substituting for both aa and bb in this equation from (1), we have

T=2π(ωaωb)1/24π2T2=ωaωb=ω2\begin{aligned} T &= \frac{2\pi}{(\omega_a\omega_b)^{1/2}}\\ \frac{4\pi^2}{T^2} &= \omega_a\omega_b = \omega^2 \end{aligned}
Area of ellipse=πab(=πr2 when a=b=r in a circle)OF=OF′=ae(where e is eccentricity)FL=semilatus rectum=l=FL′=b2a,al=b2AF=a(1−e)=l1+e=F′A′A′F=a(1+e)=l1−e1−e2=la=b2a2;a2−b2=a2e2e=(1−b2a2)0.5FB+BF′=2a=FP+PF′=FL+LF′∠FPB′=∠CPB′∠PQF′=∠CQF′\begin{aligned} \text{Area of ellipse} &= \pi ab \quad(= \pi r^2\text{ when }a=b=r\text{ in a circle})\\ OF &= OF' = ae \quad(\text{where }e\text{ is eccentricity})\\ FL &= \text{semilatus rectum}=l=FL'=\frac{b^2}{a},\quad al=b^2\\ AF &= a(1-e)=\frac{l}{1+e}=F'A'\\ A'F &= a(1+e)=\frac{l}{1-e}\\ 1-e^2 &= \frac{l}{a}=\frac{b^2}{a^2};\quad a^2-b^2=a^2e^2\\ e &= \left(1-\frac{b^2}{a^2}\right)^{0.5}\\ FB+BF' &= 2a=FP+PF'=FL+LF'\\ \angle FPB' &= \angle CPB'\\ \angle PQF' &= \angle CQF' \end{aligned}

R is a pole on which tangents drawn to the orbit (e.g. QP,QC) are of equal length, the point of their intersection joined to the adjacent focus bisecting the angle between them. If the central body (e.g. the sun) is located at F, then A represents the perihelion of the orbit and A’ represents the aphelion. As justified in this paper, elliptical orbits are energetically conservative of total energy EE, with E=T+VE=T+V equivalent per unit mass equal to −h2/2al=−h2/2b2=(dr/dt)2/2+h2/2r2−v2/2−h2/lr-h^2/2al=-h^2/2b^2=(dr/dt)^2/2+h^2/2r^2-v^2/2-h^2/lr, where h=r2ωh=r^2\omega, designated the specific action, is a critical constant for all orbits, related to the conservation of angular momentum.

Here ω˙\dot{\omega} is the average angular orbital velocity 2π/T2\pi/T; then Kepler’s third law for the constancy of the product of the cube of the orbital radius with the square of the period is shown in the following equalities for the celerity CC.

4π2a3/T2=a3ω˙2=a3ωaωb=a3ωa3/2ωl1/2=l3ωl3/2ωl1/2=l3ωl2=h2/l=C(3)\begin{aligned} 4\pi^2a^3/T^2 &= a^3\dot{\omega}^2 = a^3\omega_a\omega_b\\ &= a^3\omega_a^{3/2}\omega_l^{1/2}\\ &= l^3\omega_l^{3/2}\omega_l^{1/2}\\ &= l^3\omega_l^2\\ &= h^2/l\\ &= C \tag*{(3)} \end{aligned}

Given the dimensional form of GG (L3T−2M−1L^3T^{-2}M^{-1}), the universal gravitation constant defined by Newton as a function of total mass, we have an exact solution for CC in a function defined by the square of the specific action and the semi-latus rectum.

G(m+M)=a3ω˙2=l3ωl2=r4ω2/l=h2/l=C(L3T−2)(4)G(m+M)=a^3\dot{\omega}^2=l^3\omega_l^2=r^4\omega^2/l=h^2/l=C\quad(L^3T^{-2}) \tag*{(4)}

Note that CC is not a function solely of the mass of the central body, such as the Sun, but is a function of the total of each system of coupled masses. Where the orbits of satellites of negligible mass compared to the central body are concerned, it is possible to regard CC as an approximate constant for these satellites of different masses. However, with modern calculators orbits can be calculated precisely using the actual masses. A computation using an appropriate number of significant digits — seven for the planet mercury orbiting the Sun, but at least 28 for a satellite of 100 kg mass in orbit and 57 digits for a proton in solar orbit will be required to appreciate this difference.

In effect, Newton’s gravitational constant provides a definition of mass by its effect on celerity CC. Most significantly, the greater CC the greater the straightness of the curvature of the orbit.

Centre of mass coordinates

We use coordinates based on the centre of masses as (M+m)(M+m) lying on rr (r=rm+rMr=r_m+r_M) known as the barycentre; then mrm=MrMmr_m=Mr_M and M=mrm/rMM=mr_m/r_M.

Substituting for centre of mass coordinates in Newton’s inverse square law of universal gravitation in Equation (5)

Fg=−mMGr2=−m2rmGrMr2=−m2rmh2rM[(m+M)lr2]=−m2rmh2[(mrM+mrm)lr2]=−mrmh2lr3=−mrmω2rl=−MrMω2rl(5)\begin{aligned} F_g &= -\frac{mMG}{r^2}\\ &= -\frac{m^2r_mG}{r_Mr^2}\\ &= -\frac{m^2r_mh^2}{r_M[(m+M)lr^2]}\\ &= -\frac{m^2r_mh^2}{[(mr_M+mr_m)lr^2]}\\ &= -\frac{mr_mh^2}{lr^3}\\ &= -\frac{mr_m\omega^2r}{l}\\ &= -\frac{Mr_M\omega^2r}{l} \tag*{(5)} \end{aligned}

These symmetrical forces (mrmω2r/lmr_m\omega^2r/l and MrMω2r/lMr_M\omega^2r/l) can satisfy Newton’s third law of action and reaction in the gravitational couple between mm and MM [5]. Usually, the acceleration involved in the latter force for the solar system −rMω2r/l-r_M\omega^2r/l is negligible, so close to the Sun’s centre is the system’s centre of mass, so that rmr_m is practically equal to rr with little or no error in neglecting rMr_M.

However, for complete accuracy in binary systems such as dual stars where mm and MM are comparable, it becomes imperative to add the two accelerations −rMω2r/l-r_M\omega^2r/l and −rmω2r/l-r_m\omega^2r/l for MM and mm respectively, both directed inwards towards the centre of mass. Together, they will give the total relative acceleration, referred to earlier in Equation (6).

rMω2r/l+rmω2r/l=rω2/l(6)r_M\omega^2r/l+r_m\omega^2r/l=r\omega^2/l \tag*{(6)}

Note that this becomes the same as the gravitational acceleration derived by Newton of V2/rV^2/r in circular motion when r=lr=l. For elliptical orbits, the gravitational acceleration can also be expressed as C/r2C/r^2 or h2/lr2h^2/lr^2, where CC is the celerity ((m+M)G(m+M)G), ll is the semilatus rectum or radial axis normal to the major axis of the ellipse and hh is the specific orbital action (r2ωr^2\omega).

The celerity relationship of (m+M)G(m+M)G equals h2/lh^2/l defines any bipolar orbit [6], whether coupling a much larger central body or not. We will show how, for maximum accuracy in plotting, all orbits should be plotted as around the centre of mass of the system.

Another sound reason for using centre of mass coordinates is that this preserves the correct relationship for giving the overall action of the system. Using the centre of mass of one or other always results in an incorrect expression for action. The action is usually given as the integral of the momentum with respect to distance (∫mvds\int mv\mathrm{d}s). This can be reinterpreted as the integral of mr2ωdθmr^2\omega\mathrm{d}\theta. Given that both members of the couple will change their inertial position during an orbit with respect to the distant stars, the true action for both members for a change of dθ\mathrm{d}\theta is mrm2ωdθ+MrM2ωdθmr_m^2\omega\mathrm{d}\theta+Mr_M^2\omega\mathrm{d}\theta or mrmrωdθ=MrMrωdθ=Iωdθmr_mr\omega\mathrm{d}\theta=Mr_Mr\omega\mathrm{d}\theta=I\omega\mathrm{d}\theta. Thus, we have revealed an important symmetry involved in the action of the orbital pairing and how the total action is constant because, IωI\omega is constant. This cannot be achieved if all of the action is ascribed to the smaller member of the orbital system.

Incidentally, from Equation (6) we can also express the Newtonian gravitational force FgF_g as follows.

Fg=−mMGr2=−mrmω2rl=−mrmω2(rm+rM)l=−(mrm2+MrM2)ω2l=−Iω2l\begin{aligned} F_g &= -\frac{mMG}{r^2}\\ &= -\frac{mr_m\omega^2r}{l}\\ &= -\frac{mr_m\omega^2(r_m+r_M)}{l}\\ &= -\frac{(mr_m^2+Mr_M^2)\omega^2}{l}\\ &= -\frac{I\omega^2}{l} \end{aligned}

Here, II is (mrm2+MrM2)(mr_m^2+Mr_M^2), the total moment of inertia.

