AAS 12-213

A CLOSED FORM SOLUTION OF THE TWO BODY PROBLEM

IN NON-INERTIAL REFERENCE FRAMES
Daniel Condurache* and Vladimir Martinusi
A comprehensive analysis, together with the derivation of a closed form solution to the two-body problem in arbitrary non-inertial reference frames are
made within the present work. By using an efficient mathematical instrument,
which is closely related to the attitude kinematics methods, the motion in the
non-inertial reference frame is completely solved. The closed form solutions
for the motion in the non-inertial frame, the motion of the mass center, and the
relative motion are presented in the paper. Dynamical characteristics analogue
to the linear momentum, angular momentum and total energy are introduced.
In the general situation, these quantities may be determined as functions of
time, and their derivation is presented within the paper. In the situation where
the non-inertial frame has only a rotation motion, these quantities become first
integrals in a larger sense, with respect to an adequately defined differentiation
rule.

INTRODUCTION
The paper approaches the two-body problem in an arbitrary non-inertial reference frame, offering
its solution in the most general case, where the non-inertial frame has an arbitrary rotation and
translation. To the knowledge of the authors, such a solution is not present in textbooks [1, 2, 3, 4],
perhaps due to the unusual expression of the rst integrals.
The study of the problem in a non-inertial frame, in place of the classical study in an inertial one,
has two simple motivations here: rst, it is a problem which models real situations, for example
the way a telescope orbiting Earth sees the relative motion between two attracting bodies, or it is
the way the same motion is seen by an Earth-xed observer, from an Earth-xed reference frame,
which rotates around the polar axis, and also translates in an orbit around the sun. Second, it is
a very interesting mathematical problem, which has not been approached comprehensively in the
literature.
Assuming that the motion of the non-inertial frame is known, the solution is determined by using
a tensorial instrument introduced in 1995 by one of the authors [5]. This method is similar to the
approach used in rigid body kinematics, based on proper orthogonal and skew-symmetric tensor
valued functions, related by the Darboux equation, also known as the attitude kinematics equation.
The paper is organized as follows. After the presentation of the non-linear initial value problems
which model the motion of two interacting bodies with respect to a general non-inertial frame,
the tensor method is presented, together with a vector differentiation operator, which makes the

Professor, Department of Theoretical Mechanics, Technical University Gheorghe Asachi, 700050, Iasi, Romania.
Member AIAA, AAS. E-mail: [email protected].

Post-doctoral Fellow, Faculty of Aerospace Engineering, Technion - Israel Institute of Technology, 32000, Haifa, Israel.
E-mail: [email protected]

1649

connection between the derivative in a rotating frame and the derivative in the inertial frame. One
has to remark that all the computations are made in the non-inertial frame, and the inertial one is
used only as a catalyst in solving the problem. The study of the two-body problem in non-inertial
reference frames is made rst by studying the motion of the center of mass of the system, then by
studying the relative motion of one body with respect to a reference frame attached to the other one,
and the relative motion of both bodies with respect to a reference frame attached to the center of
mass. The closed form solution for the motion of each body with respect to the original frame is
also presented. The motion with respect to the frame attached to the mass center is comprehensively
studied, and new rst integrals of the motion are obtained. These rst integrals allow to offer a
geometrical visualization of the motion.
In the end, dynamical characteristics of the motion, analogue to the classical linear momentum,
angular momentum and total energy are introduced, and a method to determine them is also offered.
In the situation where the non-inertial frame is only a rotational one, they become rst integrals
of the motion, in a larger sense, and a generalized potential energy function is also introduced.
Particularizing to the case where the reference frame has a constant instantaneous angular velocity,
the aforementioned function becomes the classical potential energy. The present work constitutes
the basis for future research, where the full-body problem will be studied, where the internal torques
of the system are going to be incorporated into the mathematical model.
PROBLEM FORMULATION
The mathematical model for the two body problem in non-inertial reference frames is represented
by the initial value problems (IVPs):

r1 + 2m1 r 1 +m1 ( r1 ) + m1 r1 + m1 aF = F12 ,

m1


(1)
r1 (t0 ) = r01

r 1 (t0 ) = v01

r2 + 2m2 r 2 +m2 ( r2 ) + m2 r2 + m2 aF = F21 ,

m2


(2)
r2 (t0 ) = r02

r 2 (t0 ) = v02
where rk denotes the position vector, r k the velocity vector of the particle Pk , k = 1, 2, related to
the reference frame where the motion takes place, mk are the masses of the two particles, aF is
the acceleration and is the angular velocity of the non-inertial reference frame where the motion
takes place. The vector map of real variable is supposed to be differentiable, and aF is supposed
to be continuous. Fij denotes the force that acts upon the particle Pi due to the interaction with
particle Pj , i = j, i, j = 1, 2. The approach is based on the assumption:
Fij = Fij (|ri rj |) , i = j, i, j = 1, 2.
From the principle of reciprocal interactions, it follows that:

F12 + F21 = 0;
F12 (r1 r2 ) = 0.

