Ulam's Accelerator: Classical vs. Quantum ✛ ✎
Context
The Fermi–Ulam problem stands apart in dynamical systems theory: it is a rare example where the classical and quantum versions of the same model give opposite answers. Nearby in the archive lie works on quantum tides from the future, on the quantum logic of the brain, and on gravitational noise erasing superpositions, but all of them concern different mechanisms of acceleration or decoherence. This article returns to the roots, to the original 1961 model, and carries its analysis to completion.
Methods
The authors use adiabatic coordinates, in which the particle's momentum is conserved between collisions, and then construct adiabatic normal forms for the jumps at the wall's kinks. At parameter resonance (when the amplitude ratio is an integer), the dynamics on invariant circles reduces to a skew product over an interval exchange with a piecewise constant cocycle. For the quantum version, they apply a 'wall stopping' transformation that converts boundary conditions into delta-function potential impulses, after which the Schrödinger equation is solved via finite-dimensional Floquet matrices.
Results
Theorem 1: under classical resonance, the set of accelerating orbits has measure zero, and almost every orbit returns infinitely often to its initial energy level. Theorem 2: in the quantum case, energy grows quadratically with the number of periods, E(N) = aN² + bN + c, with coefficient a ≥ 0. Theorem 3: the quasienergy spectrum is absolutely continuous and consists of at most q components; degeneracy of components to a point suppresses energy growth. For the 1:1 resonance, an explicit formula is obtained: a = 2(J₂−J₁)² ∫ x²|φ|² dx > 0.
Implications
Unlike neighboring works where quantum effects merely modify the classical picture, here quantization radically changes the conclusion: classical typical recurrence gives way to quantum acceleration. This is the first mathematically proven example since the quantum rotator where a quantum system differs essentially from its classical prototype. The work also provides a complete answer to Ulam's original question, up to a set of measure zero.
Future development
The non-resonant case remains open: there the skew product disappears, and the dynamics on invariant circles becomes an infinite interval exchange, a standalone unsolved problem. For the quantum version, it is expected that for Diophantine or Liouville resonance numbers the absolutely continuous spectrum will degenerate into singular continuous or point spectrum, but the analysis of large matrices currently hits computational difficulties.
Impact
The results are important for the physics of cold atoms in dynamic traps and for mesoscopic systems with time-dependent boundaries.
Next steps
Investigate non-resonant parameters and the behavior of the quasienergy spectrum as resonance numbers approach Diophantine or Liouville values.
Key open problems
In dynamical systems theory, the typical behavior of infinite interval exchanges arising in the non-resonant case remains open; in quantum chaos, the problem is Anderson localization for systems with delta-function impulses.
The knowledge machine's view
This section is addressed to the author of the paper and their colleagues. Neighbouring works were found by meaning-based search across our archive, not by word overlap. We do not review: we show the field around the work.
What the work does
This work settles Ulam's 1961 question: in the piecewise-linear Fermi–Ulam model, accelerating orbits do exist, but they form a set of measure zero. Under a generic resonance, the classical dynamics is recurrent almost everywhere, whereas the quantum version exhibits quadratic energy growth. The authors construct an explicit procedure for finding accelerating orbits and link the energy growth to the shape of the quasienergy spectrum.
Where its strength is
A complete answer to Ulam's original question up to a set of measure zero, along with an explicit demonstration of the difference between classical and quantum dynamics in a single model, is a rare result that can serve as a foundation for further generalizations.
Where it could go next
- It might be interesting to see how accounting for gravitational noise, which erases quantum superpositions, affects the quadratic energy growth in the quantum Fermi–Ulam model.based on: Gravitational Noise: How Ripples in Space Erase Quantum Superpositions arXiv:2608.06121v1
- Nearby lies work on quantum motors at near-light speeds, where motion is treated as fuel; perhaps the methods for analyzing acceleration there intersect with your adiabatic coordinates.based on: Wind in the Sails of a Heat Engine: How Motion Bypasses the Carnot Limit arXiv:2508.14183
- In the archive there is an analysis of a quantum dispatcher for a power grid that uses finite-dimensional approximation and Floquet matrices; perhaps their computational techniques apply to your resonant cases.based on: A Quantum Sieve for the Power Grid arXiv:2505.00145
Worth trying
- Test whether quadratic energy growth persists when weak gravitational noise is added as a source of decoherence.
- Compare your procedure for finding accelerating orbits with optimization methods used in quantum dispatch for power grids.
- Consider a relativistic generalization of the Fermi–Ulam model, taking into account effects described in the work on quantum motors at near-light speeds.
Why this may matter
A thorough analysis of the original Fermi–Ulam model provides a benchmark example where quantization radically changes the system's behavior, and it opens the way to studying the stability of this difference under external perturbations.
Nearby works in our archive
- A Cosmic Net for Quantum Gravitysimilarity 0.50
- A Quantum Sieve for the Power Gridsimilarity 0.49
- Quantum Logic of the Brain: Macroscopic Analogues of Entanglement Without Particlessimilarity 0.49
- Gravitational Noise: How Ripples in Space Erase Quantum Superpositionssimilarity 0.48
- Quantum First Aid for Power Gridssimilarity 0.48
- Gravity: Classical or Quantum?similarity 0.47
- Wind in the Sails of a Heat Engine: How Motion Bypasses the Carnot Limitsimilarity 0.47
The archive is sparse around this work: the closest review is only 0.50, against a usual 0.64. It looks like few have drilled nearby.
Ulam chose the parameters A=1/√2, B=√2, T=1 almost at random, and they turned out to be a special case of the 1:1 resonance, where classical accelerating orbits exist but are extremely rare.
Key numbers
- dimension of Floquet matrices: q (integer from resonance p/q)
- number of components of the quasienergy spectrum: at most q
- energy growth exponent: 2 (quadratic)
- fraction of accelerating classical orbits: 0 (measure zero)