Thus, a re-interpretation of the orbital gravitational force as expressed by Newton is as a variable orbital torque FglF_g l proportional to the angular velocity ω\omega according to the following equation,

Fgl=−Iω2(7)F_g l=-I\omega^2 \tag*{(7)}

However, this torque is just one-half of that exerted by the couple in the opposite direction, providing a flexible link ensuring that the net torque is zero and there is therefore no change in angular momentum. Note that for a circular orbit where l=rl=r, that FgF_g will be symmetrically equal to mrmω2mr_m\omega^2 and MrMω2Mr_M\omega^2. Under these conditions, the Lagrangian remains constant and there is no variation in action. This also means that the derivative of the total action IωdθI\omega\mathrm{d}\theta with respect to time is equal to the torque for the orbit, Iω2I\omega^2, which varies with the angular velocity and the moment of inertia but not with the total angular momentum IωI\omega. This is expected, given that (m+M)r2ω=(m+M)h(m+M)r^2\omega=(m+M)h.

This symmetry of action is an essential component of gravity.

Cartesian vectors

The Cartesian differential equations of motion [1] for an attractive force, where aca_c is the centripetal or radial acceleration [1], are shown in Equation (8).

d2xdt2=−accos⁡θ=−acx/r\frac{d^2x}{dt^2}=-a_c\cos\theta=-a_c x/r
d2ydt2=−acsin⁡θ=−acy/r(8)\frac{d^2y}{dt^2}=-a_c\sin\theta=-a_c y/r \tag*{(8)}

From Equation (8), the axial components for an acceleration obeying an inverse square law of attraction where ac=−h2/lr2=−r2ω2/la_c=-h^2/lr^2=-r^2\omega^2/l are given by,

d2xdt2=−Ccos⁡θ/r2=−Cx/r3=−h2x/lr3=−xω2r/l=ax\frac{d^2x}{dt^2}=-C\cos\theta/r^2=-Cx/r^3=-h^2x/lr^3=-x\omega^2r/l=a_x
d2ydt2=−Csin⁡θ/r2=−Cy/r3=−h2y/lr3=−yω2r/l=ay\frac{d^2y}{dt^2}=-C\sin\theta/r^2=-Cy/r^3=-h^2y/lr^3=-y\omega^2r/l=a_y
d2xdt2+d2ydt2=−Cx/r3−Cy/r3=ax+ay=−(x+y)ω2r/l(9)\begin{aligned} \frac{d^2x}{dt^2}+\frac{d^2y}{dt^2}&=-Cx/r^3-Cy/r^3=a_x+a_y\\ &=-(x+y)\omega^2r/l \tag*{(9)} \end{aligned}

Note in Figure 2 that the sum of the Cartesian accelerations in the x,yx,y plane is rarely equal to the gravitational acceleration, but usually exceeds it, as (x+y)(x+y) is greater than rr (see Figure 2), except when xx or yy equals zero. This indicates that resolution of a radial vector into linear Cartesian elements infers that greater forces or impulses must operate than are actually required. This has consequences for plotting orbits increasing with longer time intervals, a point to be examined later in this article. However, when x=0x=0 for the satellite lying on the latus rectum where r=l=yr=l=y (see Figure 2), it is equal to the gravitational centripetal acceleration. This is also true when y=0y=0 at perigee and apogee in the orbit and r=xr=x.

Figure 2. The traditional Newtonian plot involves vectors in the x,yx,y plane corresponding to the centripetal acceleration whereas radial vectors operate in a rotating frame, can produce almost the same result. The Cartesian vectors axa_x and aya_y exceed the radial vector aca_c given that ax.x/r+ay.y/r=aca_x.x/r+a_y.y/r=a_c except when xx or yy equals zero.

So, to have an equation directly relating the Cartesian accelerations, identified by Newton, and the gravitational acceleration, we must correct Equation (9) by writing down (11).

d2xdt2cos⁡θ+d2ydt2sin⁡θ=−C/r2(cos⁡2θ+sin⁡2θ)=−C/r2=−rω2/l=−G(M+m)/r2=−h2/lr2=−hω/l=d2rdt2−rω2(10)\begin{aligned} \frac{d^2x}{dt^2}\cos\theta+\frac{d^2y}{dt^2}\sin\theta &=-C/r^2(\cos^2\theta+\sin^2\theta)=-C/r^2=-r\omega^2/l\\ &=-G(M+m)/r^2=-h^2/lr^2=-h\omega/l\\ &=\frac{d^2r}{dt^2}-r\omega^2 \tag*{(10)} \end{aligned}

The radius ll normal to the major axis of the ellipse where θ\theta is π/2\pi/2 radians is the semilatus rectum for the orbit (Figure 1). The fact that CC is equal precisely to h2/lh^2/l, a unique constant for each stable gravitational couple characteristic of the combined mass M+mM+m, is of great significance; the constant h2/lh^2/l can be written as the fixed value of l3ω2l^3\omega^2 or as r3ω2.r/lr^3\omega^2.r/l, a varying relationship to be explored in more detail later.

The x,yx,y plane accelerations of Equations (8), (9) must track as (11) and (12) to equal the centripetal radial acceleration.

d2xdt2⋅cos⁡θ=d2xdt2⋅xr=−G(M+m)x2r4=−h2x2lr4=−x2ω2l=axxr(11)\frac{d^2x}{dt^2}\cdot\cos\theta= \frac{d^2x}{dt^2}\cdot\frac{x}{r} = -\frac{G(M+m)x^2}{r^4} = -\frac{h^2x^2}{lr^4} = -\frac{x^2\omega^2}{l} = a_x\frac{x}{r} \tag*{(11)}

as the vector component of gravitational acceleration along the x-axis, and

d2ydt2⋅sin⁡θ=d2ydt2⋅yr=−G(M+m)y2r4=−h2y2lr4=−y2ω2l=ayyr(12)\frac{d^2y}{dt^2}\cdot\sin\theta= \frac{d^2y}{dt^2}\cdot\frac{y}{r} = -\frac{G(M+m)y^2}{r^4} = -\frac{h^2y^2}{lr^4} = -\frac{y^2\omega^2}{l} = a_y\frac{y}{r} \tag*{(12)}

as the vector component of gravitational acceleration along the y-axis.

Obviously, ax=−rω2x/la_x=-r\omega^2x/l and ay=−rω2y/la_y=-r\omega^2y/l.

Thus, we show in (13) the corrected equivalence for the Cartesian accelerations in Equations (8) and (9).

d2xdt2⋅xr+d2ydt2⋅yr=−G(M+m)r2=−r2ω2l=d2rdt2−rω2(13)\begin{aligned} \frac{d^2x}{dt^2}\cdot\frac{x}{r}+\frac{d^2y}{dt^2}\cdot\frac{y}{r} &= -\frac{G(M+m)}{r^2}=-\frac{r^2\omega^2}{l}\\ &=\frac{d^2r}{dt^2}-r\omega^2 \tag*{(13)} \end{aligned}

The Cartesian accelerations are reduced by the current ratios of x/rx/r and y/ry/r to equal the centripetal acceleration expressed radially.

From Equation (13), we can write energy potentials per unit mass as,

x⋅d2xdt2+y⋅d2ydt2=r⋅d2rdt2−r2ω2ax⋅x+ay⋅y=r⋅ar−r2ω2=−r3ω2l=−Cr\begin{aligned} x\cdot\frac{d^2x}{dt^2}+y\cdot\frac{d^2y}{dt^2} &= r\cdot\frac{d^2r}{dt^2}-r^2\omega^2\\ a_x\cdot x+a_y\cdot y &= r\cdot a_r-r^2\omega^2=-\frac{r^3\omega^2}{l}=-\frac{C}{r} \end{aligned}

When yy is equal to both rr and ll (where the satellite is placed at the latus rectum with xx equal to zero and the origin as in Figure 2, d2y/dt2d^2y/dt^2 is equal to −rω2-r\omega^2, also equal to −r2ω2/l-r^2\omega^2/l. When xx is equal to rr at apogee or perigee where yy is zero, d2x/dt2d^2x/dt^2 is equal to −r2ω2/l-r^2\omega^2/l, but not equal to −rω2-r\omega^2, often called the "centripetal acceleration", since rr is not equal to ll except twice during an orbit. In fact, the true acceleration of gravitation Newton was seeking is not −rω2-r\omega^2 but −r2ω2/l-r^2\omega^2/l, except in circular motion when r=lr=l or when the orbit is at the latus rectum, the geometric equivalent of circular motion on a conical section.

These Equations (7)-(12) refer to the sum of the relative accelerations of the satellite towards the central body −GM/r2-GM/r^2 with that of the central body towards the satellite −Gm/r2-Gm/r^2. In most discussions on gravitation with a massive central body MM, only the former acceleration is considered, neglecting that caused by the satellite mm and the component accelerations are then −GMx/r3-GMx/r^3 and −GMy/r3-GMy/r^3 [3]. One should be aware that this is merely an approximation. In fact, −GM/r2-GM/r^2 is equal to −rm2ω2/l-r_m^2\omega^2/l rather than −r2ω2/l-r^2\omega^2/l, where rmr_m is the radius of mm to its centre of mass with MM rather than the distance rr separating their centres (see Equations (6) and (7)). We can equate d2xdt2⋅xr+d2ydt2⋅yr\frac{d^2x}{dt^2}\cdot\frac{x}{r}+\frac{d^2y}{dt^2}\cdot\frac{y}{r} to d2rmdt2−rmω2\frac{d^2r_m}{dt^2}-r_m\omega^2, but this is only true where the origin is located at the centre of mass of the gravitational couple MM and mm rather than the centre of the larger body MM.