(3)

(4)

Within this paper, it is going to be proved that the two-body problem in non-inertial reference
frames may be solved exactly like in the inertial case, by solving two single-particle problems:

1650

1. The motion of the center of the center mass, denoted with C; its position vector with respect
to the non-inertial frame F is denoted with rC (also see Figure 1);
2. The relative motion of one particle related to a reference frame associated to the other one,
for example the motion of particle P2 related to P1 . This motion is described by the vector
map r = r2 r1 , where r2 is the solution to IVP (2) and r1 is the solution to IVP (1).

Figure 1. The two-body problem in a non-inertial reference frame.

The closed form solution will be presented with the help of a tensor instrument, rst introduced in
Ref. [5]. It is based on proper orthogonal and skew-symmetric tensor maps of real variable. A key
role is played by a vector differentiation operator, which relates the motion in a rotating reference
frame to the motion in an inertial frame.
By using this tensor instrument, the following results are presented herein:
An explicit solution for the general motion of the center of mass;
A representation theorem of the relative motion; the problem is reduced to the study of the
motion of one mass point in an arbitrary central force eld;
novel rst integrals, replicas to the linear momentum, angular momentum and energy conservation laws.
TENSORIAL CONSIDERATIONS
This section introduces the main mathematical instruments used in this paper. A tensorial map
and a vectorial differential operator will be dened. The following denotations are introduced:
V3 the three-dimensional space of free vectors;
VR
3 the set of functions of real variable, with values in V3 ;
SO3 the special orthogonal group of second order tensors:


SO3 = Q | QT Q = I3 , det (Q) = 1 ;
SOR
3 the set of functions of real variable, with values in SO3 ;

1651

(5)

 so3 the Lie-algebra of skew-symmetric second order tensors:



T =

;
SO3 =

(6)

soR
3 the set of functions of real variable, with values in so3 .
A Tensor Operator
The precession with arbitrary angular velocity is related to proper orthogonal tensor maps of
real variable by a tensor IVP, similar to the one from attitude kinematics, which is also referred to
as the Darboux equation.
Lemma 1 Consider the IVP:
= Q

; Q (t0 ) = I3 , t0 0.
Q

(7)

soR
Then there exist a unique solution Q SOR
3 , for any continuous map
3.
Proof. From the existence and uniqueness theorem, it follows that IVP (7) has a unique solution
T
Q = Q (t). One has to prove that Q is in SOR
3 , meaning that Q Q = I3 and det (Q) = 1.


d
T + QQ
T = Q

QT Q

QT = 03 , so QQT is a constant
QQT = QQ
We have that dt


T
differentiable function that satises QQ (t0 ) = I3 . Then, QQT = I3 . Using that det (Q) is
also a continuous function which satises det (Q) {1, 1} and det (Q (t0 )) = det I3 = 1, it
follows that det (Q) = 1. So Q SOR
3.
Remark 2 Lemma 1 is the well-known Darboux problem (also named the attitude kinematics
equation) (also see Ref [6]): determining the rotation tensor when the instantaneous angular velocity is known. The link between the rotation tensor map and the skew-symmetric tensor associated
to the angular velocity vector is given by IVP (7).
The solution to IVP (7) will be denoted F . The next result presents the properties of this
tensorial orthogonal map.
Lemma 3 The map F satises:
1. F is invertible;
2. F u F v = u v, () u, v VR
3;
3. |F u| = |u| , () u VR
3;
4. F (u v) = F u F v, () u, v VR
3;


d

5.
F u = F u + u , () u VR
3 , differentiable;
dt


d2

6. 2 F u = F u + 2 u + ( u) + u , () u VR
3 , differentiable.
dt
The proof of Lemma 3 may be performed by elementary manipulations, therefore it is omitted.

