Physlib

PhyslibAlpha.ClassicalMechanics.NortonDome.Solution

The motions of the Norton dome and the failure of uniqueness

i. Overview

A particle at rest on the apex of the Norton dome satisfies `r̈ = √r` by staying there forever, and also by staying until an arbitrary instant `T ≥ 0` and then sliding off along `r(t) = (t - T)⁴ / 144`: that curve is `C²`, vanishes with its derivative at `T`, and its second derivative `(t - T)² / 12` is `√r`. All these motions have the same initial position and velocity, so the initial data do not determine the motion. This is Norton's observation.

For contrast, the pendulum's phase-space vector field is Lipschitz and Mathlib's `ODE_solution_unique_univ` gives `SimplePendulum.equationOfMotion_unique`; the force of the dome is not Lipschitz at the apex, `NortonDome.not_lipschitzOnWith_force`. Energy conservation does not help either, since every motion in the family has zero energy (`energy_solution`). Read backwards in time, in section E, a particle sliding up the dome may arrive at the apex at rest in finite time and stay there.

ii. Key results

- `NortonDome.delayedQuartic` is the real function `τ ↦ max (τ - T) 0 ⁴ / 144`, with its derivatives `hasDerivAt_delayedQuartic` and `hasDerivAt_deriv_delayedQuartic`, its `C²` regularity `contDiff_delayedQuartic`, and the identity `sqrt_delayedQuartic`. - `NortonDome.solution T` is the motion leaving the apex at the instant `T`. It is a solution, `solution_isSolution`, starts from rest at the apex when `0 ≤ T`, `solution_zero` and `deriv_solution_zero`, and is injective in `T`, `solution_injective`. - `NortonDome.rest_isSolution` is the motion staying at the apex. - `NortonDome.exists_isSolution_ne` is the failure of uniqueness: two distinct solutions with the same initial position and velocity; `NortonDome.not_isSolution_unique` restates it as the negation of the uniqueness property of the pendulum. All these motions have zero energy, `NortonDome.energy_rest` and `NortonDome.energy_solution`. - `NortonDome.IsSolution.comp_neg` is the time-reversal symmetry of the dome, and `NortonDome.arrival T` the time-reversed motion, off the apex before the instant `-T`, `arrival_apply_pos`, and at rest on it from then on, `arrival_of_le`.

iii. Table of contents

- A. The delayed quartic - A.1. Definition and sign - A.2. Derivatives and regularity - B. The motions leaving the apex - B.1. The definition and its values - B.2. Derivatives and regularity - B.3. The equation of motion - B.4. Distinct instants give distinct motions - C. The motion staying at the apex - D. The failure of uniqueness - E. Time reversal and arrival at the apex

iv. References

- Norton, J. D., *The dome: an unexpectedly simple failure of determinism*, Philosophy of Science 75 (2008), 786–798. - `Physlib.ClassicalMechanics.Pendulum.SimplePendulum.Solution` (uniqueness for a Lipschitz force, for contrast).

A. The delayed quartic

As a function of the real time `τ`, the motion leaving the apex at the instant `T` is `max (τ - T) 0 ⁴ / 144`; its calculus is that of the positive-part powers of `NortonDome.PosPartPow`.

A.1. Definition and sign

A.2. Derivatives and regularity

B. The motions leaving the apex

The motion leaving the apex at the instant `T` is the delayed quartic read on `Time`, times the unit vector of the arc-length coordinate.

B.1. The definition and its values

B.2. Derivatives and regularity

B.3. The equation of motion

B.4. Distinct instants give distinct motions

C. The motion staying at the apex

The force vanishes at the apex, so the constant curve there satisfies the equation of motion.

D. The failure of uniqueness

Rest at the apex and departure at any instant `T ≥ 0` are distinct solutions with the same initial data. So the uniqueness property `SimplePendulum.IsSolution.eq_of_initial` of the pendulum fails for the dome, whose force violates the Picard–Lindelöf hypothesis (`not_lipschitzOnWith_force`).