Note the equality that the sum of the squares (dx/dt)2+(dy/dt)2(dx/dt)^2+(dy/dt)^2, given sin⁡2θ+cos⁡2θ=1\sin^2\theta+\cos^2\theta=1, is equal to the square of the linear velocity vv. This is also equal to the radial velocity (dr/dt)2(dr/dt)^2 or Vr2+r2ω2V_r^2+r^2\omega^2.

(dr/dt)2+r2ω2=(dx/dt)2+(dy/dt)2=v2(dr/dt)^2+r^2\omega^2=(dx/dt)^2+(dy/dt)^2=v^2

Or

Vx2+Vy2=v2=Vr2+r2ω2V_x^2 + V_y^2 = v^2 = V_r^2 + r^2\omega^2

Comparing cartesian versus radial vectors

Thus, the radial acceleration d2r/dt2d^2r/dt^2 is actually made of a larger acceleration d2rm/dt2d^2r_m/dt^2 of the smaller satellite mm towards the centre of mass rmr_m and a symmetrically opposite acceleration d2rM/dt2d^2r_M/dt^2 of the larger body MM towards the centre of mass rMr_M. Though acting in opposite directions, these must be added to obtain the total radial acceleration d2r/dt2d^2r/dt^2 towards the centre of mass. This distinction is usually overlooked, given that it provides a satisfactory approximation when the mass of the central body is orders of magnitude greater than the satellite, as for most bodies in the solar system. But for calculations with dual or binary stars or where the relative masses are less than 100, errors result if this approximation is used.

In this article, the validity of the expression above (21) relating radial (ara_r), inertial (aia_i) and centripetal accelerations for bodies mm and MM bonded by gravitation is examined in detail (Figure 2). Unlike the case of the stone whirled in a sling, d2r/dt2d^2r/dt^2 has both positive and negative values during an orbit and the difference is given by the difference between the inertial and the centripetal accelerations. We have shown the validity of the equivalence of variations in radial symmetry using polar coordinates to those using Cartesian coordinates (Figure 3).

Figure 3. Outline of a radial microcomputer program designed to numerically plot a satellite orbit. Note that the first iteration requires an initial multiplicand of 1/2, to obtain the mean velocity in δt1\delta t_1, thus avoiding a false radius at the end of period one; all subsequent iterations loop back, omitting this factor. Alternatively, this error in integration could be minimised by shortening δt\delta t.

d2r/dt2−r⋅(dθ/dt)2=d2r/dt2−h2/r3=−(M+m)G/r2=−h2/lr2=−r2ω2/l(14)d^2r/dt^2 - r\cdot(d\theta/dt)^2 = d^2r/dt^2 - h^2/r^3 = -(M+m)G/r^2 = -h^2/lr^2 = -r^2\omega^2/l \tag*{(14)}

Thus we can write in Equation (15) the relationship involving variation in dr/dtdr/dt as one between two symmetries, but out of balance during the orbit except at the semilatus rectum where ll equals rr.

d2r/dt2=h2/r3−h2/lr2=rω2−rω2⋅r/l=rω2−r2ω2/lar=ai−ac(15)\begin{aligned} d^2r/dt^2 &= h^2/r^3 - h^2/lr^2 \\ &= r\omega^2 - r\omega^2\cdot r/l = r\omega^2 - r^2\omega^2/l \\ a_r &= a_i - a_c \tag*{(15)} \end{aligned}

(15) provides the basis for calculation of net radial acceleration separate to the use of Cartesian vectors, given that the centripetal acceleration aca_c can be calculated from a knowledge of the specific action and the semilatus rectum. These two accelerations are different in kind, given that one is inertial derived from the instantaneous circular motion, independent of the influence of the central body, while the other is gravitational and causal, obeying an inverse square relationship. The net radial acceleration (ara_r) defined here differs from the radial acceleration normally referred to in the theory of gravity of as d2r/dt2−rω2d^2r/dt^2-r\omega^2; d2r/dt2d^2r/dt^2 is no more than an observation indicating the rate at which the velocity of the radius vector dr/dtdr/dt varies or reverses in sign at perigee and apogee. We emphasise that the acceleration of the radius d2r/dt2\mathrm{d}^{2}r/\mathrm{d}t^{2} is only zero except in a circular orbit whenever the orbit crosses the latus rectum when the centripetal and inertial forces are equal. The difference reaches a maximum at perigee when the maxima for both forces is achieved and a minimum at apogee.

The fact that the gravitational aca_{c} and the inertial accelerations aia_{i} are not equal in magnitude does not conflict with Newton’s third law, although it was claimed that Newton believed that they should be equal and opposite at all times, according to a recent historical analysis of his Principia Mathematica by Aiton [5]. Aiton also indicated that Leibniz recast Newton’s fluxions in a form similar to Equation (15). In fact, equality of action and reaction is maintained during orbital motion at all times between both gravitational forces exerted by each member mm and MM in the gravitational couple on the other; equality is also achieved between the inertial forces exerted by each member of the couple. However, there is no necessity for the gravitational and the inertial forces to be equal to each other unless the orbit is circular, or when r=lr=l as occurs twice in each orbit when the radius vector in normal to the major axis lying on the latus rectum. Newton’s third law is obeyed, but for the centripetal and the inertial forces between the partners in the couple considered separately and not between these two forces given that they must vary independently during each orbit.

Results and Discussion

Symmetrical binding energy potentials

From Equations (8) and (9) we can conclude that the energy potential per unit mass can be derived from these expressions for the components of acceleration, as follows, given x2+y2=r2x^{2}+y^{2}=r^{2}.

d2xdt2+d2ydt2=−C⋅xr3−C⋅yr3\frac{\mathrm{d}^{2}x}{\mathrm{d}t^{2}}+\frac{\mathrm{d}^{2}y}{\mathrm{d}t^{2}}=-\frac{C\cdot x}{r^{3}}-\frac{C\cdot y}{r^{3}}

This relationship for Cartesian accelerations was that used by Feynman [3] to illustrate planetary motions in his lectures at Caltech in 1963. Such an approach allows the motion of all planets simultaneously orbiting the Sun to be calculated using 3-dimensional Cartesian coordinates. Obviously, this process now benefits from the availability of modern computers.

Then to obtain a work function we multiply the coordinate position value by the acceleration, assuming unit mass.

xd2xdt2+yd2ydt2=−C⋅x2r3−C⋅y2r3=−C(x2+y2)r3=−Cr=−h2lr=−l3ω2r=−r3ω2l=−G(M+m)r=−GMr−Gmr(16)\begin{aligned} x\frac{\mathrm{d}^{2}x}{\mathrm{d}t^{2}}+y\frac{\mathrm{d}^{2}y}{\mathrm{d}t^{2}}&=-\frac{C\cdot x^{2}}{r^{3}}-\frac{C\cdot y^{2}}{r^{3}}=-\frac{C(x^{2}+y^{2})}{r^{3}}\\ &=-\frac{C}{r}=-\frac{h^{2}}{lr}=-\frac{l^{3}\omega^{2}}{r}=-\frac{r^{3}\omega^{2}}{l}\\ &=-\frac{G(M+m)}{r}\\ &=-\frac{GM}{r}-\frac{Gm}{r} \tag*{(16)} \end{aligned}

It is customary to give the potential energy VV for gravitational binding of a satellite as equal to −GMm/r-GMm/r and the energy potential per unit mass as −GM/r-GM/r. However, this neglects the contribution to binding energy of the attraction by the smaller satellite suggested by Equation (23). The energy binding term differs from −mh2/lr-mh^2/lr by the ratio rm/rr_m/r, being equal to −mrmr2ω2/l-mr_mr^2\omega^2/l, where rmr_m is the length of the radius vector to the centre of mass of mm and MM. Although a very minor error for nearly all the planets, this will no longer be the case when considering the orbits of binary stars, or even the binding potential of small nuclei and electrons. When sufficient precision is available for calculation of orbits, there is no reason why the total energy potential per unit mass bound −G(M+m)/r-G(M+m)/r should not be used, rather than −GM/r-GM/r. This infers that the binding energy is symmetrically derived from each process, not only from binding of the smaller mass.

Within an orbit, the total gravitational energy EE is conserved with any change in kinetic energy matched by an opposite change in potential energy. Equation (6) established that the linear kinetic energy per unit mass of v2/2v^2/2 is equal to (dr/dt)2/2+(rω)2/2(dr/dt)^2/2+(r\omega)^2/2. So with the potential energy from (23) −G(M+m)/r=−h2/lr-G(M+m)/r=-h^2/lr we can write Equation (17) for the total energy per unit mass.