1652

Comments and Remarks

1. The following denotation is introduced:

(F )1 = R

(8)

Since F is the solution to IVP (1), it follows that R is the proper orthogonal tensor map
associated to the instantaneous angular velocity , therefore it obeys the IVP:
+

R = 0; R (t0 ) = I3 .
R

(9)

2. There exists a numerical solution of the initial value problem (9) when the instantaneous
angular velocity is an arbitrary continuous vector function. This is known as the PeanoBaker solution, it is obtained by iteration [7] and it is presented as a limit of innitesimal
integrals.
t
t


1

(t1 ) , ...,

(tn )] ,
dt1 ... dtn T [
(10)
R (t) = I3 +
n!
n=1

t0

t0

with

n1
n
n 






(1)

(t1 ) , ...,

(tn )] =

t(p) ,
T [
t(k) t(k+1)

2n
P(n)


(t) =

(11)

p=1

k=1

0, t t0
,
1, t > t0

(12)

and P (n) denotes the group of permutations of the set {1, ..., n} , n 2.
In case has xed direction, =
, with
constant unit vector and : R R,

(t1 )

(t2 ) =

(t2 )

(t1 ), () t1,2 R, (also see Refs. [5, 8, 9]), then R

since
has the explicit expression:

R (t) = exp

t

(s) ds = I3 sin (t)

t0

where

+ [1 cos (t)]

 2

, (13)

t
(t) =

(s) ds.

(14)

t0

If is constant, then R has the explicit expression:

] = I3 sin [ (t t0 )]
R(t) = exp [ (t t0 )

1653

+{1 cos [ (t t0 )]}

 2

(15)

 If vector has a regular precession with angular velocity 1 around a xed axis, expressed mathematically like:

1] ,
= R1 0 ; 0 = (t0 ) ; R1 = exp [(t t0 )

(16)

then the IVP (9) still has a timeexplicit solution [10, 11], expressed like:

1 ] exp [ (t t0 ) (

1 +

0 )] ,
R (t) = exp [(t t0 )

(17)

and written explicitly:


R

 2 

=
I3 + sin [1 (t t0 )]
+ {1 cos [1 (t t0 )]}
(18)
1
1

 2 

I3 sin [ (t t0 )] + {1 cos [ (t t0 )]}

,

where:
= 1 + 0.

(19)

A comprehensive study, together with a closed-form solution to the IVP (9) in the general case, may be found in Ref. [10]
Remark 4 Equations (13), (15) and(18) provide the closed form solution to the Darboux equation
in the situation where vector has xed direction, it is constant, and it has a regular precession,
respectively.
A Vector Differentiation Operator
A vector differential operator related to the angular velocity is introduced. It relates the derivative of a vector valued function in an inertial reference frame to the derivative of the same vector
function expressed in a rotating reference frame. Like the regular derivative, it admits an inverse
operator, dened within this section. This derivation rule will prove to be useful in the study of the
motion with respect to a non-inertial reference frame.
R
Dene the vector valued function differentiation rule ( ) : VR
3 V3 by:

( ) = ( ) + ( )

(20)

For any arbitrary vectorial map u : R VR

3 , it stands:
u = u + u

(21)

The next result presents the properties of this operator, together with the link to the previously
dened tensor valued function F .
Lemma 5 The following afrmations hold:

1.  = ;

1654


2. (u + v) = u + v , () u, v C 2 V3R .

 , () u C 2 V R , () : R R, differentiable.

3. (u) = u+u
3




4. (u v) = u v + u v , () u, v C 2 V3R .

d
(u v) , () u, v C 2 V3R .
5. u v + u v = u v + u v = dt

6. u = u
+ 2 u + ( u) + u, () u C 2 V3R .

d
(F u) = F (u ) , () u C 2 V3R .
dt


d
(F u)
= u (t0 ) + (t0 ) u (t0 ) ;
8. F u|t=t0 = u (t0 ) ;
dt
7.

t=t0

The proof to Lemma 5 may be made by elementary mathematical manipulations, therefore it will
not be presented here.
The vector differentiation dened in Equation (20) makes the connection between the derivative
of a vector referred to a reference frame which rotates with angular velocity (denoted with a dot
above) and the derivative of the same vector referred to an inertial reference frame (denoted with
prime), as it is depicted in Figure 2.

Figure 2. The vector differentiation operator transports the derivative from the rotating reference frame to the inertial reference frame.

The anti-derivation rule associated to the differentiaition rule ( ) is presented below.

Lemma 6 Consider b = b (t) a continuous vector valued function. Then the solution to the IVP:
u = b, u (t0 ) = u0
is expressed like:

u = R u0 +

t

(22)

RT (s) b (s) ds

(23)

t0

where R is dened in Equation (8).

Proof. Apply the tensor operator F to IVP (22) and take into account point (8.) from Lemma 5. It
follows that:
d
(24)
(F u) = F b; F u|t=t0 = u0
dt

1655

By using point (7) from Lemma 5, together with the initial conditions from Equation (22), it follows:
t
F u = u 0 +

RT (s) b (s) ds.

