The charged particle and the twisted symplectic form

Intuition Beginner

A free particle, with no forces on it, drifts in a straight line. A charged particle in a magnetic field instead curls into circles. The usual way to explain the difference is to add a force term to the equation of motion. There is a second way, and it is the heart of this unit: leave the particle free, and instead bend the geometry of its phase space.

Phase space is the space of all position-and-momentum pairs. It carries a structure, the symplectic form, that turns an energy rule into motion. If you tilt that structure by an amount set by the magnetic field, a particle that still only "knows" its kinetic energy starts to spiral exactly as the magnetic force would make it. The magnetic field has been baked into the geometry rather than added as a push.

Visual Beginner

Picture phase space as a flat sheet ruled by a grid of position-momentum cells. The free particle slides across this flat sheet in straight tracks. Now imagine warping the sheet: each cell gets a small twist whose size is the local magnetic field. A particle gliding on the warped sheet follows curved tracks even though nothing pushes it.

The picture to hold: the same free-particle energy rule, read on a twisted sheet, produces the circling motion of a charge in a magnetic field. The amount of twist is the charge times the field.

Worked example Beginner

Take a particle in the plane in a uniform magnetic field pointing out of the page. Without the field, the energy rule sends the particle straight along whatever direction it starts moving. With the field, add a fixed twist between the two momentum directions.

Start the particle moving to the right. The twist deflects part of its motion upward. A moment later it is moving up, and the twist deflects that part to the left. Step by step the velocity vector rotates at a steady rate, and the particle traces a circle.

The radius shrinks as the field strengthens and grows as the speed increases. The sense of circulation flips if the charge flips sign.

What this tells us: a steady twist in phase space, with no force added by hand, is enough to bend a free particle into the circular orbit of a charge in a magnetic field.

Check your understanding Beginner

Formal definition Intermediate+

Let $Q$ be the configuration manifold of a particle and $T^{\ast }Q$ its cotangent bundle, the canonical phase space 05.02.05, with projection $\pi :T^{\ast }Q\to Q$. The canonical (Liouville) one-form $\theta$ gives the canonical symplectic form ${\omega }_0=-d\theta$, written ${\omega }_0={\sum }_{i}d{q}^i\wedge d{p}_i$ in cotangent coordinates 05.01.02. Electromagnetism enters through a closed two-form $F$ on $Q$, the field strength, satisfying $dF=0$ (the homogeneous Maxwell equations).

Definition (twisted symplectic form). For a particle of charge $q$, the twisted (or magnetic) form on $T^{\ast }Q$ is

$${\omega }_F={\omega }_0+q{\pi }^{\ast }F.$$

Because ${\pi }^{\ast }F$ is pulled back from the base it involves only the $dq$ directions; adding it to ${\omega }_0$ shifts the $d{q}^i\wedge d{p}_i$ block while leaving the pairing of each $d{q}^i$ with its $d{p}_i$ intact. The form ${\omega }_F$ is closed, since $d{\omega }_F=d{\omega }_0+q{\pi }^{\ast }(dF)=0$, and nondegenerate, since the canonical pairing of $p_i$ with $q^i$ survives the lower-triangular shift. Hence $(T^{\ast }Q,{\omega }_F)$ is a symplectic manifold for every closed $F$.

Definition (charged-particle dynamics). The Hamiltonian is the unmodified free kinetic energy $H=\frac{1}{2m}{g}^{ij}{p}_i{p}_j$ for a metric $g$ on $Q$ (in flat space $H=|p{|}^2/2m$). The dynamics is the Hamiltonian flow of $H$ with respect to ${\omega }_F$: the vector field $X_H$ determined by ${\iota }_{X_H}{\omega }_F=dH$. All the electromagnetic coupling lives in the symplectic form, none in $H$.

