Why a symmetry should leave a first integral
An autonomous variational problem has a Beltrami first integral. A cyclic coordinate supplies a constant momentum. These statements look different, but both arise by comparing the action with a nearby description that leaves its local density unchanged. Noether's identity brings them into one calculation and also accommodates a density that changes by a total derivative.
The word “symmetry” needs precision. A transformation may preserve the differential equation without preserving the variational density. A transformation of the independent variable also changes the integration element, which contributes an essential Jacobian term. Finally, adding a total derivative changes the canonical momentum but can leave the correctly adjusted conserved quantity unchanged.
We prove an infinitesimal identity directly for first-order functionals of one independent variable. This gives a local version of Noether's first theorem along classical Euler–Lagrange curves. It is not a minimum test or an existence theorem for those curves. Velocity-dependent generators and symmetries depending on arbitrary functions are outside this point-symmetry treatment.
State variables, generators and the density criterion
Definition 16.1.1
(First-order variational data)
Let and be open. Assume , with state variable and velocity variable . A classical curve is a map into on an interval contained in . Write
for its momentum and Euler–Lagrange equation. Gradients are columns and pairings are Euclidean dot products. For a function define
For a vector function , this notation means . The operator in this definition acts on functions of position and state; derivatives of velocity-dependent quantities along a curve use the full chain rule, including .
Definition 16.1.2
(An infinitesimal variational symmetry with a boundary term)
Let , and be . Define
We impose the identity for every triple in the stated domain. Here is the infinitesimal change of independent variable, the change of state, and the infinitesimal boundary term. At the infinitesimal level, means the density variation is zero; adding a constant to does not affect the criterion.
The all-triple identity is our criterion for an infinitesimal point variational symmetry. The conservation calculation will also yield a weaker, curve-specific statement when the density residual vanishes only along a given extremal. We take the infinitesimal identity as a hypothesis and do not construct a finite family from it. The next calculation explains its origin when a smooth local family is already available.
Suppose smooth local point transformations equal at zero and have parameter derivatives . Along a curve, assume so the transformed graph has a valid increasing independent coordinate. Its velocity and the action density pulled back to the original coordinate are
At zero, the velocity derivative is . Differentiating the displayed density gives the first four terms in . If the pulled-back density differs from the original one by with , then its derivative supplies for . This explains both the Jacobian term and the boundary correction.
For a fixed compact curve segment, and smoothness ensure positive and values in for sufficiently small parameters. No global preservation of that domain or of the positive graph Jacobian is assumed. Such a transformation can move the endpoints; it need not preserve a particular fixed-endpoint comparison class. The conservation theorem is a local identity on Euler–Lagrange arcs, rather than a claim that every transformed curve is an admissible competitor for the same boundary data.
Common mistake
(Dropping the Jacobian term when the independent variable moves)
When is not constant, the term in is easy to forget, because it does not come from differentiating . For with , the dilation , has and . Without the criterion would read , which is not zero, and a genuine symmetry would be rejected. With the term, the residual is .
The identity and its conservation consequence
Theorem 16.1.3
(A Noether quantity is constant along an extremal)
For any curve in the stated domain define
evaluating the functions along the curve. Then the exact identity is
Consequently, if is the stated variational identity and the curve satisfies Euler–Lagrange, is constant. No invertibility of is needed for this implication.
Proof
(Differentiate the proposed quantity before imposing the equation)
Write . Along any classical curve,
Differentiate and substitute these expressions:
Add and subtract , then collect the terms into . This proves the identity without assuming the equation. The two residuals vanish under the theorem's hypotheses, giving conservation.
Proof X-ray
Two identities do different work
The all-triple density identity establishes a point variational symmetry. On an individual Euler–Lagrange curve, conservation already follows if the density residual vanishes along that curve; a full domain identity is stronger than needed for that conclusion. Merely mapping solutions to other solutions establishes neither density statement. The oscillator scaling below is an equation symmetry whose density residual does not vanish along a general solution.
The field is the fixed-coordinate vertical variation. Indeed, differentiating the transformed-graph relation at a fixed new coordinate subtracts the displacement caused by moving the independent coordinate. This is the same distinction between a total endpoint displacement and a fixed-coordinate variation that appears in transversality calculations.
Two familiar special cases now follow. If , take , , ; then is constant. Its negative is the usual energy expression. If is independent of state coordinate , take , , and obtain the conserved momentum . Neither consequence supplies a general converse from a first integral to the full Euler–Lagrange system.