(25)

t0

Equation (23) is obtained by applying R to the equality (25). The proof is nalized.
Remark 7 From Lemma 6 it follows that if a vector map u : R+ VR
3 obeys the IVP:
u = 0, u (t0 ) = u0 ,

(26)

then vector u is the rotation with angular velocity of a constant vector u0 = u (t0 ):
u = R u0 .

(27)

This remark will help to offer a geometrical interpretation of the rst integrals of this particular
dynamical system.
THE STUDY OF THE TWO BODY PROBLEM IN NON-INERTIAL REFERENCE FRAMES
This is the main section of the paper, presenting the theoretical study of the two-body problem
with respect to a non-inertial reference frame. The motion of the center of mass, the relative motion
of a particle with respect to an adequately chosen reference frame attached to the other one, the
prime integrals of the dynamical system formed by the two mass points, the relative motion of
the two particles with respect to a reference frame attached to the center of mass are presented,
with closed form equations of motion and novel features. The essential result is presented within
Theorem 9, which relates the inertial and the non-inertial approaches of the studied problem via an
proper orthogonal tensor map of real variable.
The Motion of The Center of Mass
This Section presents the closed form solution for the motion of the center of mass. Introduce the
denotation:
m1 r1 + m2 r2
.
(28)
rC =
m1 + m2
Vector rC models the position of the center of mass of the dynamical system comprising the two
mass points with respect to the non-inertial reference frame to which the motion is referred. The the
following afrmation hold:
m1 r 1 + m2 r 2
r C =
(29)
m1 + m2
By summarizing eqs (1) and (2) and taking into account Equations (1) and (2), it follows that the
vector function rC obeys the IVP:

rC + 2 r C + ( rC ) + rC + aF = 0,

m1 r01 + m2 r02 0

rC (t0 ) =
= rC

(30)

m1 + m2

m v0 + m2 v02 0

r C (t0 ) = 1 1
= vC
m1 + m2

1656

Theorem 8 The solution to IVP (30) is:

t s


rC (t) = R r0C + v0C + 0 r0C (t t0 ) RT ( ) aF ( ) d ds , t t0 .

t0

where

t0

(31)

= (t0 ).

Proof. Remark that IVP (30) may be written with the help of the differential operator ( ) like:

rC = aF ,

rC (t0 ) = r0C ,
(32)

rC (t0 ) = v0C + 0 r0C .

Apply the tensor operator F to IVP (32) and make the change of variable:

C = F r C ;

(33)

then IVP (32) transforms into:

C = F aF ,


C (t0 ) = r0C ,

C (t0 ) = v0C + 0 r0C .

(34)

By solving IVP (34) through direct integration, the solution to IVP (32) is obtained by taking into
account substitution (33), and it is expressed like:

t s


(35)
rC (t) = R r0C + v0C + 0 r0C (t t0 ) RT ( ) aF ( ) d ds .

t0

t0

The proof is nalized.

REMARKS
In the situation when the acceleration
of the non-inertial frame {F} is null, aF = 0, since

|R u| = |u| , () u V3R , it follows that at any moment of time the mass center is
situated on a variable sphere that has an increasing radius:




(36)
R (t) = r0C + v0C + 0 r0C (t t0 ) , t t0 .
If v0C + 0 r0C = 0, then the motion of the center of mass takes place on a sphere with
constant radius r0C .
In case vector has a xed direction, or it has a regular precession, then the law of motion
of the mass center rC = rC (t) may be written explicitly, also by taking into account the
expressions (13) and (18) of the tensor map R .

1657

Figure 3. If aF = 0, the motion of the center of mass takes place on a ruled (conical)
surface, generated by a straight line which rotates with angular velocity .

If aF = 0, then the hodograph of the vector map that models the motion of the mass center is
a curve that is situated on a ruled surface generated by the rotation with angular velocity
of the straight line:


r = r0C + v0C + 0 r0C (t t0 ) , t t0 .
(37)
In this situation, the motion of the mass center may be decomposed into two:
a rectilinear uniform motion with velocity v0C + 0 r0C ;
a rotation with angular velocity of the straight line where the rectilinear motion
takes place.
The trajectory generated by these two independent motions is situated on a conical surface
(surface generated by a straight line with a xed point, see Figure 3).
The Mathematical Model of The Relative Motion
The motion of particle P1 (m1 ) related to particle P2 (m2 ) , referred to the non-inertial frame
{F} , is modeled by the vector valued function:
r = r2 r1 ,

(38)

where r1 , r2 are the solutions to IVPs (1) and (2). After simple manipulations, it follows that vector
r obeys the IVP:

