You are on page 1of 6

The Helmholz Conditions and Field Equations

Summary
The Helmholz conditions determine when a dynamics has a formulation by a Lagrangian. There is a
version for mechanics, as well as a version for field theory; the latter being more general. The field-
theoretic version of the Helmholz conditions is derived here.

Mark Hopkins

1. Background
An important subclass of field equations are those derivable from an action
1

( ) ( ) ( ) ,
x
S q x q x dx =

L (1.1)
through a Lagrangian density ( ) , q v L , with the corresponding Euler-Lagrange equations yielding the
system

, .
a a
a a
v q
q v
| |

= = |
|

\
L L
(1.2)

The geometry of this system is as follows. The arena in which the field is housed is an N -dimensional
manifold M locally coordinatized by
( )

: : 0, , 1 M x N = . The field configuration ( )


x
q x Q , for
each x M lies in a fibre of the bundle Q M , locally with adapted coordinates
( )

: ,
a
Q x q , the field
being defined as a section
( ) ( ) , x M x q x Q (1.3)
of this bundle. The velocities reside in the first jet
1
J Q of Q , endowed with coordinates
( )
1

: , ,
a a
J Q x q v
adapted to the fibration
1
J Q Q M . Corresponding to each section ( ) ( ) ( ) , s x x q x = is the first jet
extension ( ) ( ) ( ) ( )
1
, , j s x x q x dq x = , given by
2

.
a a a
v v dx dq = = (1.4)

The action integral is then defined as a functional of the integration region and section
( ) ( ) ( )
1
, , S s j s x dx

L (1.5)
with the requirement that the section ( ) ( ) : , s x x q x representing the dynamics of the field have a
variation ( ) ( )
, S s S q

= be a function solely of the variation on the boundary.



If we define the canonical momentum and force, respectively, by

, ,
a a a a
v q



L L
P F (1.6)
then we may express the Euler-Lagrange equations as the system


, ,
a a
a a
v q = = F P (1.7)
combined with the requirement that the canonical 1-form
( ) ( )
* 1 * 1
V J Q T J Q

1
See equation (1.12) for the definition of dx .
2
Remark on notation: In the following, the exterior derivative over the spaces of field configurations will be denoted by , while
the exterior derivative over the underlying manifold M is denoted by d . The wedge product and tensor will be denoted by simple
juxtaposition, thus for instance
a b b a
q v v q

= , dx dx dx dx

= , but
a a
q dx dx q

= . In here, and the following, the
summation convention is used whereby an index occurring once in the upstairs and downstairs position within a monomial is
understood to mark the place of the index of the summation representing the corresponding contraction; e.g.
a
a
f q means
a
a
a
f q .

a a
a a
q v + F P (1.8)
be exact, yielding the integral = L .

Under the field equations that result we may write

( ) ( ) ( )


,
a a a a a
a a a a a
q v q q q + = + = F P P P P (1.9)
so that the N -form associated with variations of the action over a spacetime region
S dx

L (1.10)
is related to the corresponding ( ) 1 N -form

a
a
S q dx

P (1.11)
associated with the integral over the boundary , where the following notation is used

0 1

, .
N
dx dx dx dx dx

(1.12)

The last integral identity captures the principle that the physics that goes on inside a region may be
characterized by a process analogous to tomography from the state on the boundary , particularly as
represented through the ( ) 1 N -form

a
a a
p q dx P . The field equations, themselves, are then
characterized as the key enabler of this process, playing the role of effectively eliminating a dimension of
redundancy.

The question being posed here is: what conditions may be posed that assure that such a Lagrangian
formulation underlie a system of field equations, when the equations are already given at the outset? The
equations are assumed to be a system of the form


,
a a b
ab a
v q v = = M Q (1.13)
with a set of coefficients of inertia

ab ba
= M M comprising a symmetric matrix and generalized force
a
Q
both being functionals of ( ) , q v as well as the coordinates, and both given at the outset. So, what were
asking is: what conditions need to be imposed in the coefficients of inertia and generalized force to yield a
Lagrangian system?

