Hamiltonian Mechanics from a contact–geometric perspective

Hamiltonian mechanics can be derived from a simple variational principle involving only a 1-form. The key idea is that the familiar action

S=(padqaHdt)

is nothing more than the restriction of the tautological 1-form on the extended cotangent bundle (T(Q×Rt)). This viewpoint unifies:


1. The Tautological 1-Form

Recall the tautological 1-form. For any manifold M, the cotangent bundle TM carries a canonical 1-form

θΩ1(TM),θη(v):=η(πv),

where ηTxM and π:TMM is the projection.
In coordinates (xi,pi),

θ=pidxi.

2. Apply this to the Extended Configuration Space

Take

M=Q×Rt.

A point of TM has coordinates (qa,t;pa,pt), and the tautological form becomes

θ=padqa+ptdt.

This is a single geometric object from which the entire Hamiltonian formalism can be recovered.

3. The Hamiltonian Constraint and the Physical 1-Form

Introduce the extended Hamiltonian constraint

H(q,p,t,pt)=pt+H(q,p,t).

This condition can be interpreted as saying: "the energy, which is the momentum corresponding to t, is given in this particular system by the function H".
The physical motions lie on the hypersurface

Σ:={H=0}T(Q×R).

Restrict the tautological 1-form to this hypersurface:

α=θ|Σ=padqaH(q,p,t)dt.

This form α is the action 1-form of Hamiltonian mechanics. Its exterior derivative is

dα=dqadpa+dtdH,

which restricts the symplectic structure to the hypersurface.

4. The Variational Principle (S=α)

Given a curve γ(τ)Σ, define

S[γ]=γα.

Varying qa(τ),pa(τ),t(τ) produces the Euler–Lagrange equations of this action. The result is:

q˙a=Hpat˙,p˙a=Hqat˙.

Choosing the parameter τ=t yields the usual Hamilton equations:

q˙a=Hpa,p˙a=Hqa.

Thus the entire structure of Hamiltonian mechanics comes from the line integral of the 1-form α.
For the physical motivation see tautological 1-form#Physical Intuition The "Action" of a Nudge, or directly, action#What is the action, intuitively?.

5. Contact Geometry on the Constraint Surface

The hypersurface Σ={H=0} is automatically a contact manifold, because:

6. The Reeb Vector Field of α

Given a contact manifold (Σ,α), the Reeb vector field R is defined by:

α(R)=1,ιRdα=0.

These equations uniquely determine R.
Let the Reeb vector field be written as:

R=aiqi+bipi+ct

We have

dα=dpidqi+Hqidtdqi+Hpidtdpi

and then

iRdα=(bi+cHqi)dqi+(ai+cHpi)dpi+(aiHqibiHpi)dt=0

Therefore:

ai=cHpibi=cHqi

(Note: The dt coefficient equation is automatically satisfied if these two are plugged in, confirming consistency.)

Now we substitute the expressions for ai and bi into the normalization constraint.

α(R)=piaiHc=1

Substitute ai=cHpi:

pi(cHpi)Hc=1

Factor out c:

c(piHpiH)=1

The coefficients (a,b,c) are given by:

c=1pkHpkHai=HpipkHpkHbi=HqipkHpkH

The key fact:

The Reeb flow of α is precisely the Hamiltonian flow of the constraint H=pt+H, modulo reparametrization.

In other words:

8. Relation Between the Reeb Field and the Hamiltonian Vector Field

Consider the Hamiltonian vector field XH on the ambient symplectic manifold:

ιXHω=dH.

Properties:

Thus the Reeb field is simply a normalization:

R=XHα(XH).

Hence:

9. Summary of the Geometric Picture

  1. Start with the extended configuration space Q×Rt.
  2. Take its cotangent bundle T(Q×R) and its tautological form θ.
  3. Impose the Hamiltonian constraint H=pt+H=0.
  4. Restrict θ to this hypersurface: $$\alpha = \theta|_\Sigma = p_a , dq^a - H , dt.$$
  5. The action is the line integral $$
    S=\int_\gamma\alpha.
  6. The hypersurface is a contact manifold whose Reeb field is the normalized Hamiltonian flow.
  7. Rovelli’s action padqa with H=0 is the projection of this contact mechanics to the (q,p)-space.

10. Conceptual Takeaways

11. Analogy with Second-Order ODEs

A helpful way to interpret the Hamiltonian constraint

Σ={H=pt+H=0}T(Q×R)

is to compare it with the geometric description of second-order ODEs on higher jet spaces.

1. Contact Geometry in Jet Spaces

The second jet bundle J2(Q), with coordinates (q,q˙,q¨), carries a canonical contact system generated by forms such as

β1=dqq˙dt,β2=dq˙q¨dt.

These express the universal kinematic identities:

dq=q˙dt,dq˙=q¨dt.

Nothing dynamical has been imposed yet; these relations only ensure that a curve in J2(Q) arises as the 2-jet of a true function q(t).

2. Dynamics Arise from a Hypersurface in J2(Q)

A second-order ODE

q¨=f(q,q˙,t)

is encoded geometrically by introducing the dynamical hypersurface

Ξ={q¨f(q,q˙,t)=0}J2(Q).

Integral curves tangent to both:

Thus:

3. Parallel Structure in Contact Hamiltonian Mechanics

In the extended cotangent bundle T(Q×Rt), the fundamental geometric object is the tautological 1-form

θ=padqa+ptdt.

Restricting to the Hamiltonian constraint

Σ={pt+H(q,p,t)=0},

one obtains the physical 1-form

α=θ|Σ=padqaHdt,

which defines a contact structure on Σ.

This mirrors the jet-space picture:

Jet-space mechanics (in J2) Contact Hamiltonian mechanics
Contact system β1,β2 gives universal kinematics Tautological 1-form θ gives universal p-dq-dt kinematics
Hypersurface Ξ:q¨=f(q,q˙,t) encodes the law Hypersurface Σ:pt+H=0 encodes the Hamiltonian
Dynamics = curves tangent to Ξ and contact system Dynamics = Reeb/Hamiltonian flow tangent to Σ

4. Conceptual Correspondence

Both constructions share the same architecture:

A contact structure provides universal kinematics.
A hypersurface specifies the physical system.

In jet spaces:

In Hamiltonian mechanics:

Thus, Hamiltonian dynamics stands to the tautological 1-form exactly as Newtonian dynamics stands to the jet-space contact system:
both arise from imposing a dynamical hypersurface on top of a universal geometric background.