F21

r + 2 r + ( r) + r =


,
(39)
0

0 r0 =

)
=
r
r
r
(t
0

2
1

r (t0 ) = v02 v01 = v0

where:

m1 m2
(40)
m1 + m2
denotes the reduced mass of the system. The present approach will refer to the situation where F21
is a central force, namely it may be written like:

m=

r,
F21 = f (r)

1658

(41)

where f : R+ R is a scalar continuous function of real positive variable and

r denotes the unit

vector associated to r, dened by:

r=r/r. It follows that the relative motion of P1 (m1 ) related to
P2 (m2 ) , referred to frame {F} , is modeled by the IVP:

f (r)

r,
r + 2 r + ( r) + r =

m
(42)


r
(t
)
=
r
0

r (t0 ) = v0

The Closed Form Solution of The Relative Motion

The IVP (42) models the motion of a particle having the mass dened in Equation (40); the
motion is related to the non-inertial reference frame {F}. The solution to IVP (42) is determined
within the next result (also see Ref. [12,13]).
Theorem 9 The solution to IVP (42) is obtained by applying the tensor operator R to the solution to the IVP:

f ()

m
,
(43)
 (t ) = r0

(t0 ) = v0 + 0 r0
where 0 = (t0 ) .
Proof. Eq (42) may be written using the previous considerations:
r =

f (r)

r.
m

(44)

Apply operator F to Equation (44) and perform the same change of variable dened in Equation

(33), =F r; then IVP (43) is obtained by simple manipulations, also by taking into account
Lemma 3. The proof is nalized.
Theorem 9 offers a simple way to solve the two body problem in non-inertial reference frames,
by following the steps: (i) solve IVP (43); (ii) apply R to the solution to IVP (43)to determine
the solution to IVP (42).
The First Integrals of The Relative Motion
The rst integrals associated to IVP (42) are determined. They are deduced with the help of the
tensor instrument previously introduced within the present work. The following denotation is made:



h0 = r0 v0 + 0 r0 .

1659

(45)

Theorem 10 The rst integrals of IVP (42) are:

r (r + r) = R h0 = h
(replica to the specic angular momentum conservation law);

m
mr2

+ m (, r, r ) + ( r)2 f (r) dr = constant = E

2
2

(46)

(47)

(replica to the specic energy conservation law).

Proof. By using operator ( ) previously introduced, it follows that:


r r = r r = 0

(48)

and by using Lemma 6 one may write:

r r = R

$

r r t=t0 = R h0 .

(49)

The existence of the rst integral dened in Equation (47) is proved by direct computations, by
differentiating with respect to the time variable t the quantity from the left hand side.
Remarks
1. The rst integral (46) shows that the hodograph of the vector map h dened in Equation
(46) is a spherical curve, which is drawn by the extremity of vector h0 = h (t0 ) while
precessing with the instantaneous angular velocity (see Figure 4, left). If vector has a

Figure 4. Left: The hodograph of h is a spherical curve; Right: If has a xed

direction, the hodograph of h is a circular section, and it sweeps the surface of a
circular cone.

xed direction, dened by the constant unit vector

, then the hodograph of h (t) is a circular
section, drawn by the extremity of vector h0 = h (t0 ) while precessing around a xed axis
dened by the unit vector
. Vector h sweeps the lateral surface of a right circular cone
with the instantaneous angular velocity (see Figure 4, right). As it follows from Equation
(13), its explicit expression is:
)
sin (t) (
0 ) cos (t) [
(
h0 )] ,
h = (h0

1660

(50)

where

t
(t) =

( ) d.

(51)

t0

2. The rst integral (47) has an energetic meaning. It emphasizes the existence of a generalized
potential energy function [14]:

m
2
(52)
V (t, r, r ) = m (, r, r ) + ( r) f (r) dr.
2
The rst integral (47) may be rewritten:


m 2

r + V t, r, r = constant.
2

(53)

If has a constant direction, then (t) = (t)

, with
constant. Since r (r + r) =
R h0 , by dot-multiplying this relation with , it follows:



(54)
, r, r + ( r)2 = R h0 = h0 (R )T = h0
(If has constant direction, then R = RT = ). Consequently,
( r)2 = h0 , and in correlation with (52) it follows that:
%
& 
1
V = m h0 ( r)2 f (r) dr.
2




, r, r +

(55)

In this situation, V = V (t, r).

