Liouville measure

In plain terms

Any theory with a deterministic underworld needs to say how much of that world is one way rather than another, meaning how big a set of possible states is, and the notion of size had better be one the motion respects. If flowing forward in time inflated some regions and shrank others, sizes would depend on when you measured them, and probabilities built from them would be meaningless.

Classical mechanics solved this long ago. The complete state of a classical system, all of its positions and all of its velocities, is a single point in a space called phase space, and the laws of motion move that point along a path. Liouville proved that this motion preserves the volume of phase space exactly, so a region can be stretched and folded into filigree but its total volume never changes, and a volume with this property is called a Liouville measure.

The residual worry is choice. If several motion-respecting volumes existed, a theorist could pick whichever delivered the desired answer, and any derivation of probabilities would be quietly rigged. The theorem here closes that gap, as on the sector CSD works with, the one whose motion is unitary, there's exactly one option.

In CSD

This is the hinge that joins the ontic story, about what exists, to the probability story. The substrate carries a volume its flow preserves, and the epistemic side, which concerns what an observer can know, needs the Fubini-Study measure to turn regions into Born weights. On the unitary-flow sector these are proved to be the same object, so the ruler used for probability is the substrate's own volume and not a second, conveniently chosen assumption.

That identification is the programme's reply to the oldest objection against reasoning from typicality, pressed from Boltzmann's day to Valentini's, that the measure is smuggled in. Here it's forced, and the forcing is a theorem with a uniqueness proof behind it.

The scope should be stated plainly. That the flow preserves volume is one of CSD's stated postulates about the substrate, carried as an explicit assumption rather than derived for constraint dynamics in general. For every flow generated by a smooth energy function on the space of states, Liouville's theorem is proved in the corpus at the level of the manifold itself, so the postulate is a theorem for all such flows, and what keeps it a postulate is that the measurement steps of the constraint dynamics aren't of that kind globally. What the corpus proves about the measure is conditional, so if the sector is this one then its volume is forced and coincides with the geometric volume of the space of states. The theorem is about the sector and not a derivation of it.

Mathematically

For the unitary-flow sector, the Liouville measure equals the Fubini-Study measure, and it's the forced Kahler volume, determined by the state space and its unitary symmetry rather than posited. It's also proved to be the symplectic volume, the normalised top power of the Fubini-Study form (unitaryFlowSetup_liouvilleMeasure_eq_fsVolumeNormalized), which the sector's flow preserves (unitaryFlowSetup_flow_measurePreserving_fsVolume), and the forced volume and the symplectic volume are the same measure (fsVolumeNormalized_isForcedKahlerVolume).

The scope is load-bearing, so it's stated plainly. Measure-preservation of the flow is carried as a structure field, an ontic postulate listed in the axiom ledger, and not derived for the constraint dynamics in general. It's a theorem for the unitary flows, by invariance, and for every Hamiltonian flow of a smooth energy on complex projective space by the manifold-level Liouville theorem (IsSymplectic.map_hamiltonianFlow_topFormMeasure_wedgePow, instantiated as fsVolume_map_hamiltonianFlow), which says the flow of the Hamiltonian vector field preserves the measure of every power of the symplectic form, proved through the flat Liouville theorem in charts with Cartan's formula and the change of variables for top forms. The measurement pieces of the constraint dynamics are only locally Hamiltonian, which is why the field remains a postulate. What's proved about the measure is the conditional, so if the sector is this one then its volume is forced and is the symplectic volume, and the theorem is about the sector and not a derivation of it.

The name

Joseph Liouville (1809-1882) was born at Saint-Omer and trained at the Ecole Polytechnique. He taught there, at the College de France and at the Sorbonne. The volume-preservation theorem dates from 1838, and he reached it through celestial mechanics rather than any inkling of statistical physics.

He left two other legacies. He founded the Journal de Mathematiques Pures et Appliquees, which is still publishing, and he used it in 1846 for an act of scientific rescue, printing the papers of Evariste Galois fourteen years after Galois died in a duel at twenty, and thereby saving the founding documents of group theory from oblivion. He was also the first to exhibit a provably transcendental number, decades before e and pi were settled.

Related entries

• Fubini-Study measureThe space of quantum states allows exactly one way of measuring size that respects its symmetry, and that one measure is what turns regions of the space into Born-rule probabilities.
• Kahler formOn the space of quantum states, distance, area and a quarter-turn coming from the complex numbers lock together, and that lock is what makes the volume of the space nobody's choice.
• Constraint surfaceThere's one arena, it's deterministic, and it doesn't grow, and everything else in the programme is bookkeeping about how much of it a measurement context lets you see.
• TypicalityTypicality is probability without randomness, as the world is deterministic, you don't know which state you have, and the odds are the sizes of the possibilities.
• Poisson bracketThe Poisson bracket is the one rule that tells classical mechanics how fast any quantity changes, and the quantum commutator obeys exactly the same algebra, so that shared grammar is what lets a deterministic flow present itself in quantum form.
• Koopman operatorClassical mechanics has a Hilbert space and a unitary time evolution of its own, so neither of those is what makes a theory quantum, and what's left as the difference is that quantum observables don't commute.