Counterexamples to common slips

• The deformation is in $\omega$, not $H$. A frequent error is to add $q{\pi }^{\ast }F$ and shift $H$ to $\frac{1}{2m}|p-qA{|}^2$; doing both double-counts the coupling. The twisted-form description keeps $H$ free and puts the field entirely in ${\omega }_F$.
• $F$ must be the field two-form, not the potential. The shift uses ${\pi }^{\ast }F$ with $F$ closed; it does not use $A$. This is what lets the construction survive when no global $A$ exists.
• Nondegeneracy is not automatic for an arbitrary added two-form. It holds here because ${\pi }^{\ast }F$ has no $dp\wedge dp$ part — it is pulled back from the base — so the momentum block of ${\omega }_F$ is still the identity pairing. A two-form mixing the momentum directions could destroy nondegeneracy.

Key theorem with proof Intermediate+

Theorem (the twisted form generates the Lorentz force). *Let $Q={\mathbb{R}}^3$ with flat metric, $H=|p{|}^2/2m$, and $F$ the magnetic field two-form $F=\frac{1}{2}{F}_{jk}d{q}^j\wedge d{q}^k$ with $B^i=\frac{1}{2}{\epsilon }^{ijk}{F}_{jk}$. The Hamiltonian flow of $H$ with respect to ${\omega }_F={\omega }_0+q{\pi }^{\ast }F$ is*

$${\dot{q}}^i=\frac{p_i}{m},{\dot{p}}_i=q{F}_{ij}{\dot{q}}^j,$$

equivalently $m\ddot{\mathbf{q}}=q\dot{\mathbf{q}}\times \mathbf{B}$. Adjoining the time-dependent electric part promotes this to the full Lorentz force $m\ddot{\mathbf{q}}=q(\mathbf{E}+\dot{\mathbf{q}}\times \mathbf{B})$.

Proof. Write $X_H={\dot{q}}^i{\partial }_{q^i}+{\dot{p}}_i{\partial }_{p_i}$. In coordinates,

$${\omega }_F=\sum _{i}d{q}^i\wedge d{p}_i+q\frac{1}{2}{F}_{jk}d{q}^j\wedge d{q}^k.$$

Contract with $X_H$. The canonical part gives ${\iota }_{X_H}{\omega }_0={\dot{q}}^{i}d{p}_i-{\dot{p}}_{i}d{q}^i$. The magnetic part gives ${\iota }_{X_H}(q\frac{1}{2}{F}_{jk}d{q}^j\wedge d{q}^k)=q{F}_{jk}{\dot{q}}^{j}d{q}^k$, using antisymmetry of $F_{jk}$. Therefore

$${\iota }_{X_H}{\omega }_F={\dot{q}}^{i}d{p}_i-{\dot{p}}_{i}d{q}^i+q{F}_{jk}{\dot{q}}^{j}d{q}^k.$$

The differential of $H=|p{|}^2/2m$ is $dH=(p_i/m)d{p}_i$. Matching ${\iota }_{X_H}{\omega }_F=dH$ coefficient by coefficient: the $d{p}_i$ terms give ${\dot{q}}^i=p_i/m$, and the $d{q}^i$ terms give $-{\dot{p}}_i+q{F}_{ji}{\dot{q}}^j=0$, that is ${\dot{p}}_i=q{F}_{ij}{\dot{q}}^j$ after using $F_{ji}=-F_{ij}$. Substituting ${\dot{q}}^j=p_j/m$ and ${\dot{p}}_i=m{\ddot{q}}^i$ yields $m{\ddot{q}}^i=q{F}_{ij}{\dot{q}}^j=q(\dot{\mathbf{q}}\times \mathbf{B}{)}^i$. A time-dependent formulation on the evolution space $T^{\ast }Q\times \mathbb{R}$ adds the electric two-form $dt\wedge (q{E}_{i}d{q}^i)$, contributing ${\dot{p}}_i\supset q{E}_i$ and completing the Lorentz force. $\square$

Bridge. This computation builds toward the Souriau program in which an "elementary system" is a symplectic manifold rather than a force law, and it appears again in the prequantisation units 05.11.01 where the same closed two-form $F$ becomes the curvature of a line bundle. The foundational reason the construction works is structural: the Lorentz force is not an extra term but the shadow of a deformation of $\omega$, and this is exactly the content of minimal coupling read geometrically. The central insight is that charge is the coupling constant of a cohomology class, $q[F]$, and putting these together shows why the description survives even when no global potential exists — the next sections turn that survival into the monopole and the Dirac quantisation condition.

