Dhaka Univ. J. Sci.

61(1): 87-92, 2013 (January)

Serret-Frenet Equations in Minkowski Space

Md. Showkat Ali1 and Md. Abu Hanif Sarkar2
1
Department of Mathematics, University of Dhaka, Dhaka- 1000, Bangladesh
2
Department of Mathematics, Jagannath University, Dhaka 1100, Bangladesh
E-mail: msa317@yahoo.com
Received on 28.03.2012. Accepted for Published on 04.11.2012
Abstract
In this paper we established the Serret-Frenet equations in Minkowski space. These equations originally formulated in Euclidean space in
, constitute a beautiful set of vector differential equations which contains all intrinsic properties of parameterized curve. From the local
theory of curves in states that a curve lies in a plane if and only if its torsion vanishes, which gives us clear geometrical insight in the
notion of torsion. This theorem has two counterparts in Minkowski space that has been focused.

I. Introduction
Or, in matrix representation,
The main objective of this paper is to look at some aspects
of the differential geometry of curves in Minkowski space
[2]. The article is organized as follows. In section 2 we start
by setting up the Serret-Frenet equations in Minkowski =
space. The Serret-Frenet equation gives the derivatives with
respect to the arc length parameter of the tangent, normal
and binormal vectors of a curve in terms of each other.
Though the Serret-Frenet equations, the evolution of a curve where , , denote, respectively, the tangent, normal and
is completely determined, up to rigid motion, by two binormal vectors of the curve , which is assumed to be
intrinsic scalars: the curvature and the torsion . smooth (at least of class ). The triad of vectors
This result is known as the fundamental theorem of space constitute an orthonormal right handed frame defined at
curves [10]. An analogous theorem holds in Minkowski each point of and the invariant scalars and
space, and proof is given in section 3. The set of equations
are called, respectively, the curvature and
has a solution in terms of , curvature and  , the angle of
rotation of the osculating plane, that indirectly solves the torsion of . The equations follows directly from the
Frenet-Serret equations, with a unique value of  for each definition of the normal vector , the binormal vector
specified value of , torsion. Explicit solutions can be [10].
generated for constant  .The equations breakdown when To adapt the above formalism to Minkowski space we need
the tangent vector aligns to one of the unit coordinate to replace the Euclidean metric for the Minkowski metric
vectors, requiring a reorientation of the local coordinate and define a second
system
binormal and a second torsion . Since usual
II. The Serret-Frenet Equations
vector products make no sense in four dimensional space we
In Euclidean space the intrinsic geometric properties of define our set of orthonormal four-vectors ( ) by
a curve (parameterized by the arc length ) are described concomitantly requiring them to satisfy a four dimensional
by the Serret-Frenet equations extension of the Frenet-Serret equations, which then governs
the evolution of the tetrad. It is also convenient to restrict
ourselves to timelike curves , i.e. those for
which , where now denotes the arc
length parameter in the sense of Minkowski metric .
(1)
Accordingly, if we denote the tetrad vectors by
, then the orthonormality conditions read
It can be shown
88 Md. Showkat Ali and Md. Abu Hanif Sarkar

