[1]
Covariance and invariance in physics
A mathematical concept became a physical principle by Einstein. Einstein’s
talent was just this: to interpret physically the mathematical results.
George Mpantes
www.mpantes.gr
The history of mathematical transformations
it is connected in physics with the Newton’s principle of relativity:
Uniform rectilinear
mechanical experiments.
1
motion cannot be intrinsically determined by
It has it’s mathematical foundations
in the
transformation of the mechanical laws, under the Galilean transformations which
(Galilean transformations) are physical presuppositions for space and time of
Newtonian Universe.
From the mathematical point of view, this physical assumption refers
to the frames of reference of classical mechanics, and the facts of immediate
concern may be put as follows: it is assumed that a frame of reference (called
1
“The motion of bodies included in a given space are the same among themselves
, whether the space is at rest , or moves uniformly in a right line
motion” Newton Corollary V to the “Laws of motion”
without circular
[2]
by the physicists in inertial frame) exist in which Newton’s laws of classical
mechanics hold. Any frame in uniform rectilinear motion in relation to the
original system can be shown to be an inertial frame. We associate a rectangular
Cartesian coordinate system with each reference frame.. The close relationship
of this principle for classifying geometries, to the demand that physical laws
have invariant mathematical forms with respect to a given transformation group,
was pointed out by Klein. The orthogonal linear transformations accomplishes
the first unification of geometry and the Galilean transformations the first
unification of physics. (classical mechanics).
Abstract
The Newtonian mechanics is invariant under the simple Galilean
transformations. This invariance was the mathematical description of Galilean
relativity of motion. The electromagnetic theory of Maxwell (the two of the
four equations) failed in invariance by terms in frame velocity, under rectilinear
transformations, because of the partial derivatives with respect to t. This
invariance was achieved by the Hertz’s invariants equation of Maxwell, published
in 1892.
Finally physicists decided that electromagnetics
was covariant (not
invariant) under the Lorentz transformations of coordinates, and this covariance
was the mathematical description of Einstein’s principle of relativity. This
behavior of electromagnetic theory is due to the non-operationalist definition of
the concept of the electromagnetic field.
The transformations
Invariance of Newton’s mechanics in frame velocity
The covariance
Objectivity and covariance
The end of invariance
[3]
The covariance as a physical principle
The operationalist point of view
Sources
The transformations
…if the form of a law is not changed by certain coordinate transformations , that
is, if it is the same law in terms of either set of coordinates , we shall call that law
invariant or covariant with respect to the transformations considered..(P.G. Bergmann
(Introduction to the theory of relativity , Prentice Hall, New York 1946 p.10-11)
This verbal confusion about the terms
“invariance” and “covariance”
arises because both are forms of mathematical form preservation. But what is
their distinction in their physical implications ? This disjoining in the above
reference is referred
to the same physical reality or represent deeper
differences in the whole course of physics?
I shall
prove that these meanings, spring from mathematics,
and
particularly from the theory of transformations, but for physics, are related to
the operational school of definition of physical concepts.
…..the transformation idea has more than historical interest. It plays a major
role in the present day study of physical laws. In fact , the use of vector analysis as a
descriptive language for physical sciences is largely based on the invariant properties of
vector relations under certain types of transformations. (Introduction of vector and
tensor analysis Robert C. Wrede (Dover p. 35)
But what is transformed?
The history of mathematical transformations is connected in physics
with the Newton’s principle of relativity:
[4]
Uniform rectilinear
mechanical experiments.
2
motion cannot be intrinsically determined by
It has it’s mathematical foundations
in the
transformation of the mechanical laws, under the Galilean transformations which
(Galilean transformations) are physical presuppositions for space and time of
Newtonian Universe.