E=T+V=−h2/2al=−h2/2b2=(dr/dt)2/2+h2/2r2−h2/lr=v2/2−h2/lr(17)\begin{aligned} E &= T+V\\ &=-h^2/2al=-h^2/2b^2=(dr/dt)^2/2+h^2/2r^2-h^2/lr\\ &=v^2/2-h^2/lr \tag*{(17)} \end{aligned}

Noted in Figure 1 is the symmetry of how a radius rbr_b equal to the semi-minor axis bb lies between the semilatus rectum ll and the radius to the minor axis equal to aa, the semi-major axis. When r=ar=a, we have the total potential energy equal to −h2/al=−h2/b2=−(rbωb)2-h^2/al=-h^2/b^2=-(r_b\omega_b)^2, or the negative of the circular velocity squared, consistent with an unchanging total orbital energy EE of −(rbωb)2/2-(r_b\omega_b)^2/2. So, at a radius equal to the semi-minor axis bb, the linear kinetic energy v2/2v^2/2 is equal to h2/b(1/l−1/2b)h^2/b(1/l-1/2b) and (dr/dt)2/2(dr/dt)^2/2, the kinetic energy of the change in radius, is equal to h2/b(1/l−1/b)h^2/b(1/l-1/b).

A reinterpretation of the potential energy might be to consider that the −h2/lr-h^2/lr is a static estimate for VV, given that the circular function for EE is constant at −h2/2al-h^2/2al but the circular T′T' at r=ar=a is −(raωa)2/2-(r_a\omega_a)^2/2. The true potential energy could be lessened from −h2/lr-h^2/lr by the increased kinetic energy (dr/dt)2/2(dr/dt)^2/2, so the result for VV is actually −h2/lr−(v2/2−h2/2a2)=h2/a(1/2a−1/l)−v2/2-h^2/lr-(v^2/2-h^2/2a^2)=h^2/a(1/2a-1/l)-v^2/2. This would mean that the potential energy around the orbit is diminished from that normally calculated, except at perigee and apogee where dr/dtdr/dt equals zero. In other words, the potential energy inferred by −GMm/r-GMm/r is for a static satellite mm which by necessity it is already diminished by the radial kinetic energy (dr/dt)2/2(dr/dt)^2/2, unless perigee or apogee is being considered.

In terms of quantised gravitational field energy which is determined by rr expressed radially, Equation (18) would be more accurate. But normally this inaccuracy does no harm, as the true zero for potential energy is not used and all arguments about conservation of energy are settled by taking differences.

E=v2/2−(dr/dt)2/2−h2/lr=h2/2r2−h2/lr(18)E = v^2/2 - (dr/dt)^2/2 - h^2/lr = h^2/2r^2 - h^2/lr \tag*{(18)}

To place a satellite into orbit it must be given sufficient velocity (rω)(r\omega) to match the square root of the potential energy in orbit [h2/lr]0.5[h^2/lr]^{0.5}. This provides the total energy needed for the kinetic energy in orbit h2/2r2h^2/2r^2 that could be recovered plus an equal aliquot of field energy thus raising the potential energy twice this value, as required by the virial theorem. Furthermore, if the kinetic energy h2/2r2h^2/2r^2 in orbit is then doubled by some form of momentum thrust, the total energy EE of gravitational binding will be zero, releasing the orbit entirely. In effect, the negative potential energy −h2/lr-h^2/lr is the loss in potential energy VV caused by gravitational bonding, made up of the increase in kinetic energy h2/2r2h^2/2r^2 plus the field energy emitted to space on binding, energy previously sustaining the new satellite in some other orbit independently. The zero value assumed is only for the particular process involved. In fact, as we explained in 2001 [7], the trajectories of bound gravitational objects will involve a series of binding processes, each diminishing the original potential energy according to a summation of losses of ΣV\Sigma V equal to [Mc2−ΣMv2][Mc^2 - \Sigma Mv^2], comprising the increase in kinetic energy ΣMv2/2\Sigma Mv^2/2 and an equal emission of radiation Σhν\Sigma h\nu.

Plotting gravitational orbits

Because of the equality of the vectorial expression for the centripetal acceleration and the polar expressions given in Equations (8) and (9) we now have available two related methods of plotting a gravitational orbit. The first of these methods uses a standard vectorial approach of estimating accelerations for the xx- and yy-axes from a knowledge of rr, xx and yy and the equations given above (4, 13-16). A knowledge of the accelerations in the xx- and yy-directions then enables calculation of new velocities vxv_x and vyv_y after a short time interval and hence a new position xx, yy and radial separation rr can be found using Pythagorus’ theorem x2+y2=r2x^2 + y^2 = r^2 and thus the new accelerations determined. The process is reiterated many times to plot a half orbit until yy becomes negative (or θ\theta exceeds π\pi) and the semi-ellipse simply reflected to complete the orbit.

Alternatively, the orbit can be plotted directly from radial coordinates (r,θ)(r,\theta) (Equation (15)) again using rr to estimate the centripetal acceleration (ac)(a_c) from Equation (5), then allowing the inertial acceleration (ai)(a_i) to be found (16), since acl/r=aia_c l/r = a_i. The net radial acceleration (ar=d2r/dt2)(a_r = d^2r/dt^2) is easily found by subtraction (see Equation 22). It is then possible to estimate the new rr, ω\omega and thus θ\theta, allowing the orbit to be plotted by a process of reiteration until θ\theta at least reaches π\pi. This method may be less prone to rounding errors, having fewer calculations, but also makes no assumptions that the elliptical motion is composed of a set of straight lines, depending on the time interval for each reiteration.

We can see from the Equations (4),(12)-(15) that the standard method of plotting an orbit using Newtonian vectors with Cartesian coordinates and the alternative method using accelerations restricted to those along the radius vector subtracting the inertial acceleration with polar coordinates in the infinitesimal limit should yield the same result; they both allow the rate of change in the radius to be estimated. The Cartesian method adheres more closely in principle to Newton’s first law, since it appeals only to the central force, inversely proportional to the square of the radius r2r^2 (Equations (2)-(3)). However, the calculus above shows that employing the Cartesian accelerations axa_x and aya_y to plot the orbits is formally equivalent to the more direct approach of taking the difference between the inertial (aia_i) and centripetal (aca_c) accelerations inversely proportional to the cube and the square of the radius respectively and then directly plotting rr and δθ\delta\theta.

A radial frame of reference thus allows estimation of the variation in rr with θ\theta, to establish the orbit, resolving two opposed accelerations. The position of the central body and the satellite can then be considered as stationary with the radius vector increasing in length from perigee to apogee and decreasing from apogee to perigee, similar to a body projected upwards vertically from the earth’s surface in its rise and fall, a result of its initial inertia carrying it to the point where its newly acquired potential energy equals its initial kinetic energy, where gravity succeeds in reversing its motion. Furthermore, the whole solar system can be modelled simultaneously using this approach using polar coordinates (r,θ,ϕ)(r, \theta, \phi). The centripetal accelerations for all gravitational attractions can be resolved although the Sun will dominate at all times despite temporary perturbations by planets. The sole inertial effect proportional to rω2r\omega^2 is dependent on the composite gravitational effects and not on direct interaction with other planets.

It is obvious that the inertial acceleration h2/r3h^2/r^3 or rω2/r=rω2r\omega^2/r = r\omega^2 is directed along the radius effectively to the stars and necessarily derived from circular motion, normal to the radius. This is not a direct inertial response on the tangent to the linear velocity vv demonstrated in Equation (6). Instead, vv must be considered as the result of instantaneously circular but stepwise adjustments by quanta regarding kinetic and potential energy, conserving total energy and keeping hh constant; changes of rr and ω\omega occur infinitesimally according to the principle of least variation in action. In any conservative system the variation in potential energy is matched by an opposite variation in kinetic energy, keeping their sum constant. This is the requirement for the specific action hh to remain constant. This contrasts with the uptake or emission of field energy from the system giving a different orbit, where the variation of total energy and kinetic energy must have the same magnitude but opposite sign and the change in potential energy is double the change for total energy or kinetic energy, consistent with the virial principle. The total energy is the actual energy causing the field effects, expressed as quanta delivering impulses [7].

Computer simulations

In using a microcomputer and numerical methods to plot the orbit — most conveniently beginning from either apogee or perigee on the major axis where dr/dtdr/dt (vrv_r, cm sec−1^{-1}) and θ\theta (radians) are both zero — it is necessary to consider that the change in rr in the first time period δt\delta t is equal to the average velocity in the period multiplied by δt\delta t, approximately vrδt/2v_r\delta t/2 or arδt2/2a_r\delta t^2/2. In reiterated loops, the lengthening or shortening of radius vector is then arδt2a_r\delta t^2, the mean velocity vrv_r (arδta_r\delta t) multiplied by the subsequent time period. The change in ara_r is estimated by recalculating aca_c and aia_i for each period. Schema for separate computer programs using the radial and vectorial methods to plot elliptical orbits are given in Figure 3 and Figure 4, with all coding in Supplementary Material.

Figure 4. Outline of a radial microcomputer program designed to numerically plot a satellite orbit. Note that the first iteration requires an initial multiplicand of 1/21/2, to obtain the mean velocity in δt1\delta t_1, thus avoiding a false radius at the end of period one; all subsequent iterations loop back, omitting this factor. Alternatively, this error in integration could be minimised by shortening δt\delta t.

