Astronomy & Celestial Mechanics
the sky, computed
Concept

Regularisation — where it appears

A change of variables that removes the singularity of a two-body collision orbit. It turns the equations into those of a harmonic oscillator, so a step uniform in the new variable is exactly the step the trajectory needs near perihelion, with no adaptive control at all.

Named by 5 essays across 4 fields — each of them below, with the objects they name alongside it.

One squaring, and the singularity is gone. Left, a harmonic oscillator: an ellipse centred on the origin, marked at 24 equal steps of its own phase. Right, the same points after squaring as complex numbers, u ↦ u². The image is an ellipse with the origin at a focus — checked here by the focal property, r₁ + r₂ = 2a to nine figures at four points — with semi-axes (A²+B²)/2 = 2.5000 and AB = 1.5000 and the focus at (A²−B²)/2 = 2.0000, which is ae exactly. Three things follow at once. The angle doubles, so one turn of the oscillator is one whole orbit traversed twice as fast in phase; the equal phase steps on the left arrive as the eccentric anomaly on the right, which is why that anomaly and not the true one is what the equations want; and the collision at r = 0, where the inverse square is infinite, is the point u = 0, where the oscillator has a perfectly ordinary velocity. The singularity was a property of the coordinates.

The singularity that is a change of variable

The Kepler problem blows up at zero separation, and a fixed-step integrator falls apart long before it gets there. Divide time by the radius and the equations become a harmonic oscillator — exactly, for every conic at once.

orbits · Universal variables The adaptive step starts 5.7× more accurate and ends 7.5× worse. The envelope of the relative energy error against elapsed revolutions — the worst error within each plotted interval rather than the error at one sample in it — for the same second-order symplectic integrator run two ways on a Kepler orbit at e = 0.5. The flat band is a fixed step of one 200th of a period: its error oscillates once per revolution and the envelope does not grow, because a symplectic method at constant h is the exact solution of a nearby Hamiltonian and is conserving that one. The rising curve is the same method with the step refined where the orbit is fast — h ∝ r^3/2, the free-fall time, varying by a factor of 5 around the orbit — which is the first thing anybody reaches for at a close encounter and which is strictly more accurate step for step: over its first three revolutions it stays a factor of 5.7 below the fixed run. By 1500 revolutions it is a factor of 7.5 above it and still climbing. Changing the step changes which Hamiltonian is being conserved, the errors from successive steps stop cancelling, and what is left is a random walk with no bound at all. The practical consequence is that a solar-system integration cannot adapt its step: it either keeps a step short enough for the closest encounter it will ever meet, or it detects the encounter and hands that piece of the trajectory to an entirely different, non-symplectic method for the duration — which is what every long-term integration of the planets actually does.

A step that must not be adapted

A symplectic integrator's bounded energy error is a property of a fixed step. It is conserving a Hamiltonian a step-size away from the intended one, and changing the step changes which Hamiltonian — so refining the step at a close encounter, which is the first thing anybody does, destroys the only property the method was chosen for.

gravitation · Numerical integration Where each mode turns back: ℓ = 0 through the centre, ℓ = 300 in the outer 2 per cent. Why a set of frequencies is a depth profile and a single frequency is not. An acoustic wave travelling into a star meets a rising sound speed and is refracted back; it turns where its horizontal phase speed matches the local sound speed, which happens at c(r)/r = 2πν/√(ℓ(ℓ+1)). The horizontal axis is the angular degree on a logarithmic scale and the vertical axis is the fractional radius of that turning point, drawn at 2000, 3090, 4000 microhertz. The ordering is the content. A radial mode, ℓ = 0, has no horizontal phase speed at all and passes straight through the centre. Degrees one and two turn deep in the core. By ℓ = 300 the mode is trapped in the outer 2 per cent and knows nothing about anything below. So a frequency measured to a part in ten thousand constrains an average of the interior weighted in a way the mode itself decides, and measuring thousands of modes of different degree gives thousands of differently weighted averages — which is a solvable inverse problem, and is how the base of the convection zone was located at 0.713 of the radius rather than assumed. The sound speed here is a polytrope's rather than a tabulated solar model's, so the curve is the right shape and the wrong star in its outer tenth, where the real Sun is convective and this one is not. What the picture cannot show is the frequency dependence at fixed degree, which is weaker but not negligible: a higher-frequency mode of the same degree turns slightly deeper, and the three curves separating toward the right is that effect.

Every note turns back at its own depth

A sound wave heading into a star is refracted by the rising sound speed and turns around before it reaches the middle. Where it turns depends on which mode it is — so a list of frequencies is not one average of the interior but thousands of differently weighted ones, and that is a solvable problem.

stars · Asteroseismology Three pictures, one set of measurements. A one-dimensional sky, the transform of it sampled at 10 spatial frequencies out of 128, and three images reconstructed from those samples. Every one of the three reproduces every measured visibility exactly — checked to a part in a million — so the data cannot choose between them. The dirty image is the samples transformed back and nothing else, complete with the negative sidelobes that incomplete sampling produces. The other two are the members of the same family that minimise a penalty: one the total curvature, one the total absolute brightness. They differ from each other by 7 per cent of the peak, and where they differ is exactly where the array did not look. The smooth reconstruction blurs the two compact sources into one another and keeps the broad component; the sparse one splits them cleanly and loses the broad one. Neither is a lie and neither is the sky. A published image is the data plus a criterion, and the criterion is a choice made by whoever ran the reduction.

How much of the picture is the prior

An array measures the sky's transform at a few dozen spatial frequencies and at no others, so the data do not determine an image. They determine a family of them, and the reconstruction picks one member by a criterion that is not in the data — smoothness, or sparsity, or entropy. Three reconstructions of one measurement fit it exactly and disagree by a tenth of the peak.

starlight · Interferometry A feature read back from the limb. A bump in the source function at optical depth 0.1, a quarter of a decade wide (black dashed), and what an inversion of the limb profile recovers on average when the profile is measured at 24 angles with errors of 1, 0.1, 0.01, 0.001 per cent, the inversion assuming only that the departure from a grey atmosphere is smooth. At one per cent the recovered bump has 56 per cent of its height and is spread over about twice its width; at a thousandth of a per cent, 82 per cent, and still wider than the truth. A thousandfold gain in precision buys a modest gain in resolution — each factor of ten a step, not a factor.

A profile that averages a decade of depth

Each point on a star's disc reports the temperature at a different depth, so a limb-darkening profile looks like a depth sounding of the atmosphere waiting to be read. It is one, but a blurred one. Every angle averages a whole decade of optical depth, the averages overlap, and recovering the temperature structure from them is the inversion of a Laplace transform — a problem whose resolution grows only as the logarithm of the data's precision.

starlight · Limb darkening

Named alongside it

The objects these essays reach for when they reach for this one.

ConditioningEnergy driftAcoustic cavityAdaptive stepAsteroseismologyAveraging kernelClose encounterClosure phaseCollision singularityConvectionDeconvolutionDegeneracy

All concepts