From the mathematical point of view, this physical assumption refers
to the frames of reference of classical mechanics, and the facts of immediate
concern may be put as follows: it is assumed that a frame of reference (called
by the physicists in inertial frame) exist in which Newton’s laws of classical
mechanics hold. Any frame in uniform rectilinear motion in relation to the
original system can be shown to be an inertial frame. We associate a rectangular
Cartesian coordinate system with each reference frame.. The close relationship
of this principle for classifying geometries, to the demand that physical laws
have invariant mathematical forms with respect to a given transformation group,
was pointed out by Klein. The orthogonal linear transformations accomplishes
the first unification of geometry and the Galilean transformations the first
unification of physics. (classical mechanics).
Invariance of Newton’s mechanics in frame velocity
Example 1. In mathematics we say that the laws of mechanics3 are
invariant under the simple Galilean transformations
r΄ =r-υt t΄=t …………..(1)
2
“The motion of bodies included in a given space are the same among themselves
, whether the space is at rest , or moves uniformly in a right line
without circular
motion” Newton Corollary V to the “Laws of motion”
I. every body tends to remain in a state of rest of uniform rectilinear motion unless compelled to
change it’s state by action of an impressed force.
II . The “rate of change of motion”, that is the change of momentum , is proportional to the
impressed force and occurs in the direction of the applied force.
III. To every action there is an equal and opposite reaction.
3
[5]
where we associate the coordinates r΄ and r with a particle P moving
uniformly and rectilinearly with respect to either system, the axes of the two
rectangular Cartesian coordinate systems are parallel (have the same orientation
in space) and the O’ system is in uniform translatory motion with respect to O.
Newton’s second law of motion is
d 2r
m 2 F .......... (2)
dt
and it is a measure of force. So we must to prove that force in relation
(2) is invariant.
Since m is invariant in Newtonian mechanics we have to prove the
invariance of acceleration.
Differentiating (1) twice we get the left-hand term
dr dr ΄
d 2r d 2r΄
,......... ... 2
dt
dt
dt
dt΄ 2
d 2r
d 2r
So m 2 (m 2 )΄ F F΄ viz
dt
dt
(2) becomes
d 2r
(m 2 )΄ F΄ .......... (2΄ ) in the primed system
dt
We see that the form of the equation is the same, and each term is
unaltered by the Galilean transformation. (by term we mean any additive
quantity). We call this equation Galilean invariant under this
transformation.
We say that the second law of motion (and hence all the three) are
Galilean invariant.
[6]
Example 2. the covariance
We shall study
now the transformation
of equation (2)
(now in the form
Fk
d
(m k )......( 2΄ )
dt
under coordinate axis rotation. The two systems are in uniform
rectilinear relative motion but have different orientation in space. It is assumed
as before that at t=0 the origins coincide. The path of the of motion of one
frame with respect to the other is illustrated in figure 1 and the complete
transformation is given
x c x t
i
i
j
j
i
x i
wit h c
.......... ..( 4)
x j
i
j
The transformation of (2) under (4) will result from the axis rotation,
as the translation does not vary the equation (2).
This rotation is given by the linear transformations
x1 c11
2 2
x c1
x3 c3
1
c12
c 22
c 23
c31 x 1
c32 x 2 .
c33 x 3
Multiplying (2΄) with
x k
and summating for κ we have
x r
x
x k
d
d
Fk (m k k ) Fr (m r )......... .......( 5)
x r
dt
x r
dt
Comparison (2΄) and (5) shows that the two terms of (2΄) undergo
similar “scrambling” of components in the barred system with the same
i
coefficients c j
[7]
For example this scrambling (linear combination of all components) is
written for one component
F
x1
x 2
x 3
F1
F2
F3 .......... ...( 6) .
x
x
x
This scrambling is a consequence of the vector character of the
Newtonian force and not of the fact that F satisfies the Newton equation of
motion.
So the individual terms of the Newtonian law of motion, transform under
an axis rotation to the same rule (of linear combination). The form of equation
(2΄) , is the same in (5), but it’s terms are altered by the
transformation, in the same way. We call this equation Galilean covariant.
The law’s form expressed in the new symbols is independent of the orientation
of the frame of reference.