E. Time reversal and arrival at the apex

The equation of motion has no velocity term, so time reversal maps solutions to solutions. Reversing the departure at the instant `T` gives a motion sliding up the dome, arriving at the apex at the instant `-T` with zero velocity, and staying there. For a locally Lipschitz force this is impossible; for the dome it is the non-uniqueness read backwards.

38 declarations

definition

Delayed quartic function max⁡(τ−T,0)4144\frac{\max(\tau - T, 0)^4}{144}

For a given departure time T∈RT \in \mathbb{R} and an instant τ∈R\tau \in \mathbb{R}, the delayed quartic function is defined as (max⁡(τ−T,0))4144\frac{(\max(\tau - T, 0))^4}{144} This function is zero for all τ≤T\tau \le T and follows the power law (τ−T)4144\frac{(\tau - T)^4}{144} for τ>T\tau > T. It represents the position of a particle on the Norton dome that remains at rest at the apex until time TT.

theorem

delayedQuartic(T,τ)=max⁡(τ−T,0)4144\text{delayedQuartic}(T, \tau) = \frac{\max(\tau - T, 0)^4}{144}

For any departure time T∈RT \in \mathbb{R} and any instant τ∈R\tau \in \mathbb{R}, the delayed quartic function is given by the expression delayedQuartic(T,τ)=(max⁡(τ−T,0))4144.\text{delayedQuartic}(T, \tau) = \frac{(\max(\tau - T, 0))^4}{144}.

theorem

The Delayed Quartic Function is Non-negative

For any real numbers TT and τ\tau, the delayed quartic function (max⁡(τ−T,0))4144\frac{(\max(\tau - T, 0))^4}{144} is non-negative, i.e., 0≤(max⁡(τ−T,0))41440 \leq \frac{(\max(\tau - T, 0))^4}{144} where TT represents the instant the particle leaves the apex and τ\tau is the current time.

theorem

delayedQuartic(T,τ)=0\text{delayedQuartic}(T, \tau) = 0 for τ≤T\tau \le T

For any departure time T∈RT \in \mathbb{R} and any instant τ∈R\tau \in \mathbb{R} such that τ≤T\tau \leq T, the delayed quartic function delayedQuartic(T,τ)\text{delayedQuartic}(T, \tau), defined as (max⁡(τ−T,0))4144\frac{(\max(\tau - T, 0))^4}{144}, is equal to 00.

theorem

The delayed quartic is positive for τ>T\tau > T

For any departure time T∈RT \in \mathbb{R} and any instant τ∈R\tau \in \mathbb{R}, if τ>T\tau > T, then the delayed quartic function max⁡(τ−T,0)4144\frac{\max(\tau - T, 0)^4}{144} is strictly positive.

theorem

The square root of the delayed quartic function is max⁡(τ−T,0)212\frac{\max(\tau - T, 0)^2}{12}

For any departure time T∈RT \in \mathbb{R} and any instant τ∈R\tau \in \mathbb{R}, the square root of the delayed quartic function r(τ)=(max⁡(τ−T,0))4144r(\tau) = \frac{(\max(\tau - T, 0))^4}{144} is given by r(τ)=(max⁡(τ−T,0))212.\sqrt{r(\tau)} = \frac{(\max(\tau - T, 0))^2}{12}. This identity shows that the square root of the position function matches its second derivative r¨(τ)\ddot{r}(\tau), thereby satisfying the equation of motion r¨=r\ddot{r} = \sqrt{r} for a particle on the Norton dome.

theorem

The derivative of the delayed quartic rT(τ)r_T(\tau) is max⁡(τ−T,0)336\frac{\max(\tau - T, 0)^3}{36}