We present here outlines of two programs (OrbitRadial and OrbitVector) to run numerical calculations, using an interpreter for Windows emulation as TRS32, emulating the Texas Instruments programmable SR52SR52; having established the utility of these programs, we also used RR for Windows, an integrated suite of software facilities for data manipulation, calculation and graphical display available in the public domain. RR is an implementation of the SS language which was developed at Bell Laboratories. The code for these programs are given as appendices in supplementary material. Similar programs using centre of mass coordinates (MrM=mrmM r_{\mathrm{M}} = m r_{\mathrm{m}}) suitable for plotting orbits for double stars require only minor modification. Another supplementary radial action program we have prepared performs a numerical calculation for the three-body Earth-Moon systems orbiting the Sun. This model can be applied to simulating Newton’s Proposition 66, Theorem 26, Case 1, Corr. 4 in his Principia [8] for the Moon’s orbit of the Earth in that its motion “is more curved in the quadratures than at the conjunction or the opposition” caused by the Sun’s tidal effects when the Moon is on the same radial axis. This observation involving the perturbing gravitational effect of the Sun on the elliptical orbit of the Moon and earth is readily modelled using the radial vector approach.

The Astrocal SR52 program has the advantage for astronomical calculations of much greater accuracy, its 127 significant digits allowing any effect of the mass of the satellite as well as the central body to be included in calculations, despite differences in mass of more than 10 to 100 orders of magnitude between central body and satellite. Also, its operations are highly accessible in practice and the code is more easily developed and understood. But RR software is readily available to other users since it can be downloaded from the internet free of cost. Full details of these four programs are given in the Appendices to this publication and more information on accessing the software is available on request.

These numerical programs all assume that the change in radius with time is given by the series:

∑δr=ar0δt2/2+ar1δt2+ar2δt2+arnδt2……(19)\sum\delta r = a_{r0}\delta t^2/2 + a_{r1}\delta t^2 + a_{r2}\delta t^2 + a_{rn}\delta t^2 \quad\ldots\ldots \tag*{(19)}

The accelerations (arn)(a_{rn}) acting are easily calculated from the relationships above and the position of the satellite can then be estimated in either polar or Cartesian coordinates. The change in θ\theta is then easily estimated from the relationship.

δθ=(ai/rn)0.5δt,where ai=rω2\delta\theta= (a_i/r_n)^{0.5}\delta t, \qquad\text{where } a_i = r\omega^2

Then

∑δθ=∑(ai/rn)0.5δt=θ(20)\sum\delta\theta= \sum(a_i/r_n)^{0.5}\delta t = \theta \tag*{(20)}

Given that the specific action or angular momentum per unit mass is a constant as long as total energy is conserved, the relationship of the radial method to least variation of action (mvδs=mr2ωδφ)(mv\delta s=mr^2\omega\delta\varphi) is more obvious. For Maupertuis’ version of action as the time integral of momentum with respect to differential space, given orbital mr2ω=mhmr^2\omega=mh is constant with time, then the minimum action per orbit is 2πmh2\pi mh or 2πJ2\pi J. Any other orbital path proposed must exceed 2π2\pi of angular motion. The same result would be obtained by the customary Lagrangian of ∫(T−V)dt\int(T-V)dt equals ∫(mh2/2r2+mh2/rl)dt\int(mh^2/2r^2+mh^2/rl)dt or ∫ω(1/2+r/l)dt\int\omega(1/2+r/l)dt. This action integral implies that most time in the orbit will be spent where rr is greater than ll with ω\omega in its minimum values. Rather than least action being least time, the competing forces seek greatest time, where potential energy exceeds kinetic energy, like any pendulum.

Although numerical approaches are an approximation, a very high degree of accuracy is possible if the period δt\delta t is chosen sufficiently small. There is only slight improvement to accuracy when the number of iterations per orbit is increased beyond 50 or 100. However, the radial and the Cartesian methods do not give precisely the same result in terms of estimating the correct radius at apogee when the same data inputs to the two programs are made at perigee (celerity, radius, masses, etc.), although the agreement is better with smaller time intervals δt\delta t.

The radial method is more accurate as a result of lower error in estimating the change in radius directly from the acceleration directly on the radius vector compared to the Cartesian method requiring separate estimations of changes in xx and yy first using two finite accelerations at each iteration (Figure 2).

Some representative outputs from Astrocal for rr and θ\theta from data calculated to 79 significant figures for the orbit of the planet Mercury and for hypothetical planet of a range of differing masses are shown in Table 1, allowing a comparison of the radial (Leibnizian) and the Cartesian (Newtonian) models. Also shown are the radii generated for these orbits on the day when a half orbit is completed, close to aphelion. The main limitation to accuracy is that of the input data, but the results for the radial model are highly consistent with expectations, with the semilatus rectum reached when θ\theta is π/2\pi/2 radians after 18 days and aphelion or maximum radius reached between 87.5 and 88.5 siderial earth days, where θ\theta exceeds π\pi radians, just as expected for the planet Mercury.

Initially surprising to the authors, it was found that the two numerical methods do not automatically produce exactly the same outputs. The agreement between the two was excellent as long as the mass of the satellite was negligible compared to that of the Sun, or when the total mass (m+Mm+M) was kept constant. But when the mass of the hypothetical satellite approached that of the Sun, very large discrepancies are generated between these two reiterative numerical models. So the program recommended by Eisberg and Hyde [4] is not appropriate where mm is significant, shown in Table 1 by the discrepancies in the vectorial values of hh. Cartesian coordinates using the centre of mass as the origin are clearly needed to deal with larger masses or orbiting bodies.

SatelliteMercuryOne-third Sun’s massTwo Suns
MM (g) (Sun)1.9891×10331.9891 \times10^{33}6.6303×10326.6303 \times10^{32}1.9891×10331.9891 \times10^{33}
mm (g)3.285×10263.285 \times10^{26}9.995×10329.995 \times10^{32}1.9891×10331.9891 \times10^{33}
rpr_p (cm) ×10−12\times10^{-12}4.60012004.60012004.6001200
rar_a cm, rθr\theta ×10−12\times10^{-12}6.96364066.96315316.9631531
rar_a cm, XYXY ×10−12\times10^{-12}7.01001374.95081162.1127324
h=r2ωh=r^2\omega, ×10−19\times10^{-19}2.71320323.12901853.8354824
h=vy.x−vx.yh=v_y.x-v_x.y, ×10−19\times10^{-19}2.71320322.71320322.7132032
T/2T/2 days, rθr\thetaca. 44ca. 41ca. 32
T/2T/2 days, XYXYca. 44ca. 31ca. 14

Table 1. Orbital properties using radial (rθr\theta) and cartesian (XYXY) models for numerical calculation of aphelion (rar_a), specific action (hh) and approximate period (TT), with a radius at perihelion (rpr_p) of 4.6001200×10124.6001200 \times10^{12} cm

Semi-latus rectum = 5.5405×10125.5405 \times10^{12} cm.

In fact, the large discrepancies in Table 1 are trivial results of the different constraints placed on the orbits using the two methods, radial or polar and rectangular. In the case of the former, the program uses the length of the semilatus rectum to constrain the orbit whereas the rectangular method uses the orbital velocity vyv_y crossing the x-axis at perigee, to constrain the orbit, requiring that the specific action r2ωr^2\omega or hh should remain constant (see Equation (5)). The radial method, more correctly, requires that the specific action should increase with the square root of the extra mass, given that (M+m)G=h2/l(M+m)G = h^2/l (see Equation (14)) and ll is held constant for the polar calculations in Table 1, so hh must vary instead. In the case of the orbit of two bodies with the Sun’s mass, the value of the squared specific action h2h^2 is twice that with one major body, so no other change of inputs is necessary in the radial model.

Orbital plots for mercury and the other cases in Table 1 are shown in Figure 5. An advantage of the radial method is that centre of mass bipolar orbits around can be made around the barycentre with only small modifications of the code, separating rmr_m and rMr_M. The programmable SR-52 has a special function to interconvert polar and rectangular coordinates using the addressable memory register 00, so Cartesian plots as shown in the figure can readily be executed in the same program. In the plot for Mercury the centre of the Sun is 7.6 km from the barycentre at perigee and 11.5 km at apogee, showing that the Sun is also tracing out an ellipse about 4 km across, based on the common relationship with the semi-latus rectum. Figure 5 also demonstrates the symmetry of orbital motion with mass and the central role of the latus rectum.

Figure 5. Symmetrical radial orbital plots for planet Mercury but substituting masses equal to the Sun and one-third the Sun. For the Sun-Mercury system, only the planetary orbital motion can be shown, given the true barycentre is always within several km of the SUN’s centre. Note that the inertial acceleration (blue) is greater than the centripetal or gravitational acceleration at perigee while the reverse is true at apogee. Full data for these plots in Table 1 and for Mercury is given in appendices.