If is a constant vector, from (55) it follows that V = V (r) (i.e. V represents the classical
specic potential energy).
The Laws of Motion in the Non-Inertial Reference Frame
From Equations (28) and (28) it follows that:

m2

r (t) ;
r (t) = rC (t)

1
m1 + m2

r2 (t) = rC (t) +

m1
r (t) .
m1 + m2

(56)

The vector function rC (t) is the solution to IVP (30) and it has the explicit expression given in
Equation (31). The vector function r (t) is the solution to IVP (42).
The First Integrals of the Two-Body System in The Non-Inertial Frame
The global dynamical characteristics of the system of two particles is studied herein. By introducing the analogues of the classic dynamical characteristics of a system of particles (linear momentum,
angular momentum, kinetic energy), time-varying quantities are described in place of the classic inertial conservation laws. By solving the differential equations derived within this Section, one may

1661

determine the expressions for the analogue quantities dened herein. However, a complete set of
rst integrals may be determined if the non-inertial frame {F} where the motion is referred to has
only a rotation motion, namely when aF = 0.
It is known that the classic (inertial) dynamical characteristics of a two-particle system are:
P=

2


mk r k

(57)

mk (rk r k )

(58)

k=1

(linear momentum);
K=

2

k=1

(angular momentum);

Ekin =

1
mk r 2k
2

(59)

k=1

(kinetic energy).
In a non-inertial reference frame, the conservation laws of the quantities dened above are not
valid anymore.
Dene:
H=

2


mk (rk + rk )

(60)

mk [rk (rk + rk )]

(61)

k=1

(generalized linear momentum);

L=

2

k=1

(generalized angular momentum);

2

T =

1
mk (rk + rk )2
2

(62)

k=1

(generalized kinetic energy).

It is natural to search for a link between generalized linear momentum, angular momentum and
energy dened in Equations (60) (62) and their classic counterparts expressed in Equations (57)
(59). One may write:

H = P + (m1 + m2 ) rC
L = K + I0
(63)

T = Ekin + VC
where the following notations were used:
I0 =
VC

2


mk
rk
rTk ;

k=1
'
2

k=1

( rk )2
mk (, rk , r k ) + mk
2

1662

(64)
(
(65)

(rC is dened in Equation (28), I0 is analogous with the the inertia tensor related to the non-inertial
frame).
The following result may be stated:
Theorem 11 The following afrmations hold:
+ H = (m1 + m2 ) aF ;
H

(66)

+ L = (m1 + m2 ) rC aF ;
L
%
&

d
T f (r) dr = H aF .
dt

(67)
(68)

Proof. Remark that the differential equations from the two IVPs (1,2) which model the motion of
the two particles with respect to {F} may be written compactly:
mk rk + mk aF = (1)k F21 , k = 1, 2.

(69)

Remark that the vector function H dened in Equation (60) may be rewritten like:
H=

2


mk rk .

(70)

k=1

By adding the two relations expressed in Equation (69), Equation (66) is obtained, also keeping in
mind the denition of the vector differential operator ( ) from Equation (20).
By cross multiplying with rk in Equation (69) and then adding the two relations, it follows:
+ L = L =
L

2


mk rk aF + (r2 r1 ) F21 = (m1 + m2 ) rC aF ,

(71)

k=1

where Equations (4) and (28) were taken into account.

Notice that:
T =

2


mk rk

k=1

d 
r
dt k


=

2


mk rk

k=1

d 
r + rk
dt k


=

2


mk rk rk .

(72)

k=1

By dot-multiplying with rk in Equation (69) and adding the obtained equalities, it follows that:
)
T =

2


*
mk rk


r,
aF + r2 r1 F21 = H aF + f (r) r

(73)

k=1

where Equations (38), (40) and (41) were taken into account. Now take into account that:




dr
d
r r d
=
f (r) dr .
r =f (r)
=
f (r) dr
f (r) r
dt
dt
r
dr
By replacing Equation (74) into (73), Equation (68) is obtained. The proof is nalized.

1663

(74)

Remark 12 Assuming that the acceleration aF of the non-inertial frame is known, from Equations
(66) (68), one may determine H, L, T by direct integration, by using Lemma 6.
Corollary 13 In the situation when the non-inertial reference frame has only a rotation motion, i.e.
aF = 0, then the following afrmations hold:
+H=0
H

(75)

(replica to the linear momentum conservation law);

+L=0
L

(76)

(replica to the angular momentum conservation law);


T f (r) dr = constant

(77)

(replica to the energy conservation law).

Corrolary 13 shows that in the situation when aF = 0, the vector functions H and L precess
with the angular velocity :
(
' 2

 0

0
0
(78)
H = R H0 = R
mk v k + r k ;

L = R L0 = R

k=1
2


mk r0k

v0k

r0k


.