that if we chose , i.e. being the components same as in the case of , requiring only minor
modifications. The proof of the uniqueness part, however,
of the unit tangent vector, then we can easily construct an
differs from its counterpart in , since the latter makes use
orthonormal basis vectors , defined along the curve,
of the positiveness of the Euclidean metric. Let us first state
which obey the following four dimensional Serret-Frenet the theorem.
equations, given in matrix representation by
Theorem 1. Given differentiable functions
and , there exists a regular
parameterized timelike curve such that is the
curvature, and are, respectively, the first and
second torsion of . Any other curve satisfying the same
conditions differs from by a Poincare transformation. i.e.,
by a transformation of the type , where
Of course the above procedure may be easily generalized to represents a proper Lorentz matrix and is a constant
-dimensional Riemannian (or pseudo-Riemannian) spaces four vector.
by changing from ordinary differentiation to absolute
differentiation [4, 8]. Here, two points are worth Proof. Let us assume that two time-like curves and
mentioning. First, due to the Lorentzian signature, the satisfy the conditions
matrix that governs the evolution of the tetrad vectors and , , where is an open interval of .
is not anti-symmetric, as in the case of Euclidean signature. Let and be the Serret-Frenet tetrads
Secondly, in order to construct of the tetrad it is not at of and , respectively. It is clear that it is
necessary, as a matter of fact, to assume that the curvature always possible, by a Poincare transformation, to bring
and the torsions have non-zero values. If , then the of into of in such a way
curve is a timelike geodesic and a triad of constant spacelike that . Now, the two Serret-Frenet
orthonormal vectors orthogonal to
tetrads , satisfy the equations
may be chosen. In this case, and are
zero, if , but , then we can choose an
orthonormal basis in such a way
that and are constant spacelike vectors.

III. The Fundamental Theorem in Minkowski Space

A most important result in the local theory of curves in
Euclidean space , known as the fundamental theorem of
and
curves, states the following:
Given differentiable functions and , there
exists a regular parameterized curve such that is the
curvature, is the torsion of . Any other curve
satisfying the same conditions differs from by a rigid
motion . It would be natural to expect this theorem to
hold when appropriately transposed to Minkowski space. In
this case, a rigid motion would correspond to a Poincare which can be written in a more compact form as
transformation and the curve would be expected to be
determined by the three differentiable functions (6)
and . Here we give a simple proof of
the fundamental theorem of curves in Minkowski space. We
omit the proof of the existence part since it is almost the
Serret-Frenet Equations In Minkowski Space 89

with denoting the elements of the Serret-Frenet . From we

matrix. Clearly, the two tetrads , are
have . Given that we conclude that
related by an equation of the type
has no components in the -direction, i.e.,
(7)
. Thus
with the elements of the matrix satisfying the
condition , since we are assuming that hence from the
. From (6) and (7) we obtain a system equation we conclude
of first order differential equations for the elements of that . If , then also must vanish, for in
given by
this case is chosen to be constant. If , then
, hence . From the

By assumption, are differentiable functions of the equation and

proper parameter . From the theory of ordinary differential
the third Serret-Frenet equation
equations [11], we know that if we are given a set of initial
conditions , then the above system admits a unique we are led to conclude that . However,
solution defined in a open interval cannot be zero, otherwise the set of vectors
containing . On the other hand, it is easily seen that would not be linearly independent.
is a solution of . Therefore, we conclude Therefore, must vanish.
that . Let us turn to sufficient condition. Suppose that .
Other extensions of known theorems of the differential Then, the fourth Serret-Frenet equation implies that is a
geometry of curves in Euclidean space can easily be constant vector. Let us conveniently choose our coordinate
carried over into Minkowski space. As an example, let us system in such a way that . Now, since is
consider the following results on curves lying in : orthogonal to we must have , which means
A curve, with non vanishing curvature, is plane if and only that lies in the hyperplane . This completes
if its torsion vanish identically Natural extensions of the proof.
this result to Minkowski space are given by the following
Another extension of theorem 2 leads directly to the
theorem:
following proposition:
Theorem A time-like curve , with non vanishing
Proposition. A time-like curve , with non-vanishing
curvature, lies in a hyperplane if and only if the second
curvature, is plane if and only if first and second torsions
torsion vanishes identically.
vanish identically.
Proof. Again we restrict ourselves to time-like curves. Let
Since it follows the same lines of reasoning presented in the
us start with the necessary condition. Suppose the curve proof of theorem 2 that’s why we omit the proof.
lies in a hyperplane. Then, by a Lorentz rotation we can
align one of the coordinate axes with the normal direction to IV. Mathematical Development
the hyperplane. For the sake of the argument, let us assume A local coordinate system having the property
that we can bring to lie, say, in the - T  i  supports the definition of N :
hyperplane. Then the parametric equations of are of the
N  j  cos   k  sin  (9)
form . Let
denote the vectors of the canonical coordinate basis. The curvature k and the angle of rotation  of the plane
Thus, in these coordinates, (containing N and T ) characterize the curve. When the
and
90 Md. Showkat Ali and Md. Abu Hanif Sarkar