2. Helmholz Conditions Necessity
Since these equations are to apply through the functional dependence of the quantities ( ) , q v on the
coordinates, we will distinguish the total partial derivatives

, ,
a b a
a b a
D v a d v
q v q

+ + +

(2.1)
where

b b
a v ,

a a
v q = . Part of our task, therefore, will be to eliminate the explicit reference to the
acceleration terms a in favor of the (more restricted) set of terms comprising the generalized force,

b
a ab
a = Q M . A Lagrangian system is then characterized by the relations


, , , .
a a
a a a a a a
D v q
q v

= = = =

L L
F P F P (2.2)
The integrability conditions for ( ) , F P , are straightforward to pose



, , .
a b a b a b
b a b a b a
q q q v v v

= = =

F F P F P P
(2.3)
The coefficients of inertia enter at this point through the definition

,
b
ab a
v

P
M (2.4)
which in virtue of the last relation in (2.3) is symmetric. Combined with its integrability condition, the
result is the first two of the Helmholz conditions




, .
ab ac
ab ba c b
v v

= =

M M
M M (2.5)

The acceleration may be eliminated in favor of the generalized force by explicitly working out the force
law in the following form,

( )

.
b b a
a a a ab a b
d D d a a
v

= = = =

P
F P P M Q (2.6)
The remaining integrability conditions in (2.3) are therefore rewritten explicitly local coordinate form in
terms of the operator

d and generalized force


a
Q as




, .
b a a b
a b b b a a b b a a
d d d
q q q q q v v

= = +

Q Q P Q
P P P (2.7)

At this point, we need the following commutators

, 0, , .
a b b
d d
q v q
( (
= =
( (


(2.8)
Applying them to (2.7), we obtain


, ,
a b b
ab ab ab b a a
d d
q q v

= =

Q Q Q
M (2.9)
where

.
b a
ab a b
q q



P P
(2.10)
The last of (2.9) can be split into parts symmetric and anti-symmetric in the field indices, yielding

( ) ( )



1 1
, .
2 2
a b b a
ab ba ab ab ba b a a b
d d
v v v v
| | | |

= + + = + | |
| |

\ \
Q Q Q Q
M M M M (2.11)
Finally, the explicit reference to is eliminated by substituting the first of (2.11) into the first of (2.9),
resulting in



, 2 .
b a b a
ab a b a b
d d d
v v q q
| |
| |
( + = |
|

|

\
\
Q Q Q Q
M (2.12)

Noting that

, 0 d d ( =

, the result combined with the other conditions is the following
Helmholz Conditions

( )


, ,
, 2 .
b a
ab ba ab ba a b
ab ac b a b a
c b a b a b
d
v v
d
v v v v q q
| |

= + = + |
|

\
| |
| |
= = |
|
|

\
\
Q Q
M M M M
M M Q Q Q Q
(2.13)

3. Helmholz Conditions Sufficiency
In the following, we will assume that the underlying spaces involved are simply connected.

To verify the sufficiency of these conditions, we should be able to derive from this (locally) the existence
of a Lagrangian formulation by judicious use of the Poincar Lemma. Already, from the two conditions of
(2.13) involving the coefficients of inertia, we obtain the integrals



, .
b
ab a a a
v v

= =

P T
M P (3.1)
Substituting this into the first of (2.13) that involves the generalized force, we obtain


.
b a b a
a b a b
d
v v v v
| | | |

+ = + | |
| |

\ \
P P Q Q
(3.2)
We make use of the commutators (2.8) to rewrite this as





,
b a b a
b a a b a b a b b a a b
d d
v v v v q q v q v q
| | | |
+ + + = + = +
| |

\ \
Q Q P P T T
P P (3.3)
resulting in the equation


0 .
b a
a a a a b a
d
v v q
| |
+ = +
|

\
E E T
E Q P (3.4)

Substituting into the remaining of the conditions (2.13), we obtain for the left-hand side




2 2




b a
b b a a a b a b b a
b a
a b a b b a a b b a
b a a b a b a b
a b b a b a b
d d
v v v q v q
d d
v v v v v q v q
d
v v v v q q q
| | | |
= + +
| |

\ \

= + +

| |

= + + + |
|

\
Q Q T T
E P E P
E E T T
P P
E E P P P P P P
( )

2 .
a
b a b
ba ab a b a
q
d
v q q

| |

= + + |
|

\
E P P
M M
(3.5)
For the right-hand side, we obtain

( ) ( )




.
b a
b b a a a b a b b a
b a
a b a b b a
b a a b
a b b a
d d
q q q q q q
d d
q q q q
d
q q q q
| | | |
= + +
| |

\ \

= +

| |
= +
|

\
Q Q T T
E P E P
E E
P P
E E P P
(3.6)
Combining the results (3.5) and (3.6) into the last Helmholz condition (2.13), we obtain