(79)

k=1

The following result offers a new insight to the variation of the classical dynamical quantities
P, K,Ekin in the situation where aF = 0.
Corollary 14 The rst integrals of the two-particle system in a rotating reference frame {F} may
be expressed like:
|P+ (m1 + m2 ) rC | = constant;
(80)


2





mk [rk ( rk )] = constant;
(81)
K+


k=1
( 
'
2

( rk )2
Ekin +
f (r) dr = constant.
(82)
mk (, rk , r k ) + mk
2
k=1

If vector has a xed direction, then the vectorial rst integrals (75) and (76) have explicit
formulations:
H = (H0
) sin (t)
H0 cos (t)
(
H0 ) ;

(83)

) sin (t)
L0 cos (t)
(
L0 ) ,
L = (L0

(84)

where H0 = H (t0 ) , L0 = L (t0 ) . The hodographs of vectors H and L are circular sections. H
and L sweep the lateral surface of circular cones, with angular velocity (see Figure 5).

1664

Figure 5. If vector has a xed direction, vectors H, L sweep the lateral surface of circular cones.

The Motion Referred to the Mass Center Reference Frame

Consider the general situation where the translation acceleration of the non-inertial reference
frame {F} is non-null, aF = 0. When referred to a non-inertial reference frame {FC } originated
in the center of mass of the the system, having the same axes orientation like the original frame
{F} , the motion of P1 (m1 ), P2 (m2 ) may be described completely by the solutions to the IVPs
(30) and (42). Denote:
(85)
rk = rk rC , k = 1, 2.
Then from Equations (56) it follows that:
rk = (1)k

mk
r, k = 1, 2.
m1 + m2

(86)

Equations (86) describe the motion of particles P1 (m1 ), P2 (m2 ) related to the non-inertial reference frame of the mass center. The vector valued functions rk , k = 1, 2 obey the IVPs:

f (r)

r ,
r + 2 r k + ( rk ) + rk =

mk k
k
(87)


0 r0 ;

(t
)
=
r
r

0
C
k
k

r k (t0 ) = v0k v0C .

The dynamical global characteristics of the system in this reference frame are:
HC =

2


mk (rk + rk )

(88)

mk [rk (rk + rk )]

(89)

k=1

(generalized linear momentum);

LC =

2

k=1

(generalized angular momentum);

2

1
=
mk (rk + rk )2
2
k=1

1665

(90)

(generalized kinetic energy).

From Equations (88)(90), also by taking into account Equations (86), it follows that:
HC = 0;

(91)

LC =m [r (r + r)] ;

(92)

1
T C = m (r + r)2 ,
(93)
2
where m is the reduced mass of the system, dened in Equation (40), and r is the solution to the
IVP (42).
It follows that the vectorial rst integral LC obeys:
LC = R LC
0,

(94)

Equation (94) shows that the hodograph of vector L is a spherical curve. If has constant direction,
this hodograph is a circular section; vector L sweeps the surface of a circular cone, with angular
velocity . From (92) it follows:
r LC = 0,
(95)
therefore a geometrical visualization of the motion may be given: at any moment of time, the two
particles are situated in a variable plane that is normal on vector L. It also follows that the trajectories
of the two particles are curves which are homotetical to the center of mass C, proven by:
r1 =

m2
r .
m1 2

(96)

The homotety ratio is m2 /m1 .

The trajectories are plane curves if and only if vector LC has xed direction (also see Figure 6),
as follows from (95). The following result may be stated:
Lemma 15 The trajectories in the two-body problem in the mass center non-inertial reference
frame {FC } are planar if and only if the following conditions below are simultaneously satised:
(i) Vector has a xed direction, = (t)
, with (t) continuous and
constant unit vector;
C
C
(ii)
LC
0 = 0, where L0 = L (t0 ) .

Proof. If the trajectory is a plane curve contained in plane , it follows that vector LC is
constant, as it is orthogonal on plane and has a constant magnitude. It follows that R LC
0 =

d 

R LC
constant, therefore
R LC
0 = 0. It follows that R L0 = 0, so
0 = 0. Further:
dt
C
R L0 = 0. It implies:
LC = 0.
(97)
, Equation (97)
Since LC is constant, then has xed direction, that of LC . Since = (t)
=
0.
implies u LC
0
If conditions (i) and (ii) are satised, then:
From (i), it follows: R = and (R )1 = R .

1666

$
#
1 C
C
From (ii), it follows: 0 = u LC
L = (R u)
0 = (R u) L0 = (R u) (R )






R LC . It follows that: 0 = (R u) R LC = R u LC , and consequently u LC =
0. Then vector LC has a constant direction. Since it also has a constant magnitude, it is a constant
vector, therefore the trajectory is a planar curve. The proof is nalized.

Figure 6. The trajectories are planar homothetical curves, situated in a variable

plane, orthogonal on vector LC .