In a simplistic sense, black holes may even be considered as a spherical surface orbiting quanta around a barycentre with a diameter equivalent to rr with radii ror_o equal to the gravitational radius where the celerity CC or h2/lh^2/l is reduced to roc2r_oc^2. Newton apparently assumed that gravity was transmitted instantaneously between inertial systems so the concept of a gravitational radius or its derivative identified by Einstein, the Schwarzschild radius was not imaginable.

Given that the radial method produces results that are more accurate when compared with reality as observed with orbits of the Sun’s planets (see Table 2), this difference may teach us something about the nature of gravitation so far overlooked. In no case does the polar or radial method lead to an estimate for the radius at aphelion exceeding observations. However, the Cartesian method using rectilinear vectors slightly overestimates both the radius at aphelion and the period of the orbit, although it does give a consistent estimate of the specific action that is exactly in agreement with the data produced by the polar method. This discrepancy in the radius at aphelion is small, being about 1.7% in the case of Mercury and decreasing to 0.48% for Mars and 0.046% for Pluto. But it is significant given the need to obtain maximum accuracy for astronomical purposes. Figure 5 explains why this discrepancy occurs more readily with the Cartesian approach.

SatelliteMercuryVenusEarth-MoonMarsPlutoUnits
MM (Sun)1.9891×10331.9891 \times10^{33}1.9891×10331.9891 \times10^{33}1.9891×10331.9891 \times10^{33}1.9891×10331.9891 \times10^{33}1.9891×10331.9891 \times10^{33}g
MM (Planet)3.3022×10263.3022 \times10^{26}4.8685×10274.8685 \times10^{27}5.9737×10275.9737 \times10^{27}6.4185×10266.4185 \times10^{26}1.3050×10251.3050 \times10^{25}g
h2/l×10−26h^2/l \times10^{-26}1.327581261.327584281.327585021.327581461.32758104cm3^3 sec−2^{-2}
rp×10−12r_p \times10^{-12}4.6001200010.747625914.7098000020.6669000443.6824613cm
rar_a, rθ×10−12r\theta\times10^{-12}6.981069810.894139915.20978504124.9188772737.4297680cm
rar_a XY×10−12XY \times10^{-12}7.010013710.911194015.24624291025.03916669740.8509053cm
h=r2ωh=r^2\omega, ×10−19\times10^{-19}2.713203223.790116524.45588315925.47645123127.118944372cm2^2sec−1^{-1}
h=vy.x−vx.yh=v_y.x-v_x.y,2.713203223.790116524.45545586145.47645123127.118944372cm2^2sec−1^{-1}
T/2T/2, rθr\theta44112182340Ca. 45000days
T/2T/2, XYXY44112182350Ca. 46000days
Actual4411218234345260

Table 2. Orbital properties using radial (rθr\theta) and Cartesian (XYXY) models for numerical calculation of aphelion (rar_a), specific action (hh) and approximate period, for the planets

Accuracy depends largely on the accuracy of the inputs, but the radial method is inherently more accurate.

Calculations made in RR software have the advantage of giving graphical outputs. Some results for the Mercury-Sun system using the Leibnizian radial approach are shown in Figure 6. Plots indicating the cumulative effect of the radial accelerations in the attractive, centripetal direction (vcv_c) and in the repulsive, inertial direction (viv_i) are shown. These are obtained from the reiteratively summing the products of the net radial acceleration towards and away from the Sun, being the velocities (vrv_r) with which Mercury is either attracted towards (falling values of vrv_r) or propelled away (increasing values of vrv_r) from the Sun is shown. According to Leibniz’s model, these radial vectors are in opposition to each other, with the net radial accelerations and changes in momentum exactly cancelling each other during a complete orbit. The accelerations as indicated by vcv_c and viv_i are of equal magnitude when the satellite is on the latus rectum, at right angles to the major axis of the orbit, with the inertial acceleration (viv_i) being greater at perigee and the centripetal acceleration (vcv_c) in the rotating frame being greater at apogee (see Figure 6).

Figure 6. Orbital plots from use of Leibnizian radial accelerations to calculate the orbital parameters, centripetal velocity (vcv_c), inertial velocity (viv_i), radial velocity (vrv_r) radius (rr), for two orbital periods (θ=4π\theta= 4\pi) of the Mercury-Sun system in sec. Note the greater variation in viv_i compared to vcv_c, consistent with dependence on inverse radius cubed versus inverse radius squared; vrv_r is the difference between these two.

The plot in RR (Figure 6) shows that the maximum positive and negative values of vrv_r occur at these positions, as predicted by theory. The corresponding oscillation in the radius of the orbit (rr) is also shown. The plot for cumulative angular motion is not shown in the figure, but it has exactly the same form as the plot for vcv_c, rather than viv_i, since the variation in dθ/dtd\theta/dt or angular velocity ω\omega is inversely proportional to 1/r21/r^2 for r2ωr^2\omega constant during the orbit, just as the variation in vcv_c is inversely proportional to 1/r21/r^2. The fact that Newton's centripetal force is directly proportional to the angular velocity ω\omega no less than 1/r21/r^2 can hardly be without mechanistic significance.

The corresponding plots for Mercury-Sun using Cartesian vectors are shown in Figure 7, with the origin (x=0x=0, y=0y=0) placed at the centre of the Sun and the latus rectum being at right angles to the major axis that lies on the x-axis. When xx is zero, the magnitude of dr/dtd r/d t is maximal and it is zero when yy is zero, at perigee and apogee.

Figure 7. Plots from use of Cartesian vectors to calculate the orbital parameters, xx, yy (cm) and vxv_x, vyv_y (cm/sec) for two orbital periods of the Mercury-Sun system.

Examination of the outputs from RR by both methods indicate that the results are closely similar, provided the time interval chosen is small (say >50 reiterations per orbit) and within the limits of rounding errors for each method. However, they are not identical.

Conclusion

We were intrigued to find in October 2010 [4, 8], after having written down these relationships in 2005 based on earlier theory [5], that Leibniz had already described the motion of a planet in the 17th century as a contest between the attractive force and the inertial force caused by the rotation, acting to move the body along the rotating radius vector. Leibniz even wrote down one of the first differential equations d2r/dt2=a/r3−b/r2d^2r/dt^2 = a/r^3 - b/r^2, formally equivalent to Equation (15), to describe the resultant acceleration of the radius, where rr is the distance from the centre of attraction and aa and bb are different constants than the semi-major and -minor axes. He drew attention to the significant fact that the inertial force and the centripetal gravitational force were inversely proportional to the cube and the square of the radius respectively.

Leibniz’s constants aa and bb in his equation would have values of h2h^2 and h2/lh^2/l respectively to be considered exactly the same as Equation (15). But it is not clear to the authors if Leibniz considered any significance for the semilatus rectum. Newton objected to Leibniz’s solution [5, 9], since he claimed the centrifugal force must be equal and opposite to the attraction by his third law of motion, pointing out that Leibniz’s equation implied that d2r/dt2\mathrm{d}^2r/\mathrm{d}t^2 — logically expressed by Leibniz in his original notation as ddr/ddt\mathrm{d}\mathrm{d}r/\mathrm{d}\mathrm{d}t — was zero. Meli [9] discussed Newton’s insistence on his third law as operating between the forces without a clear resolution. However, we have shown that the equality might be better accepted as equations of inertial and gravitational forces between the orbiting bodies around the centre of mass rather than equation of the inertial and centripetal forces. Given Leibniz’s role in developing Equation (15) for the second derivative of the radius with time, the fact that he was the first to correct the prevailing Cartesian opinion of the conservation of momentum to the conservation of energy as in the vis viva (mv2mv^2) is not surprising. Leibniz is also credited with being the first to perceive the role of least action, later taken up by Maupertuis and others.

This paper shows the mathematical equivalence of Newton’s idea (discussed with Hooke) that only one gravitational or centripetal force varying with the inverse of the radius squared is needed to produce the orbital motion and Leibniz’s idea that it is the result of an ongoing contest between gravitational and inertial effects. We conclude that for the assumption of a rectilinear inertial reference system with a stationary origin (or one moving in a straight line with uniform velocity), Newton is correct; and for a radial frame of reference with the origin located at either the centre of MM or the centre of mass of MM and mm, Leibniz is correct. Newton’s reference frame requires Kepler’s second law of equal areas and equal times to ensure that the elliptical orbit will be closed whereas Leibniz’s rotating frame allows the gravitational acceleration to exceed the inertial acceleration for most of the orbit, causing its eventual closure. Thus, the two points of view differ only in their initial assumptions and can be reconciled. Although similar in principle, these two methods to plot orbits involve subtle differences that require explanation by further analysis. This is despite the facts that both models used exactly the same numerical inputs and the same total accelerations are expected to be exerted from Newton’s law of gravity.

An obvious difference is that the radial method makes no assumptions regarding the resolution of linear vectors at each reiterative step required by the Cartesian method to mimic curvilinear motion. However, the inherent inaccuracy using Newton's approach can be much diminished by choosing shorter periods, though at the cost of more computing time. In the radial method, only the acceleration of the radial velocity need be considered, the rotating frame of reference obviating any need to approximate curved trajectories. It should be observed that the net radial acceleration is zero twice during each orbit, when the radius vector is normal to the major axis falling on the latus rectum; the radial acceleration reaches its maximum at perihelion with another peak at aphelion, although of lower magnitude and opposite in direction.