For any departure time T∈RT \in \mathbb{R} and any instant τ∈R\tau \in \mathbb{R}, the delayed quartic function rT(τ)=(max⁡(τ−T,0))4144r_T(\tau) = \frac{(\max(\tau - T, 0))^4}{144} is differentiable at τ\tau with derivative ddτrT(τ)=(max⁡(τ−T,0))336\frac{d}{d\tau} r_T(\tau) = \frac{(\max(\tau - T, 0))^3}{36}

theorem

The derivative of the delayed quartic equals max⁡(τ−T,0)336\frac{\max(\tau - T, 0)^3}{36}

For a given departure time T∈RT \in \mathbb{R}, let the delayed quartic function be defined as fT(τ)=max⁡(τ−T,0)4144f_T(\tau) = \frac{\max(\tau - T, 0)^4}{144}. The derivative of this function with respect to time τ\tau is given by: ddτfT(τ)=(max⁡(τ−T,0))336\frac{d}{d\tau} f_T(\tau) = \frac{(\max(\tau - T, 0))^3}{36} This represents the velocity of a particle on the Norton dome that begins to slide off the apex at time TT.

theorem

The second derivative of the delayed quartic function is max⁡(τ−T,0)212\frac{\max(\tau - T, 0)^2}{12}

For any departure time T∈RT \in \mathbb{R} and any instant τ∈R\tau \in \mathbb{R}, the second derivative of the delayed quartic function fT(τ)=max⁡(τ−T,0)4144f_T(\tau) = \frac{\max(\tau - T, 0)^4}{144} exists and is given by: d2dτ2(max⁡(τ−T,0)4144)=max⁡(τ−T,0)212\frac{d^2}{d\tau^2} \left( \frac{\max(\tau - T, 0)^4}{144} \right) = \frac{\max(\tau - T, 0)^2}{12}

theorem

The Delayed Quartic Function is C2C^2

For any T∈RT \in \mathbb{R}, the delayed quartic function fT(τ)=max⁡(τ−T,0)4144f_T(\tau) = \frac{\max(\tau - T, 0)^4}{144} is of class C2C^2 (twice continuously differentiable) on R\mathbb{R}.

definition

Motion on the Norton dome leaving the apex at time TT

For a given departure time T∈RT \in \mathbb{R}, the function maps an instant tt to the position of a particle on the Norton dome in a one-dimensional Euclidean space R1\mathbb{R}^1. The position is given by the vector: rT(t)=(max⁡(t−T,0))4144e0r_T(t) = \frac{(\max(t - T, 0))^4}{144} \mathbf{e}_0 where e0\mathbf{e}_0 is the unit basis vector of R1\mathbb{R}^1. This trajectory describes a particle that remains at rest at the apex (arc length 00) until time TT, and subsequently slides away such that its arc length follows the quartic path (t−T)4144\frac{(t - T)^4}{144}.

theorem

The arc length of the Norton dome solution equals the delayed quartic

For any departure time T∈RT \in \mathbb{R} and any instant tt, the arc length of the motion leaving the apex of the Norton dome at time TT (represented as the 00-th component of the position vector rT(t)r_T(t)) is equal to the delayed quartic function evaluated at tt: (rT(t))0=(max⁡(t−T,0))4144(r_T(t))_0 = \frac{(\max(t - T, 0))^4}{144}

theorem

The Norton dome solution rT(t)r_T(t) is at the apex for t≤Tt \le T

Let rT(t)r_T(t) be the trajectory of a particle on the Norton dome that leaves the apex at time T∈RT \in \mathbb{R}. For any time tt such that t≤Tt \le T, the particle remains at the apex, that is, rT(t)=0r_T(t) = 0.

theorem

For T≥0T \ge 0, the motion rT(t)r_T(t) starts at the apex

Let rT(t)r_T(t) denote the trajectory of a particle on the Norton dome that is scheduled to leave the apex at time T∈RT \in \mathbb{R}. For any departure time T≥0T \ge 0, the position of the particle at the initial time t=0t = 0 is at the apex, i.e., rT(0)=0r_T(0) = 0.