From the assumption (3), it follows that a scalar rst integral with energetic meaning may be
deduced:

T C f (r) dr = constant.
(98)
Equation (98) may be rewritten like:
m
m 2
r + m (, r, r ) + ( r)2
2
2


f (r) dr = constant.

(99)

Dene:

m 2
r ,
(100)
2
which represents the kinetic energy of the system related to the non-inertial frame {FC } , and denote:

m
V (t, r, r ) = m (, r, r ) + ( r)2 f (r) dr
(101)
2
the generalized potential energy. Then the conservation law (99) may be written like:
Ekin =

Ekin + V (t, r, r ) = constant.

(102)

Equation (102) shows that in the two-body problem referred to the rotating non-inertial reference
frame of the mass center {FC }, there exists the rst integral (102), with the generalized potential
energy introduced in Equation (101).
If the angular velocity has constant direction, Equation (102) becomes:
Ekin + V (t, r) = constant,

1667

(103)

& 
%
1
2
f (r) dr,
(

r)
V (t, r) = m LC

0
2
and if is constant:
m
Ekin + V (r) = constant, V (r) = ( r)2
2

(104)


f (r) dr.

(105)

CONCLUSIONS
The study of the two-body problem in an arbitrary non-inertial reference frames was approached
comprehensively, and its closed form solution was determined. The motion of the center of mass,
the relative motion of one body with respect to a frame attached to the other one, the relative motion with respect to the center of mass and the motion of the two bodies in the non-inertial frame
were approached by using the same tensor instrument, together with a vector differentiation rule
adequately dened. The results were presented in vectorial coordinate-free expressions. First integrals, equivalent to the classical dynamical characteristics of the motion (linear momentum, angular
momentum, total energy), for each body, as well as for the two-body system, were determined. The
present approach generalizes the classic two body-problem. The interaction force between the two
bodies was considered to be a general central positional force, and the reference frame to which all
the quantities were referred to was considered to have both rotational and translational motion, both
assumed to be known a priori. Future works will study the same context for particular forms of the
interaction force (including the classic gravitational case), as well as several particular situations of
other central forces. Also, the system of full bodies will be studied, by incorporating the internal
torques which might act upon the considered celestial objects.
REFERENCES
[1] T. Levi Civita and U. Amaldi, Lezioni di Mecanica Razionale. Nicola Zanichelli (Ed.), 19221926.
[2] H. Goldstein and C. P. Poole, Classical Mechanics. Addison Wesley, 2001.
[3] V. I. Arnold, V. Kozlov, and A. Neishtadt, Mathematical Aspects of Classical and Celestial Mechanics.
Springer, 3rd ed., 2006.
[4] J. E. Marsden and T. S. Ratiu, Introduction to Mechanics and Symmetry. Springer, 2nd ed., 1999.
[5] D. Condurache, New Symbolic Procedures in The Study of the Dynamical Systems. Ph.D. Thesis, Technical University Gheorghe Asachi, Iasi, Romania, 1995.
[6] G. Darboux, Lecons sur la theorie generale des surfaces et les applications geometriques du calcul
innitesimal, Vol. 1. Gauthier-Villars, Paris, 1887.
[7] A. Fasano and S. Marmi, Analytical Mechanics: An Introduction. Oxford University Press, 2006.
[8] D. Condurache and V. Martinusi, A Quaternionic Exact Solution to The Relative Orbital Motion Problem, Journal of Guidance, Control and Dynamics, Vol. 30, No. 1, 2010, pp. 201213.
[9] J. Angeles, Fundamentals of Robotic Mechanical Systems. Springer, 2nd ed., 2002.
[10] V. Martinusi, Lagrangian and Hamiltonian Formulations in Relative Orbital Dynamics. Applications
to Spacecraft Formation Flying and Satellite Constellations. Iasi, Romania: Ph.D. Thesis, Technical
University Gheorghe Asachi, January 2010.
[11] D. Condurache and V. Martinusi, A Tensorial Explicit Solution to Darboux Equation, The 2nd International Conference Advanced Concepts in Mechanical Engineering, Technical University Gheorghe
Asachi, Iasi, Romania, June 2006.
[12] D. Condurache and V. Martinusi, Relative Spacecraft Motion in A Central Force Field, Journal of
Guidance, Control, and Dynamics, Vol. 30, No. 3, 2007, pp. 873876.
[13] D. Condurache and V. Martinusi, Exact solution to the relative orbital motion in a central force eld,
2nd IEEE/AIAA International Symposium on Systems and Control in Aerospace and Astronautics, Shenzen, China, December 2008.
[14] D. Condurache and V. Martinusi, Foucault Pendulum-like problems: A tensorial approach, International Journal of Non Linear Mechanics, Vol. 43, 2008, pp. 743760.

1668