Exercises Intermediate+

Advanced results Master

The evolution-space (presymplectic) formulation. Souriau's original setting is not $T^{\ast }Q$ but the evolution space — for a relativistic charge, the mass shell inside $T^{\ast }{\mathbb{R}}^{1,3}$ with a degenerate two-form $\sigma ={\sigma }_0+qF$. The kernel of $\sigma$ is one-dimensional and integrates to the equations of motion: the worldlines are the characteristic curves $\ker \sigma$, with no Hamiltonian chosen at all. Quotienting by this characteristic foliation returns a genuine symplectic manifold, the space of motions, whose points are entire trajectories. The Lorentz force appears as the statement that the characteristic direction of ${\sigma }_0+qF$ is the charged worldline. This presymplectic viewpoint is the one that generalises cleanly to the relativistic and Yang-Mills cases and is the same reduction-of-a-degenerate-form mechanism used for the relativistic particle 05.11.09.

Non-abelian generalisation (Sternberg phase space). When $F$ is replaced by the curvature of a principal $G$-bundle (a Yang-Mills field), Sternberg showed that the correct phase space is the associated bundle $T^{\ast }Q{\times }_G\mathcal{O}$, where $\mathcal{O}$ is a coadjoint orbit carrying the particle's internal (isospin) degrees of freedom, with a twisted symplectic form built from the connection. Wong's equations — the non-abelian Lorentz force — are the Hamiltonian flow of the free kinetic energy on this space ^{[Sternberg 1977]}. The abelian charge $q$ becomes a point of $\mathcal{O}$, and Dirac quantisation becomes the integrality condition selecting prequantisable orbits.

Globality and the integrality lattice. The twisted form ${\omega }_F$ is globally defined for any closed $F$, but the quantum theory requires more: a prequantum line bundle of curvature $qF/\hslash$ exists over $Q$ iff $[qF/2\pi \hslash ]\in H^2(Q,\mathbb{Z})$. For $Q$ with $H^2(Q,\mathbb{R})\ne 0$ — the monopole exterior ${\mathbb{R}}^3\setminus \{0\}\simeq S^2$ — this is a genuine constraint, Dirac's $qg\in 2\pi \hslash \mathbb{Z}$. The same integral class governs whether the symplectic manifold is prequantisable, tying classical magnetic coupling to the geometric-quantisation integrality condition 05.11.01, 05.11.02.

Proposition (gauge independence of the twisted form). Two vector potentials $A$, $A^{\prime }=A+d\chi$ with the same field $F=dA=d{A}^{\prime }$ give symplectomorphic minimal-coupling systems, and both equal the single twisted form ${\omega }_F$. The shift ${\Psi }_{d\chi }$ is fibre translation by $qd\chi$, a symplectomorphism of ${\omega }_0$; the twisted form depends only on $F$, so it is manifestly gauge-invariant — this is why ${\omega }_F$, not the $A$-dependent canonical description, is the intrinsic object.

Synthesis. The twisted symplectic form is the foundational reason minimal coupling is a geometric rather than a dynamical operation: the magnetic field is a closed two-form on configuration space, and coupling a charge to it is the deformation ${\omega }_0\mapsto {\omega }_0+q{\pi }^{\ast }F$ of phase-space geometry, with the Hamiltonian left untouched. This is exactly the inversion Souriau builds his whole structure on, and it generalises in two directions at once. Toward quantisation, the class $[qF/2\pi ]$ that must be integral for a global prequantum bundle is dual to the Dirac quantisation of magnetic charge — the central insight that a topological constraint on the field is identical to a quantisation of charge. Toward gauge theory, replacing $F$ by a Yang-Mills curvature produces Wong's equations on the Sternberg phase space, so the abelian charged particle is the simplest face of a uniform construction. Putting these together, the bridge is unmistakable: the canonical form 05.02.05, the Lagrangian charged particle 05.00.09, and prequantisation 05.11.01 are three readings of one object, ${\omega }_0+q{\pi }^{\ast }F$, and the monopole shows that when the gauge potential disappears the symplectic form is what remains.