Adding a total derivative
Theorem 16.1.4
(The corrected charge survives a boundary modification of the integrand)
Let and replace the integrand by . For the same generators define
If the original density criterion holds, the modified one holds with . The corresponding Noether quantity is exactly unchanged: . The Euler–Lagrange equations of the two integrands also agree.
The chain rule gives the useful identity
It proves the modified criterion. For the charge, and . The added terms in the charge therefore cancel against the added boundary term. Finally,
Thus the same classical curves solve both equations. The assumption on keeps the modified integrand and boundary data in the stated classes.
Concept lensAnalytic
Momentum depends on the density; the corrected quantity can remain the same
A total derivative changes the formula for the momentum by . The symmetry's boundary term changes as well. Keeping only the modified momentum would lose that compensation. The theorem compares both changes explicitly, explaining why different variational densities for the same equation can produce the same conserved quantity.
Worked examples
Worked example 16.1.5
(Rotations and a planar angular quantity)
Let , let and take
Choose , and . Then , and . Both and vanish, so the density criterion holds. The conserved quantity is
For , with , the curve solves Euler–Lagrange. Substitution gives . Time translation also gives . In this family, , with equality exactly when .
Worked example 16.1.6
(A boost changes the free density by a boundary term)
For the scalar free integrand , consider , . The transformed velocity is , and
Thus , and . The infinitesimal criterion reads , and the charge is . Euler–Lagrange gives , for which . Dropping the boundary term would give , whose derivative is generally the nonzero constant .
Worked example 16.1.7
(Translation conserves a corrected momentum)
Let , where . This is the free integrand plus for . For state translation take , and . Since , the criterion holds.
The momentum is , but the conserved quantity is . Indeed the Euler equation is . Along a nonconstant affine solution, need not vanish, even though the corrected quantity is constant. The integrand is invariant under translation only up to a boundary term.
A symmetry of the equation can fail the variational criterion
Counterexample mode
Scaling oscillator solutions does not conserve the proposed scaling charge
For with , the equation is . Multiplying a solution by a constant produces another solution. For a finite scale factor , however, ; with the independent variable unchanged, the density scales rather than remaining invariant. The state-scaling generator , gives the density expression .
No boundary function of our stated type can have for all velocities, since is affine in and the right side has a nonzero quadratic coefficient. This particular equation symmetry is therefore not a variational symmetry of the specified type. The naive quantity is not conserved: for , its derivative at is .
The oscillator still has its energy first integral from , , . What failed was the attempt to obtain a new charge from an unverified action-symmetry condition.
Practice
Quick check
(For with real and , test , and . Derive the equation and the conserved quantity, and decide whether that quantity must be a positive energy.)
The transformation combines translation of the independent variable with state scaling.
Solution · Guided solution
Here , so Euler–Lagrange is . The density criterion is
The two charge contributions are and . Thus the Noether quantity is , where
Thus is constant. Direct differentiation and substitution of the equation verify the same result. The usual expression instead has derivative along solutions; the mixed term in supplies the correction. Moreover, need not be positive: if , its quadratic form in height and velocity is indefinite. Conservation alone does not turn an arbitrary charge into a positive energy or a minimum certificate.
Quick check
(For the planar integrand with real , derive the Euler equation and the rotational Noether quantity. Compare it with alone.)
Use the rotation matrix from the example. The velocity-linear term changes the momentum.
Solution · Guided solution
The momentum and state gradient are and . Consequently the equation is . Under the rotation generator, the two velocity-linear contributions to the density criterion cancel, giving and
The derivative of the first two terms alone is ; the final term has derivative . Their sum is constant. In particular, discarding the momentum correction would generally lose conservation.
Quick check
(For on with , verify the generator , , . Combine its conserved quantity with the autonomous energy to determine the form of every positive classical extremal.)
Integrate the derivative of . Keep the positive square root and check the relation among the resulting constants.
Solution · Guided solution
Now and . The density criterion is . The two first integrals are
The two constants are fixed on the whole connected interval of the extremal. Thus throughout that interval, and integration gives a single quadratic there:
The relation follows by substituting into the energy equation. Conversely, choose and , set , and take the positive square root of the displayed quadratic. The quadratic is , and differentiation verifies , the Euler equation. Every positive classical extremal has this form on its interval. This reconstructs stationary curves; it does not prove that each arc minimizes an arbitrary fixed-endpoint problem.