The covariance above was a mathematical demonstration of the abilities
of transformations, as the rotations of axes had no motion, so it has no relation
with the Galilean relativity.
i
It might be added that if c j vary with time –so that a transformation to
rotating axes is described-, equation (2) is neither invariant nor covariant. The
physical explanation is that rotation induces extra “non-inertial” forces , so the
simplest form of motional relativity fails.
But such scrambling of components in (6) as the result of covariance we
get in the formula of transformation of the electric and magnetic three vectors
between two Galilean frames, under (now) the simple Lorentz transformation
E1 ' E1 ,
H1 ' H1 ,
E 2 ' ( E 2 H 3 / c),
H 2 ' ( H 2 E3 / c),
E 3 ' ( E3 H 2 / c).......
H 3 ' ( H 3 E 2 / c)........( 7)
It should be noted that the transformation given above is a consequence
of the tensor character of Frs (electromagnetic tensor) as t Lorentz
[8]
transformations, and depends in no way on the fact that Frs satisfies Maxwell’s
equations. (J. L.Synge: Relativity: the special theory p 317)
Objectivity and covariance
There is a well defined mathematical distinction that we find in vector
and tensor entities. They are composed of components which transform in a
change of the coordinate system, they co-vary with the system, but the laws of
transformation are such that if all the components are zero in the initial system
, they will be zero in every system. So a tensor equation holds in every system if
it holds in one.
This consideration expresses a basic idea of modern science. The system of the
tensor components is covariant , i.e it has a different numerical composition for each
coordinate system. Yet we express in this fashion a state that it is independent of the
coordinate system, i.e.
an invariant state. The tensor as a whole is an invariant
magnitude. We can recognize this property from its representation by means of
components , since the components can be calculated for every coordinate system , if
they are known for one.. it is unfortunate that the physical terminology does not reflect
this well-defined mathematical distinction.
(the philosophy of space and time p.236
Hans Reichenbach)
We saw that in systems with different orientations in space, we have a
covariant form of the second law of Newton, where the description of the
terms, can be represented differently in different coordinate systems.
……Each of these descriptions presents the objective state in a particular way.
The totality of these descriptions , however, defines an invariant situation , so to speak,
namely,
whereas one description gives only one component of the situation, it’s
projection on a particular coordinate system. Among these components there is no
difference with regard in truth. (Reichenbach p.241)
Here Reichenbach denotes that an objective state, here the Newtonian
laws, is the totality of the covariant states, as we have seen that
[9]
mathematically, -in linear transformations4-, the covariant states are produced
from the various orientations of the systems of coordinates, and of course,
the orientations in space are infinite. But
We can visualize this as analogous to the geometrical meanings of
equality and similarity. Invariance is as equality, as the scheme of the idol
(image) of an object in a flat mirror, and covariance is as similarity, as the
geometrical similar scheme of this idol. We can study the properties of the
object in either of the two schemes, the similarity does not affect the causal
chains of geometry of the schemes, as covariance does not affect the
Newtonian laws of physics in various systems of coordinates.
These as for philosophy but in physics there are different opinions for
invariance and covariance from Hamilton’s time till today, in Thomas Phipps book
(HERETICAL VERITIES Thomas E. Phipps p.120)
…Hamilton had an opportunity to make the opposite realization; namely that the
covariance of the Newtonian form of the equations of mechanics (under rotations as we
demonstrated)was not an asset but a clue to imperfection of the form. Working from
this or other clues , Hamilton produced “canonical” forms of the mechanical equations
that were genuinely invariant under a huge group of (so called “contact” )
transformations – a group containing coordinate transformations , rotations etc. as small
subgroups. ..in consequence physics flowered into canonical mechanics , the Hamilton Jacobi equation
and (a short step onward) the Schroedinger equation and quantum
mechanics……
Beyond the Hamilton’s canonical forms of equations of mechanics, there
is a basic denial of symmetry in the edifice of physics. Maxwell’s equations hold
in it’s form, only in the inertia frame of the observer- for his era in the frame
of the immovable ether – and nowhere else. In all others inertial frames their
terms alter , because alter the transformations , which now express different
4
Einstein claimed that the coordinate transformations transformations (between two Lorentz
charts with a common origin) must be linear “on account of the properties of homogeneity which
we attribute to space and time.