Full proof set Master

Proposition (nondegeneracy of the twisted form). *For any closed two-form $F$ on $Q$ and any $q$, the form ${\omega }_F={\omega }_0+q{\pi }^{\ast }F$ is nondegenerate, hence symplectic.*

Proof. Work at a point in cotangent coordinates $(q^i,p_i)$. The canonical form is ${\omega }_0=d{q}^i\wedge d{p}_i$, and ${\pi }^{\ast }F=\frac{1}{2}{F}_{jk}(q)d{q}^j\wedge d{q}^k$ involves only the $dq$ directions. In the ordered basis $({\partial }_{q^1},\dots ,{\partial }_{q^n},{\partial }_{p_1},\dots ,{\partial }_{p_n})$ the matrix of ${\omega }_F$ is the block form

$$\left(\begin{array}{cc}qF& I\\ -I& 0\end{array}\right),$$

where $F$ is the antisymmetric matrix $(F_{ij})$ and $I$ the $n\times n$ identity. Its determinant is computed by the block formula: with the lower-right block zero and the off-diagonal blocks $\pm I$ invertible, $\det \left(\begin{array}{cc}qF& I\\ -I& 0\end{array}\right)=\det (I)\det (0-(-I)(I{)}^{-1}(qF))$... more directly, row-reduce using the $-I$ block to clear $qF$: adding $qF$ times the lower block rows to the upper rows annihilates the $qF$ entry without touching $\det$, leaving $\left(\begin{array}{cc}0& I\\ -I& 0\end{array}\right)$, whose determinant is $(\pm 1{)}^n\ne 0$. Hence ${\omega }_F$ is nondegenerate at every point, and being closed (Exercise 1) it is symplectic. $\square$

Proposition (the magnetic Poisson brackets). Under ${\omega }_F$ on ${\mathbb{R}}^3$ with $H=|p{|}^2/2m$, the kinetic momenta ${\Pi }_i:=p_i$ satisfy $\{q^i,{\Pi }_j{\}}_F={\delta }_j^i$ and $\{{\Pi }_i,{\Pi }_j{\}}_F=-q{F}_{ij}$.

Proof. The Poisson bracket of ${\omega }_F$ is the inverse of its matrix. Inverting the block matrix above gives the bivector $\{q^i,p_j\}={\delta }_j^i$, $\{q^i,q^j\}=0$, and $\{p_i,p_j\}=-q{F}_{ij}$. Concretely, from ${\iota }_{X_f}{\omega }_F=df$ one solves $X_{q^i}=-{\partial }_{p_i}$ (up to the magnetic correction, which vanishes on $q^i$) and $X_{p_i}={\partial }_{q^i}+q{F}_{ij}{\partial }_{p_j}$; then $\{p_i,p_j{\}}_F={\omega }_F(X_{p_i},X_{p_j})=-q{F}_{ij}$. The nonvanishing momentum-momentum bracket is the algebraic signature of the magnetic field: the kinetic momenta no longer commute, and their bracket is exactly $-q{F}_{ij}$. $\square$

Connections Master

• Cotangent bundle and the canonical form 05.02.05. The twisted form is the canonical ${\omega }_0$ of that unit plus the base-pulled-back field term $q{\pi }^{\ast }F$. Everything about Darboux coordinates, the Liouville one-form, and nondegeneracy is inherited from there; the present unit deforms exactly one block of that structure and tracks what survives.