plane containing T and the global coordinate j is normal s

sin 1 T j  sin 1 T j 0    cosd ;
to k (figure 2) then s0

s
j j T j  sin sin 1 T j 0    cos d   sin  ;
 s0 
s s
T j  T j 0 cos  k cos  d  1  T j2 sin  k cos d
s0 s0
(16)
i
Equation (12) is solved by noting that
i 2 2
Tk  1  T  Ti  cos   Ti and 2 2
introducing
j

the variable  so that

Ti  cos  cos  ; (17)

k TK  cos  sin  ; (18)

k Substituting into (12):

dTi d
Fig. 2. Angular orientation of the local coordinate system   cos sin  cos   cos  sin 
with respect to the global coordinate system. ds ds
T  j  iTk  kTi   sin  cos  sin    cos  cos  cos  sin 
k   (10) 
T j cos 
1  T j2

Equation (10) breaks down when T   j , requiring an

  sin  sin    cos cos  sin  . (19)

alternate expression. However, when T   j , Equation (19) simplifies to

d  sin 
 iTiT j  j (1  T j2 )  kT j Tk  ; (20)
j  k   T  (11) ds cos 
1  T j2
Or,
Substituting (11) and (10) into (9): s  sin   T 
  0   d  cos 1  i ; (21)
1 dTi  Tk sin   Ti T j cos  s0 cos   cos  
Ni   ; (12)
 ds 1  T j2 so that
cos  s  sin  cos  s  sin 
1 dT j Ti  Ti 0 cos  d  Tk 0 cos  d ;
Nj   cos  1  T j2 ; (13) cos  0 s 0 cos  cos  0 s 0 cos 

 ds (22)
1 dTk Ti sin   T j Tk cos  where
Nk   ; (14)
 ds 1  T j2 s
cos   1  T j20 cos   cos d  T j 0 sin   cos d .
s

s0 s0
Equation (13) can be integrated directly: (23)

Tj dT j s The solution for Tk follows from (18) and (20):

    cos d ; (15)
Tj0 s0
1  T j2
Leading to
Serret-Frenet Equations In Minkowski Space 91

cos s  sin cos s  sin Bj Nj Bj

Tk  Tk0 cos d Ti 0 cos d;    sin 1  cos 1   tan 1 .
cos0 s0 cos cos0 s0 cos
1  T j2 1  T j2 Nj
(24)
(35)
It can be easily verified that (16), (22), and (24) meet the An explicit solution often results when  is constant.
requirement:
Setting Ti 0  1 , so that  0   0  0 (and setting s 0  0 )
dT leads to
 . (25)
ds s
  0    cos  0 d ; (36)
Generating an expression for the torsion  requires first
computing N by substituting (16) - (18) into (12) - (14):   2 tan  0 tanh 1 tan  / 2 ; (37)
N i   cos sin  cos   sin  sin  ; (26) so that
N j  cos cos  ; (27)
 
Ti  cos 2 tan0 tanh1 tan / 2 cos     cos0 d ;
s

0
N k   cos  sin  sin   cos  sin  ; (28) (38)
s
Next, B T N : T j  sin     cos  0 d ; (39)
0

Bi  sin  sin  cos   cos sin  ; (29)

B j   cos  sin  ; (30)

  s
Tk  sin 2 tan 0 tanh1 tan / 2 cos     cos 0 d ;
0
Bk  sin  sin  sin   cos  cos  ; (31) (40)
The torsion becomes
Equation (32) expresses the torsion as a function of : s
 s    s  sin  0 tan 0    cos 0 d (41)
dB d
    tan  sin  . (32) As an example, when
ds ds 2