theorem

The particle on the Norton dome is off the apex for t>Tt > T

For any departure time T∈RT \in \mathbb{R} and any time t>Tt > T, the position of the particle on the Norton dome rT(t)r_T(t) has a strictly positive arc-length coordinate, i.e., rT(t)0>0r_T(t)_0 > 0.

theorem

The Norton dome motion rTr_T is C2C^2

For any departure time T∈RT \in \mathbb{R}, the motion rT:Time→R1r_T: \text{Time} \to \mathbb{R}^1 leaving the apex of the Norton dome at time TT is twice continuously differentiable (i.e., it is a C2C^2 function).

theorem

The velocity of the Norton dome motion rTr_T is (max⁡(t−T,0))336e0\frac{(\max(t - T, 0))^3}{36} \mathbf{e}_0

For any departure time T∈RT \in \mathbb{R}, the velocity of the motion rT(t)r_T(t) leaving the apex of the Norton dome at time TT is given by the time derivative: ddtrT(t)=(max⁡(t−T,0))336e0\frac{d}{dt} r_T(t) = \frac{(\max(t - T, 0))^3}{36} \mathbf{e}_0 where e0\mathbf{e}_0 is the standard unit basis vector in the one-dimensional Euclidean space R1\mathbb{R}^1.

theorem

d2dt2rT(t)=(max⁡(t−T,0))212e0\frac{d^2}{dt^2} r_T(t) = \frac{(\max(t - T, 0))^2}{12} \mathbf{e}_0

For any departure time T∈RT \in \mathbb{R}, the second time derivative (acceleration) of the motion rT(t)r_T(t) on the Norton dome is given by d2dt2rT(t)=(max⁡(t−T,0))212e0\frac{d^2}{dt^2} r_T(t) = \frac{(\max(t - T, 0))^2}{12} \mathbf{e}_0 where tt is the time and e0\mathbf{e}_0 is the unit basis vector in the one-dimensional Euclidean space R1\mathbb{R}^1.

theorem

Initial velocity of the Norton dome motion is zero for T≥0T \ge 0

For any departure time T≥0T \ge 0, the motion rT(t)r_T(t) on the Norton dome, which describes a particle leaving the apex at instant TT, has a velocity of zero at the initial time t=0t = 0. That is, r˙T(0)=0\dot{r}_T(0) = 0.

theorem

The motion rT(t)r_T(t) satisfies the Norton dome equation of motion

For any departure time T∈RT \in \mathbb{R}, the motion rT(t)r_T(t) defined as the trajectory leaving the apex of the Norton dome at instant TT satisfies the equation of motion r¨=r\ddot{r} = \sqrt{r}. Specifically, its acceleration d2dt2rT(t)=(max⁡(t−T,0))212\frac{d^2}{dt^2} r_T(t) = \frac{(\max(t - T, 0))^2}{12} is equal to the square root of its arc length rT(t)\sqrt{r_T(t)}.

theorem

The motion leaving the apex at time TT is a solution for the Norton dome system.

For any real number TT, the trajectory rT(t)r_T(t) defined by rT(t)=(max⁡(t−T,0))4144e0r_T(t) = \frac{(\max(t - T, 0))^4}{144} \mathbf{e}_0 (where e0\mathbf{e}_0 is the unit basis vector in R1\mathbb{R}^1) is a solution to the equation of motion for the Norton dome. This represents a particle that remains at rest at the apex until time TT and then begins to slide down the dome.

theorem

T<T′  ⟹  rT≠rT′T < T' \implies r_T \neq r_{T'} for Norton dome motions

Let rTr_T denote the trajectory of a particle on the Norton dome that leaves the apex at time TT, defined as rT(t)=(max⁡(t−T,0))4144e0r_T(t) = \frac{(\max(t - T, 0))^4}{144} \mathbf{e}_0 where e0\mathbf{e}_0 is the unit basis vector. For any two departure times T,T′∈RT, T' \in \mathbb{R} such that T<T′T < T', the corresponding trajectories rTr_T and rT′r_{T'} are distinct.