• Lagrangian charged particle 05.00.09. When $F=dA$ the twisted-form system is symplectomorphic (Exercise 2-3) to the minimal-coupling Lagrangian system with $L=\frac{1}{2}m|\dot{\mathbf{q}}{|}^2-q\mathbf{A}\cdot \dot{\mathbf{q}}$ treated there. The two units describe the same helical and cyclotron motion; this unit supplies the intrinsic, potential-free formulation that 05.00.09 could not give, and it removes the dangling forward-reference that pointed at minimal coupling with no symplectic home.

• Symplectic manifold 05.01.02. The unit is a concrete and physically central family of symplectic manifolds $(T^{\ast }Q,{\omega }_F)$ parametrised by a charge and a closed two-form, illustrating that the symplectic structure is data to be chosen, not a fixed background.

• Prequantum line bundle and integrality 05.11.01, 05.11.02. The Dirac quantisation condition $[qF/2\pi ]\in H^2(Q,\mathbb{Z})$ is the prequantisability condition of those units applied to the magnetic class. The monopole is the canonical example where integrality of a magnetic flux becomes quantisation of a coupling constant.

• Relativistic particle and presymplectic reduction 05.11.09. The evolution-space formulation with the degenerate ${\sigma }_0+qF$ and its characteristic foliation is the same reduce-a-presymplectic-form mechanism used to build the relativistic massive particle as a coadjoint orbit.

Historical & philosophical context Master

The geometric reading of magnetic coupling is due to Jean-Marie Souriau, who in Structure des systèmes dynamiques (1970) ^{[Souriau 1970]} §13 placed the charge directly into the two-form governing the evolution space, treating an "elementary system" as a symplectic manifold rather than as a particle subject to a force law. This inverts the Newtonian order of explanation: the Lorentz force is derived from the geometry, not posited. Shlomo Sternberg's 1977 paper ^{[Sternberg 1977]} extended the construction to Yang-Mills fields, producing what is now called the Sternberg phase space and recovering Wong's equations, and the abelian case became a textbook fixture through Marsden and Ratiu ^{[Marsden-Ratiu 1999]}.

The monopole strand traces to Paul Dirac's 1931 argument ^{[Dirac 1931]} that the consistency of quantum mechanics in the field of a magnetic pole forces the product of electric and magnetic charge to be quantised. In the symplectic language this is the integrality of the cohomology class $[qF/2\pi ]$ — the same condition that, in geometric quantisation, decides whether a symplectic manifold admits a prequantum line bundle. The philosophical content is that a topological fact about a classical field (a nonzero flux class) and a quantisation rule for charge are one statement seen from two sides, a unification that the force-law formulation cannot express because it never names the underlying two-form.

Bibliography Master

@book{Souriau1970Structure,
  author    = {Souriau, Jean-Marie},
  title     = {Structure des syst{\`e}mes dynamiques},
  publisher = {Dunod},
  address   = {Paris},
  year      = {1970},
  note      = {English transl.: Structure of Dynamical Systems, Birkh{\"a}user, 1997}
}

@article{Sternberg1977Minimal,
  author  = {Sternberg, Shlomo},
  title   = {Minimal coupling and the symplectic mechanics of a classical particle in the presence of a Yang-Mills field},
  journal = {Proceedings of the National Academy of Sciences},
  volume  = {74},
  number  = {12},
  pages   = {5253--5254},
  year    = {1977}
}

@book{MarsdenRatiu1999Introduction,
  author    = {Marsden, Jerrold E. and Ratiu, Tudor S.},
  title     = {Introduction to Mechanics and Symmetry},
  series    = {Texts in Applied Mathematics},
  volume    = {17},
  edition   = {2nd},
  publisher = {Springer},
  year      = {1999}
}

@article{Dirac1931Quantised,
  author  = {Dirac, Paul A. M.},
  title   = {Quantised singularities in the electromagnetic field},
  journal = {Proceedings of the Royal Society of London A},
  volume  = {133},
  number  = {821},
  pages   = {60--72},
  year    = {1931}
}

@book{GuilleminSternberg1984Symplectic,
  author    = {Guillemin, Victor and Sternberg, Shlomo},
  title     = {Symplectic Techniques in Physics},
  publisher = {Cambridge University Press},
  year      = {1984}
}