Equation (33) expresses  in terms of components

   e s , (42)
0
of T and B :   
Ti  cos  0 erf s  cos  0  cos
d T j B j  2 
  . (33)
ds 1  T j2
     
Integrating (33) leads to the following expression for  :
 2 tan  0 tanh 1  tan  0 erf s  cos  0   ; (43)
   4   
 
s  T j B j    
   0      d . (34) T j  sin  0 erf s  cos  0 ; (44)
s0  1  T 
2
 j  2 
Equation (34) indicates a unique value of  for each   
specified value of  when T j  1 . Thus, (18), (22), and Tk  cos  0 erf s  cos  0  sin
(24) indirectly solve Serret-Frenet equations.  2 
     
The angle  can also be expressed in terms of components 2 tan  0 tanh 1  tan  0 erf s  cos  0   ; (45)
of T , N , B :    4 
    
92 Md. Showkat Ali and Md. Abu Hanif Sarkar

 0   V. Conclusion
2
 s    0 e  s sin  0 tan  erf s  cos  0  ( 46 ) Due to the fundamental theorem of curves, if a curve
 2  represents the motion of a particle, one can look at the
Serret-Frenet equations as containing complete information
on the dynamics of the particle. Such correspondence
When    0 but    0 , the solution will typically between the geometry of curves and the dynamics of
involve undetermined integrals. For example, when particle can be nicely explored in the context of special
   0 and    0 s , relativity to study the intrinsic geometry of world lines in
Minkowski space. In fact, compared to the Newtonian
s  0 sin  0 formalism, special relativity is a more natural setting for a
Ti  cossin  0 s  cos  d ; (47) description of motion through the Serret-Frenet equations,
0 cossin  0  since the worldliness of particles are usually parameterized
by the arc length parameter , turning to the equations into a
T j  sin sin  0 s ; (48) much simpler form.
.....................
s  0 sin  0
Tk  cossin  0 s  sin  d ; (49) 1. Hausrath, A.C. & Goriely, A. 2006. Protein Science 15, 753-760.
0 cossin  0 
2. A nice account of the geometry of Minkowski space may be
and found in G. L. Naber, The Geometry of Minkowski spacetime,
Dover, New York, 2003.
 s    0   0 tan sin  0 s sin  0 s. (50) 3. Divjak, B. 1997. Mathematical Communications 2, 143-147.
[3] J. L. Synge and A. Schild, Tensor Calculus, Dover, New
When    0 and    0 , (38)-(41) York, 1978, Ch.2
4. Lanczos, C., 1970. Space through the ages, Academic press,
Ti  cos 0 s cos 0  cos London and New York.
become:   0 s   5. Kreyszig, E. 1991. Differential Geometry, Dover, New York, Ch. 2.
1 
(51)
2 tan  0 tanh  tan  cos  0  ; 6. Arreaga-Garcia, G., Villegas-Brena, H. & Saucedo-Morales,
   2   J., 2004. J. Phys. A: Math. Gen. 37, 9419-9438.
7. Hasimoto, H. 1972. J. Fluid Mech. 51, 477-485.
T j  sin  0 s cos  0 ; (52)
8. Synge, J. L. and Schild, A., 1978. Tensor Calculus, Dover,
New York, Ch.2.
Tk  cos 0 s cos 0 sin
9. Nakayama, K., Segur, H. & Wadati, M. 1992. Phys. Rev. Lett.
  0 s   69, 2603-2606.
1 
(53)
2 tan  0 tanh  tan  cos  0  ; 10. Do Carmo M.P. 1976. Differential Geometry of Curves and
   2   Surfaces. Prentice-Hall, Englewood Cliff, NJ.
11. Lang, S. 1968. Analysis I, Addison-Wesley, Reading, Mass,
 s    0 sin  0 tan  0 s cos 0  (54) 383-386.
12. Kats, Y., Kessler, D. A. & Rabin, Y. 2002. Phys Rev. E 65,
When  0   / 2 , s   0 , confining T and N to a 020801(R).
plane. When T align with j,    in (54), and the
equations break down.