theorem

Injectivity of the mapping from departure time to trajectory on the Norton dome

Let rTr_T denote the trajectory of a particle on the Norton dome that remains at the apex until time T∈RT \in \mathbb{R} and then slides off, defined by the function: rT(t)=(max⁡(t−T,0))4144e0r_T(t) = \frac{(\max(t - T, 0))^4}{144} \mathbf{e}_0 where e0\mathbf{e}_0 is the unit basis vector in R1\mathbb{R}^1. The mapping from the departure time TT to the trajectory rTr_T is injective; that is, distinct departure times T1≠T2T_1 \neq T_2 result in distinct physical motions rT1≠rT2r_{T_1} \neq r_{T_2}.

theorem

rT≠0r_T \neq 0 for the Norton dome solutions

For any departure time T∈RT \in \mathbb{R}, the motion rTr_T representing a particle that leaves the apex of the Norton dome at time TT is not identical to the zero motion (the constant trajectory 00 where the particle remains at rest at the apex).

theorem

Rest at the Apex is a Solution for the Norton Dome

The constant trajectory r(t)=0r(t) = 0, representing a particle remaining at rest at the apex of the Norton dome, satisfies the equation of motion for the system SS.

theorem

The energy of staying at the apex is zero

For the Norton dome system, the motion that consists of the particle remaining at rest at the apex (represented by the constant zero function) has total energy equal to 00 for all times tt.

theorem

Failure of uniqueness for the Norton dome

There exist two trajectories x,y:Time→R1x, y : \text{Time} \to \mathbb{R}^1 that are solutions to the equations of motion for the Norton dome system SS such that they have the same initial position x(0)=y(0)x(0) = y(0) and the same initial velocity x˙(0)=y˙(0)\dot{x}(0) = \dot{y}(0), but are distinct trajectories (x≠yx \neq y).

theorem

Non-uniqueness of Solutions for the Norton Dome

In the Norton dome system SS, the uniqueness of solutions given initial conditions does not hold. Specifically, it is not the case that for any two solutions x,y:Time→R1x, y : \text{Time} \to \mathbb{R}^1, the conditions x(0)=y(0)x(0) = y(0) and ∂tx(0)=∂ty(0)\partial_t x(0) = \partial_t y(0) imply that x=yx = y.

theorem

Existence of solutions for Norton's dome with arbitrary departure time T≥0T \geq 0

For any real number T≥0T \geq 0, there exists a trajectory r:Time→Rr : \text{Time} \to \mathbb{R} that is a solution to the Norton dome's equation of motion such that the particle starts from rest at the apex at time t=0t = 0 (i.e., r(0)=0r(0) = 0 and its time derivative r˙(0)=0\dot{r}(0) = 0). This trajectory has the property that the particle remains at the apex for all times t≤Tt \leq T (satisfying r(t)=0r(t) = 0) and moves away from the apex for all times t>Tt > T (satisfying r(t)>0r(t) > 0).

theorem

The Total Energy of the Norton Dome Motion rTr_T is Zero

For any departure time T∈RT \in \mathbb{R} and any instant tt, the total energy of the motion rTr_T (the trajectory where a particle leaves the apex of the Norton dome at time TT) is equal to 00.

theorem

The second derivative of f(−t)f(-t) is f¨(−t)\ddot{f}(-t)