[10]
ideas for space and time , than Newton’s. But the invariance was rejected from
physics. It is like Lorentz’s transformations to impose a rotation of axes in
inertial frames and this is represented in four-dimensional systems in
Minkowski’s space-time.
The covariance as a physical principle. The end of invariance
Special relativity interpreted for physics, exactly these mathematical
results. It took almost three centuries of development of physics, before the
principle of relativity was extended by Albert Einstein to the case of optical
and electromagnetic phenomena , as observed in a closed , uniformly moving
cabin.
It was Einstein’s principle of relativity: Uniform rectilinear motion cannot
be intrinsically determined by any experiment, mechanical or electromagnetical or
optical . But Maxwell’s equations were not invariant as were Newton’s, under the
Galilean transformations. Now the equations of physics ought to be Lorentz
covariant because this occurred for the equations of Maxwell. Newton’s
equations had to change because this covariance was now the physical state of
the equations.
Why was impossible a Galilean invariance of the Maxwell’s equations, as
with Newtonian equations of mechanics? because the partial derivative with
respect to t, does not transform invariantly under
such a
rectilinear
transformation.
We see that applying (1)and the chain rule of differentiation, we have
x' y ' z ' t '
.
t t x' t y ' t z ' t t '
( x
y
z
)
t΄
x΄
y΄
z΄
Or
.΄ .......... .......... (8)
t t΄
[11]
So partial derivative with respect to t, as two of the four Maxwell
equations, fail of invariance by terms of first order in frame velocity. So there
is not Galilean invariance for Maxwell equations. The non-invariance of Maxwell’s
equations under Galilean transformation, was interpreted (a) that an ether
existed for electromagnetism (though not for mechanics) (b) that this ether
existed in a physically detectable state of motion.
The invariance now became covariance, under a new system of
transformations, the Lorentz transformations.
The operationalist view of point
What if, we replaced partial time derivatives, whenever they appeared in
Maxwell’s equations , with total time derivatives? These were Hertz’s equations,
which become rigorously invariant under Galilean transformation, just as
Newton’s equations are. But we had to change our views for the field.
What physically is a field?
We read in Thomas Phipps (heretical verities p.109)
Definition: A field is what is “measured” by (i.e produces quantifiable
response in) a field detector.
Definition: A “field detector is any object
that causes the field to
manifest its presence locally.
Phipps claims that this definition is better than: “fields are mathematical
vectors” (taught at a prominent east coast U.S. university)
How Hertz’s equations rendered Galilean invariant? He replaced partial
time derivatives with total time derivatives, where the total time derivative
inserts (8) a term proportional to some kind of velocity -the velocity υd of the
detector in the new-Hertzian interpretation-,
and the Galilean velocity
transformation generates a new term which cancels the velocity υd, that spoils
the Galilean invariance of Maxwell’s equations. So the operationalist definition
[12]
of the field abolishes the necessity for the Lorentz transformations for the
covariance of Maxwell’s equations, giving a new direction in the whole of physics.
But
as we Know today the Galilean invariance was converted
to
Lorentz covariance and the proper term for both is “invariance in form”.
So we have the final formula
……Any relation between physical quantities must be expressed by means of
form-invariant or covariant equations C.Moller (the theory of relativity Oxford 1972
p.96)
Sources:
HERETICAL VERITIES: MATHEMATICAL THEMES IN PHYSICAL
DESCRIPTION (Thomas E .Phipps, Jr, classic non-fiction library, Urbana)
Relativity: the special theory: (J.L.Synge North-Holland publishing
company Amsterdam New York Oxford)
Introduction to vector and tensor Analysis: (Robert C.Wrede,
Dover)
Relativity and geometry : (Roberto Torretti, Dover)
The philosophy of space and time : (Hans Reichenbach, Dover)
Space, time , matter: (Hermann Weyl, Dover)
www. mpantes.gr
END
[13]