The rectangular accelerations given in Equations (8) and (9) are considered very accurate as infinitesimals but average values of these for each period are actually used. But very little additional orbital accuracy is gained by increasing the number of periods beyond the 35-50 used in these simulations for semi-orbits with eccentricity less than 0.25.

For highly elliptical orbits such as that of Halley’s comet with an eccentricity of 0.967, it is important when using the numerical methods given in the appendices to shorten the time interval near perihelion and to lengthen it near aphelion, to avoid inaccuracy or too long a plot, no matter whether the polar or the Cartesian method is employed. In the case of Halley’s comet, an extremely small proportion of its orbit is occupied by that section where the radius in less than the semilatus rectum, so that the centripetal acceleration is nearly always greater than the inertial acceleration, except when the comet lies inside the Earth’s orbit close to the Sun. Numerical plots using varying time periods so that δθ\delta\theta is constant for each iterative step would be preferable in such cases, to maximise accuracy and minimise computing time.

Does acceptance of the validity of the Leibnizian approach after all have further implications for gravitational theory? Some physicists consider that rectilinear inertial reference systems are to be preferred as mathematically more correct — despite the fact that none can be identified in reality, either stationary or moving in a straight line — even the Sun orbits the galaxy which also orbits its neighbours. Yet they declare that the centrifugal force associated with such curvilinear motion is fictitious. Certainly, it can be agreed that the specific accelerations from gravitation caused by massive bodies and that resulting from the tendency of inertial motion to continue are not in the same class as forces. But inertial accelerations are both real and vary in intensity, as shown by the ability to place satellites in higher orbits using the accelerative thrust of rockets. The inertial tendency to continue in a straight line is clearly a matter of degree and not absolute in effect.

Given that the inertial acceleration directed outwards is calculated continuously as a derivative of the gravitational or centripetal acceleration inwards, there is also the prospect that the influence of many gravitational bodies acting at once can be modelled more effectively in the Leibnizian model used here. Estimated centripetal accelerations can be directed normally and resolved on every significant current radial separation, all diminishing by the inverse square rule. Since the inertial acceleration is exerted normal to the current circular position, it seems to use the distant stars as reference in a Machian sense and nearby objects have no influence on inertia which is self-referencing.

Extending this discussion to relativity is simplified if inertial and gravitational accelerations are considered separately. It is clear that Einstein’s general relativity is a function of massive bodies. Yet special relativity concerns inertial effects, though these do interact with gravity. In separate papers (Kennedy and Hodzic, in preparation) we have proposed that accurate relativistic corrections for GPS clocks can be made with algorithms combining gravitational, with reference to the shortening of space by the gravitational radius ror_o, including inertial corrections. It may well be that Leibniz’s perception that these effects can be considered independently can resonate four centuries later.

Acknowledgements

We would like to acknowledge encouragement and criticism from colleagues, particularly James McCaughan (regarding physics) and the late Barrie Fraser (regarding applied mathematics). A supplementary paper in which we focus on least action, written with Barrie Fraser, provides a justification of the radial approach used here by confirming that Leibniz’s conclusion is consistent with Newtonian vectorial theory.

Conflicts of Interest: The authors declare no conflict of interest.

Supplementary material

Table S1: OrbitRadial

Locn.CodeInstruct.CommentLocn.CodeInstruct.Comment
000096LABEL042096LABEL
001073I043087W# Run program
002107STO MEM044040(
0030491045109RCL MEM
0040480# Interval0460480
005119printδt\delta t sec0470524# M
006002HALT048043+
007096LABEL049109RCL MEM
008082R0500480
009107STO MEM0510535# m
0100480052041)
0110491# Radius at t0t_{0}053042x
012119printrr cm0540546# G
013002HALT055046.
014096LABEL0560546
015076L0570557
016107STO MEM0580524
01704800590502
0180502# Semilatus rectum0600568
019119printll cm061047/
020002HALT0620491
021096LABEL063101EE# expon.
022084T0640568
023107STO MEM065061=
0240480066107STO MEM
0250513# Angular motion0670480
026119printθ\theta radians0680546# Celerity
027002HALT069119printC cm3^{3}s−2^{-2}
028096LABEL0700502
029077M0711141/x
030107STO MEM072042x
0310480073109RCL MEM
0320524# Mass central body0740480
033119printM g0750491# r
034002HALT076113Square
035096LABEL0771141/x
036078N078042x
037107STO MEM079109RCL MEM
03804800800480
0390535# Mass minor body0810546
Locn.CodeInstruct.CommentLocn.CodeIntruct.Comment
040033printm g082041)
041002HALT083061=

Table S1. OrbitRadial

Program OrbitRadial to calculate rr, θ\theta using a Leibnizian radial model by integrating the differential equation d2r/dt2=h2/r3−h2/lr2\mathrm{d}^{2}r/\mathrm{d}t^{2}=h^{2}/r^{3}-h^{2}/lr^{2} (see Equation (15)).

ASTROCAL CODE

OrbitRadial (contnd.)

Locn.CodeInstruct.CommentLocn.CodeIntruct.Comment
084107STO MEM125109RCL MEM
08504801260491
0860557# Centripetal1270502# vrv_{\mathrm{r}}
087042xacceleration aca_{\mathrm{c}}128040(
088109RCL MEM129109RCL MEM
08904801300491
0900502# 1 semilatus rectum1310491# δvr\delta v_{\mathrm{r}}
091047\132042x
092109RCL MEM133109RCL MEM
09304801340491
0940491# r current radius1350480# δt\delta t
095041)136041)
096107STO MEM137117SUM MEM
0970480# Inertial1380491
0980568acceleration aia_{\mathrm{i}}1390513# δr\delta r
099109RCL MEM140109RCL MEM
10004801410480
1010568# aia_{\mathrm{i}}1420491# r
102045-143107STO MEM
103109RCL MEM1440491
10404801450524# r old
1050557# aca_{\mathrm{c}}146043+
106061=147109RCL MEM
107107STO MEM1480491
10804801490513# δr\delta r
1090579# Radial150061=
110040(acceleration ara_{\mathrm{r}}151107STO MEM
111109RCL MEM1520480
11204801530491# r new
1130579# ara_{\mathrm{r}}154119print
114042x155109RCL MEM
115109RCL MEM1560480
11604911570568# ai=rω2a_{\mathrm{i}} = r\omega^2
1170480# δt\delta t158047/
118041)159109RCL MEM
119107STO MEM1600491
12004911610524# r old
1210491# δvr\delta v_{\mathrm{r}}162061=
122117SUM MEM163107STO MEM
12304911640491
1240502# vr=dr/dtv_{\mathrm{r}} = \mathrm{d}r/\mathrm{d}t1650535# ω2\omega^2

Table 1 (PDF p. 28).

NB The first iteration is a half interval (070-072), to avoid overestimating δr\delta r by using the maximum acceleration for the whole interval.

OrbitRadial (contnd.)

Locn.CodeInstruct.CommentLocn.CodeInstruct.Comment
166092Root193043+
1670502194109RCL MEM
168061=# ω=dθ/dt\omega= \mathrm{d}\theta/\mathrm{d}t1950480
169040(1960513# θ\theta old
170107STO MEM197061=
1710491198117SUM MEM
1720546# ω\omega1990491
173040(2000557# θ\theta new
174109RCL MEM201109RCL MEM
17504912020491
1760524# rr old2030557# θ\theta new
177113Square204119print
178042x# r2r^2 old205040(
179109RCL MEM2060524
1800491207042x
1810546# ω\omega208112pi# 2 orbits
182041)# r2ωr^2\omega209045-
183119print210109RCL MEM
184040(2110491
185109RCL MEM2120557# θ\theta new
1860491213041)
1870546# ω\omega214037If pos
188042x2150480
189109RCL MEM2160480
19004912170557
1910480# δt\delta t2180513# Goto 73
192041)# δθ\delta\theta219002HALT

Table 3.

The program may be continued to any number of orbits by increasing the factor on location 206.

Locn.CodeInstruct.CommentLocn.CodeInstruct.Comment
000096LABEL042096LABEL
001065A043078N
002107STO MEM044107STO MEM
00304800450491
0040491x, radius at perigee0460502minor mass m
005119print4.60012x10^12047119print
006002HALT048002HALT
007096LABEL049096LABEL
008066B050080P
009107STO MEM051040(
0100480052109RCL MEM
0110502vxv_x at perigee=00530491
012119print0540491M
013002HALT055043+
014096LABEL056109RCL MEM
015067C0570491
016107STO MEM0580502m
0170480059041)
0180513y=0 at perigee060042x
019119print0610546G
020002HALT062046.
021096LABEL0630546
022068D0640557
023107STO MEM0650524
02404800660502
0250524vyv_y=5898657 cm s−1^{-1}0670568
026119print068047/
027002HALT0690491
028096LABEL070101EE
029069E0710568
030107STO MEM072061=
0310480073107STO MEM
0320535δt\delta t=866164 s0740491
033119print0750513Celerity
034002HALT076119print
035096LABEL0770552
036070M0781141/xInvert
037107STO MEM079042x
0380491080040(
0390491Sun’s mass M081040(
040119print082109RCL MEM
041002HALT0830480

Table S2. ASTROCAL CODE OrbitVector. Program to calculate xx, yy using a Newtonian vectorial model d2x/dt2+d2y/dt2=−C.x/r3−C.y/r3\mathrm{d}^2x/\mathrm{d}t^2+\mathrm{d}^2y/\mathrm{d}t^2=-C.x/r^3-C.y/r^3 (see Equations (15) and (16)).