Let MM be a normed real vector space and f:Time→Mf : \text{Time} \to M be a function representing a curve. If ff is differentiable and its time derivative ∂tf\partial_t f is also differentiable (i.e., ff is twice differentiable), then for any t∈Timet \in \text{Time}, the second time derivative of the time-reversed curve s↦f(−s)s \mapsto f(-s) evaluated at tt is equal to the second time derivative of ff evaluated at −t-t. In mathematical notation: ddt(ddtf(−t))=d2fdτ2∣τ=−t\frac{d}{dt} \left( \frac{d}{dt} f(-t) \right) = \left. \frac{d^2 f}{d\tau^2} \right|_{\tau = -t} where the left-hand side represents the second derivative of the composition of ff with the negation operation.

theorem

Time reversal t↦r(−t)t \mapsto r(-t) satisfies the Norton dome equation of motion

Let SS be a Norton dome system and r:Time→R1r: \text{Time} \to \mathbb{R}^1 be a trajectory. If rr is twice differentiable (specifically, rr and its time derivative ∂tr\partial_t r are differentiable) and satisfies the equation of motion for SS, then the time-reversed trajectory t↦r(−t)t \mapsto r(-t) also satisfies the equation of motion. This symmetry arises because the equation of motion for the Norton dome has no velocity-dependent terms.

theorem

Time reversal t↦r(−t)t \mapsto r(-t) of a Norton dome solution is a solution

Let SS be a Norton dome system. If a motion r:Time→R1r: \text{Time} \to \mathbb{R}^1 is a solution to the equations of motion for SS, then the time-reversed motion defined by t↦r(−t)t \mapsto r(-t) is also a solution for SS.

definition

Motion on the Norton dome arriving at the apex at time −T-T

For a given parameter T∈RT \in \mathbb{R}, this function describes the trajectory of a particle on the Norton dome that arrives at the apex at time −T-T. It is defined as the time reversal of the motion leaving the apex at time TT. The position at time tt in the one-dimensional Euclidean space R1\mathbb{R}^1 is given by: arrival(T)(t)=(max⁡(−t−T,0))4144e0\text{arrival}(T)(t) = \frac{(\max(-t - T, 0))^4}{144} \mathbf{e}_0 where e0\mathbf{e}_0 is the unit basis vector. This motion represents a particle sliding up the dome and coming to a permanent rest at the apex at the instant t=−Tt = -T.

theorem

The motion arriving at the apex at time −T-T is a solution of the Norton dome equations.

For any real number TT, the trajectory r:Time→R1r: \text{Time} \to \mathbb{R}^1 that arrives at the apex of the Norton dome at time −T-T, defined as r(t)=(max⁡(−t−T,0))4144e0,r(t) = \frac{(\max(-t - T, 0))^4}{144} \mathbf{e}_0, is a solution to the equations of motion for the Norton dome system SS.

theorem

The motion arriving at the apex at −T-T is off the apex for t<−Tt < -T

For any T∈RT \in \mathbb{R}, let arrival(T)\text{arrival}(T) denote the trajectory of a particle on the Norton dome that arrives at the apex at time −T-T. For any time t<−Tt < -T, the particle is off the apex, meaning its radial position coordinate satisfies (arrival(T)(t))0>0(\text{arrival}(T)(t))_0 > 0.

theorem

arrival(T)(t)=0\text{arrival}(T)(t) = 0 for t≥−Tt \geq -T

For any real parameter TT and any time t≥−Tt \geq -T, the position of the particle in the motion arrival(T)\text{arrival}(T) is 00. This means that a particle arriving at the apex of the Norton dome at the instant −T-T remains at rest at the apex for all subsequent times.

theorem

The velocity of arrival(T)\text{arrival}(T) is zero for t≥−Tt \geq -T

For any real parameter TT, let arrival(T)\text{arrival}(T) denote the trajectory of a particle on the Norton dome that arrives at the apex at time −T-T. For all times t≥−Tt \geq -T, the velocity of the particle (the time derivative of its position) is zero: ddt(arrival(T))(t)=0\frac{d}{dt}(\text{arrival}(T))(t) = 0 This implies that a particle arriving at the apex at time −T-T does so with zero velocity and remains at rest for all subsequent times.