1
, .
2
b a b b
ba a b a a
d d d d
q q v v
| | | |

( = + = | |

| |

\ \
E E E E
M (3.7)

The first condition on E , (3.4), results in the equation

2 2

0.
a a
b c c b
v v v v

+ =

E E
(3.8)
This implies that E is a polynomial in v , at most, of order N

1
1 1
1

, , , ,
,
2! 3! !
N
N N
N
b b b c b c d
b
a a b a bc a bcd a b b a
v v v v v v v
v
N
= + + + + +

E B B B B B (3.9)
with the coefficients B functions of q and the coordinates alone. Without loss of generality, it may be
assumed that they are symmetric in the first sets of index pairs,


, , , , ,
, , .
bc a cb a bcd a cbd a dbc a
= = = B B B B B (3.10)
The condition (3.5) implies that the coefficients are anti-symmetric with respect to the last field index


, , , , , ,
, , ,
b a a b bc a ac b bcd a acd b
= = = B B B B B B (3.11)
Substituting into the last condition for E , (3.7), we obtain

( ) ( )

, , , ,


, , ,
2!
.
2!
c d
c b c a cd b cd a c b a
a b a b a b
c d
c
a b ac b acd b
v v
v
q q q q q q
v v
d d v d
| | | |
+ + +
| |
| |

\ \
= + + +

B B B B B B
B B B
(3.12)
Noting the explicit dependence of

d on

a
v

( ) ( )

, , , ,


, ,
, , ,
2!
,
2!
c d
c b c a cd b cd a c b a
a b a b a b
c d
a b ca b c c c d
a b ac b acd b c d
v v
v
q q q q q q
v v
v v v v
q q
| | | |
+ + +
| |
| |

\ \
| | | |
= + + + + + + | |
| |

\ \


B B B B B B
B B
B B B
(3.13)
and separating out the coefficients of the different degrees (noting the symmetry of the coefficients in the
index pairs), we have the following system

,

, , ,
,

, , , ,
,
,
,
,
.
b a
a b a b
c b c a a b
ac b a b c
cd b cd a ca b ad b
acd b a b d c
q q
q q q
q q q q

=


= +


= + +

B B
B
B B B
B
B B B B
B
(3.14)

We also note that under the field equations, the expression for the generalized force yields the following
equation




,
b a
a a a a a b a a a
d a d D
q v q v q
| |

= + = + = |
|

\
P T T T T
E Q P P (3.15)
which is almost already in Lagrangian form. Our task is to use the system (3.14) to reduce the remainder,
involving E in like fashion.

Indeed, such a form may be quite readily arrived at by first writing the coefficients as differential forms
over the space spanned by the q s


, , ,
, , , ,
2! 3! 4!
b a b c a b c d a
a
a b a bc a bcd a
q q q q q q q q q
q

B B B B B B B B (3.16)
we may then write



, , , . = = = B B B B B B (3.17)
This is a sequence that terminates at order N
3
. So, by repeated application of the Poincar Lemma,
working from this highest order on back, we obtain the following integrals



, , , = + = + = + B A A B A A B A A (3.18)
involving differential forms


, , , , ,
2! 3!
a b a b c
a
a ab abc
q q q q q
q

A A A A A A A (3.19)
each set of coefficients being functions of q and the coordinates, alone, anti-symmetric separately with
respect to the field indices and space-time indices. Substituting this in the expression for E yields the result



.
2! 3! 2!
b c b c d b c
b b
a b bc bcd a ab abc a
v v v v v v v
v d v
q
| | | |

= + + + + | |
| |

\ \
E A A A A A A A (3.20)

3
The symmetry conditions on the original coefficients makes the corresponding differential forms anti-symmetric in their space-time
indices.
Noting that

d D = for functionals independent of v , we have, finally, the expression


.
2! 3!
b c b c d
b
a b bc bcd a a
v v v v v
D v
q v
| | | |

= + + | |
| |

\ \

L L
E L A A A A (3.21)
Thus, defining the Lagrangian by + L L T , we arrive at the following form for the field equations



0.
a a a a a a a
D D D
v q q v q v
| | | | | |

= = = | | |
| | |

\ \ \
T T L L L L
E (3.22)

Furthermore, though we wont work out the steps explicitly here, we will find that applying the two
processes described in the last two sections, one after the other, will result in the original Lagrangian being
recovered, up to a total divergence.

You might also like