NB The first iteration is a half interval (locations 077-079), to avoid overestimating δr\delta r by using the maximum acceleration for the whole interval.

OrbitVector

(contnd.)

Locn.CodeInstruct.CommentLocn.CodeInstruct.Comment
0840411# x124109RCL MEM
085113Square# x2x^21250480
086043+1260557
087109RCL MEM127042x
0880480128109RCL MEM
0890513# y1290480
090113Square# y2y^21300513# y
091041(131061=
092092Root132117SUM MEM
09305021330480
094041)1340524# New vyv_y
0951141/x135109RCL MEM
096091Exponent1360480
0970513137
098061=# 1/x31/x^31380524
099042x139109RCL MEM
100109RCL MEM1400480
10104911410535# δt\delta t
1020513# C142117SUM MEM
103111+/-# Change sign1430480
104042x1440568# δy\delta y
105109RCL MEM145109RCL MEM
10604801460480
1070535# δt\delta t1470568
108061=# Cδt/x3C\delta t/x^3148119print
109107STO MEM149109RCL MEM
11004801500480
1110557#1510535# δt\delta t
112042x152042x
113109RCL MEM153109RCL MEM
11404801540480
1150491# x1550502# vxv_x
116061=# δvx\delta v_x156061=
117117SUM MEM157117SUM MEM
11804801580480
1190502# New vyv_y1590491# New x
120109RCL MEM160109RCL MEM
12104801610480
1220502# vxv_x1620491
123119print163119print# x

Table 1 (PDF p. 31).

OrbitVector

(contnd.)

Locn.CodeInstruct.CommentLocn.CodeInstruct.Comment
164109RCL MEM1950480
16504801960502# vxv_x
1660535# δt\delta t197041)
167042x198061=# h = r2ωr^2\omega
168109RCL MEM199109RCL MEM
16904802000480
1700524# vyv_y2010491# x
171061=# δy\delta y202107STO MEM
172117SUM MEM2030480
17304802040480
1740513# New y205109RCL MEM
175109RCL MEM2060480
17604802070513# y
1770513# y208064Inverse
178119print209106Pol/Rect
179040(210119print# θ\theta radians
180109RCL MEM211109RCL MEM
18104802120480
1820491# x2130480# r cm
183042x214119print# r
184109RCL MEM215109RCL MEM
18504802160480
1860524# vyv_y2170513# y
187041)218037If pos# continues
188045-2190480while y is
189040(2200480positive
190109RCL MEM2210568
19104802220480
1920513# y223002HALT
193042x
194109RCL MEM

Table 4 (PDF p. 32).

The program can be run longer than π\pi radians by linking the condition in OrbitVectorial location 218 to θ\theta, as in OrbitRadial (locations 205-214).

Table S3. R script for OrbitRadial using differential equation dvr/dt=C∗l/r3−C/r2d v_r/dt = C*l/r^3-C/r^2 where C=h2/lC=h^2/l. (see Equation (15)).

R CODE

OrbitRadial

require(simecol)

orbitpolar <- new("odeModel",
  main = function(time, init, parms, inputs=NULL) {
    m   <- parms["m"]       # mass of minor body
    M   <- parms["M"]       # mass of major body
    G   <- parms["G"]       # universal gravitational constant
    as  <- parms["as"]      # semi-major axis
    ec  <- parms["ec"]      # eccentricity
    l   <- parms["l"]       # semi-latus rectum
    vc  <- init[1]          # centripetal velocity
    vi  <- init[2]          # inertial velocity
    vr  <- init[3]          # net radial velocity
    r   <- init[4]          # radius
    T   <- init[5]          # phi in radians to major axis

    l   <- as*(1-ec^2)      # semi-latus rectum
    C   <- (m+M)*G          # celerity = h^2/l
    h   <- (C*l)^(1/2)       # specific action = r^2dT/dt

    dvi <- C*l/r^3          # inertial acceleration
    dvc <- C/r^2            # centripetal acceleration
    dvr <- C*l/r^3-C/r^2    # net radial acceleration
    dr  <- vi-vc            # radial velocity
    dT  <- (h/r^2)          # angular velocity

    list(c(dvc, dvi, dvr, dr, dT))
  },
parms = c(
    m  = 3.3022*10^26,             # mass of minor body
    M  = 1.9891*10^33,             # mass of major body
    G  = 6.67428*10^-8,             # Gravitational constant
    as = 5.79091*10^12,             # Semimajor axis
    ec = 0.205630,       # Eccentricity
    l  = 5.5460489*10^12 # Semilatus rectum
),
times = c(from=0, to=86164*176, by=86164),
init  = c(vc=0, vi=0, vr=0, r=4.60012*10^12, T=0),
solver = "lsoda"
)
orbpol <- sim(orbitpolar)     # initial simulation
plot(orbpol)
print(orbpol, all=TRUE)       # returns all values in the all "slots"
R script for OrbitVector integrating differential equations
dvx/dt=−C∗x/((x2+y2)(3/2))dv_x/dt = -C*x/((x^2+y^2)^{(3/2)}) and dvy/dt=−C∗y/((x2+y2)(3/2))dv_y/dt = -C*y/((x^2+y^2)^{(3/2)}),
or dvx/dt=−Cx/r3dv_x/dt = -Cx/r^3 and dvy/dt=−Cy/r3dv_y/dt = -Cy/r^3 (see Equations (15) and (16)).

Table S4.

R

OrbitVector

require(simecol)

orbitvector <- new("odeModel",

    main = function(time, init, parms, inputs=NULL) {
        m    <- parms["m"]           # mass of minor body
        M    <- parms["M"]           # mass of major body
        G    <- parms["G"]           # universal gravitational constant
        C    <- parms["C"]           # celerity = (m+M)G
x <- parms["x"]             # x-vector distance
y <- parms["y"]             # y-vector distance
vx <- parms["vx"]           # x-vector velocity
vy <- parms["vy"]           # y-vector velocity

vx <- init[1]
vy <- init[2]
x <- init[3]
y <- init[4]

C <- (m+M)*G

dvx<- -C*x/((x^2+y^2)^(3/2))
dvy<- -C*y/((x^2+y^2)^(3/2))
dx<- vx
dy<- vy

list(c(dvx, dvy, dx, dy))
},

parms = c(
    m = 3.3022*10^26,          # mass of minor body
    M = 1.9891*10^33,          # mass of major body
    G = 6.67428*10^-8,         # Gravitational constant
    l = 5.5460489*10^12       # Semilatus rectum
),
times = c(from=0, to=86164*176, by=86164),
init = c(vx=0, vy=5898657, x= 4.60012*10^12, y=0),
solver = "lsoda"
)

orbvec <- sim(orbitvector)             # initial simulation
plot(orbvec)
print(orbvec, all=TRUE)                 # returns all values in the all "slots"

References

  1. [1]F.R Moulton, An Introduction to Celestial Mechanics, 1914. Second Revised Edition, (Republished Dover Publications New York 1970).
  2. [2]W.M. Kaula, Theory of Satellite Geodesy. 2000. Dover publications, Mineola, New York.
  3. [3]R.P. Feynman, R.B. Leighton and M. Sands, The Feynman Lectures on Physics, New Millenium Edition, Vol. 1, pp. 9.6-9.9. California Institute of Technology 1963.DOI
  4. [4]R. Eisberg, and W. Hyde, Countdown, Skydiver, Rocket and Satellite Motion on Programmable Calculators. (Dilithium Press, Portland Oregon, 1979), pp. 69-80.
  5. [5]E.J. Aiton, An episode in the history of celestial mechanics and Its utility in the teaching of applied mathematics, In Learn from the Masters, Frank Swetz, John Fauvel, Bengt Johansson and Victor Katz, Eds., (Mathematical Association of America, 1997).
  6. [6]C.G. Lambe, Dynamics. Teach Yourself Books (The English Universities Press, London, United Kingdom, 1972).
  7. [7]I.R. Kennedy, Action in Ecosystems: Biothermodynamics for Sustainability. (See Glossary entry on Gravity in particular) (Research Studies Press, 251 pages, Baldock, United Kingdom, 2001).
  8. [8]I. Newton, Mathematical principles of natural philosophy. Great Books of the Western World 34. Newton, Huygens, R.M. Hutchins Ed. Encyclopedia Britannica Inc. 1982 Chicago.
  9. [9]D.B. Meli, Inherent and centrifugal forces in Newton. Archives Historical Exact Sciences 60, 319-335. (DOI) 10.1007/s00407-006-0108-6, 2006.DOI

Paper details

vlnc:W_ajta37e2
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