Astronomy & Celestial Mechanics
the sky, computed
Concept

Periodogram — where it appears

A measure of how much of a time series' variance a sinusoid at each trial period explains. Its tallest peak is where a period search starts, and noise alone produces peaks, so a peak's height means nothing until a false-alarm probability is attached to it.

Named by 7 essays across 3 fields — each of them below, with the objects they name alongside it.

A peak worth 10.8 in a narrow search is worth nothing in a wide one. The probability that noise alone produces a peak at least as tall as a given power, for searches over four different numbers of independent frequencies. A single frequency examined in isolation gives a one-per-cent chance at a power of 4.6; searching fifty thousand frequencies for the same one-per-cent chance requires 15.4. The threshold rises as the logarithm of the width of the search, which is why the penalty is survivable — but it is a penalty, it is often not applied, and the number of independent frequencies in an unevenly sampled time series is not the number of frequencies on the grid. Overestimating that count is conservative and underestimating it is not, which is the one asymmetry worth remembering.

The tallest peak in nothing at all

A periodogram of pure noise has peaks in it, and the tallest is not small. How tall it has to be before it means something depends on how many frequencies were searched and on what the noise actually is — and astronomical noise is almost never the white noise the standard formula assumes.

starlight · Periodograms One eccentric planet, drawn as two circular ones. Top: the velocity of a star with one companion on an orbit of eccentricity 0.25 (argument of periastron 0.9 radians), in units of its semi-amplitude K, over two orbits — and on top of it the sum of two circular orbits, one at the same period and one at half of it, fitted by taking the first two terms of the exact curve's Fourier series. The second orbit's amplitude comes out at 0.246 of the first, which is the eccentricity to first order. Bottom: the difference between the two, on a scale 12 times magnified; its largest excursion is 0.0889 K and its rms 0.0473 K, second order in the eccentricity. For a planet with K = 30 m/s that is 1.42 m/s rms — below the noise of most instruments — so an eccentric single planet and a pair of planets in a 2:1 period ratio are the same measurement unless the data are good enough to see a term of order e².

One eccentric planet, or two circular ones

The velocity curve of a star with one planet on an eccentric orbit is, to first order in the eccentricity, exactly the curve of two planets on circular orbits with periods in the ratio two to one. The difference between the two readings is a term in the eccentricity squared, and for the modest eccentricities most planets have, it is smaller than the noise.

exoplanets · Reflex velocity The evidence for a period, one trial period at a time. The natural logarithm of the Bayes factor for a sinusoid at each trial period against no signal, for 120 radial velocities spread over 359 days with 2.0 m/s white noise per point and a planet of semi-amplitude 1.0 m/s at 11.3 days. The amplitude prior is Gaussian with scale 3.0 m/s in each component. At the planet's period the evidence is 14.4 — odds of about 1.8·10⁶ to one — and almost everywhere else it is near zero or negative, because a sinusoid that fits nothing is penalised for the amplitude range it wasted. The quantity that answers the question "is there a planet at some period" is not the peak but the average of the Bayes factor over the prior on the period, here uniform in the logarithm of the period from 1.5 to 1,000 days, and that average is ln B = 7.7: 6.7 lower than the peak, the Bayesian form of the look-elsewhere penalty, paid because the data had to find the period as well as the amplitude.

Odds for a planet that depend on a guess about its size

A false-alarm probability asks how often noise alone would produce a peak this tall. A Bayes factor asks something better — how much more probable the data are with a planet than without one — and pays for it with two numbers nobody measured. The width allowed for the period barely matters. The width allowed for the amplitude can turn strong evidence into none, from the same velocities.

starlight · Periodograms Amplitudes measured near the noise, against the amplitudes that were there. A simulated survey of 6000 stars, each with one planet whose true radial-velocity semi-amplitude K is drawn from a population rising towards small amplitudes (dN/d ln K ∝ K^−0.6, from 0.2 to 20 m/s), each measured with an amplitude uncertainty of 0.6 m/s; 1500 are drawn. Across: the true amplitude; up: the amplitude fitted at the planet's period; both logarithmic. The dashed diagonal is a perfect measurement; the horizontal line is the detection threshold, 4 times the uncertainty. Points above it are the detections (filled). Well above the threshold they sit on the diagonal; near it they sit above it — the detections within one unit of the threshold have measured amplitudes 1.33 times their true ones on average — and below it the measured amplitudes stop falling at all, because an amplitude is a length and noise alone gives it about 0.75 m/s.

A prior the survey measures through its own noise

Whether a weak periodic signal is a planet depends on how large planets are expected to be, and the best source for that expectation is the population the surveys themselves have measured. But the population's small end is built from the marginal detections whose reality depends on the prior. Breaking the circle means measuring the population and the planets together, with the noise and the threshold written into the model — and doing it that way changes the answers by factors of two where they matter most.

starlight · Periodograms The summer sunshine at Mars's north pole over the last million years. The sunshine reaching Mars's north pole at the summer solstice, relative to its average, over the last 1,000 thousand years (blue), from a forcing built from the main periods of Mars's recent orbit: the tilt varying by about 3.5 degrees in 120,000 years and 1 degree in 99,000, the eccentricity by 0.03 in 95,000 years and 0.02 in 2.4 million, and the perihelion going round the seasons once in 51,000 years. The part due to the tilt alone is dashed. The sunshine swings by a factor of 2.06 between its extremes: the tilt sets a slow swell with a period of 120,000 years, and the precession, whose strength the eccentricity sets, cuts a faster ripple of 51,000 years into it, strongest when the orbit is most eccentric. These two periods, and their ratio of 2.35, are what a record of the sunshine should carry.

Layers that keep the rhythm and lose the dates

Mars's polar caps are stacks of ice and dust laid down while the planet's tilt and orbit changed the summer sunshine, and the question is whether they can be read as a record of that sunshine. They can — for the rhythm. Even when the ice accumulates faster under weak sunshine than under strong, the cycles of the tilt and the precession survive in the depth of the stack in their correct proportion. What does not survive is the date of any one layer.

sky · Obliquity The same data searched with a box and with a sine. The signal-to-noise ratio of the best fit at each trial period, for the light curve of the previous figure, searched two ways: with a box of three trial durations at every phase, and with a sinusoid. The box search peaks at 3.85 days with a signal-to-noise ratio of 7.3, against a true period of 3.85. The sinusoid's best peak anywhere is 3.3, at 1.59 days, and at the true period it reaches only 1.5 — no higher than the peaks the noise makes elsewhere in the search. The planet is a clear detection to one method and invisible to the other, in the same numbers.

A periodogram shaped like a shadow

A periodogram that fits sinusoids sees a transiting planet only through the fundamental harmonic of its box-shaped dip, which for a short transit is a few per cent of the signal. Folding the data at each trial period and fitting a box recovers all of it — five times the signal-to-noise for a transit two per cent of the orbit long, eighteen times for an Earth crossing the Sun. The box pays for its extra parameters in trials, and the price is about one unit of signal-to-noise.

starlight · Periodograms Two transits two years apart, and the periods left. Two single transits of the same 6-hour duration 730 days apart allow only periods of 730/n days. Each allowed period is weighted by how well it fits the duration, with the ingress measured and the eccentricity unknown. Of the first 120 aliases, the three most probable are n = 29 (25.2 days, 4 per cent), n = 27 (27.0 days, 4 per cent), n = 30 (24.3 days, 4 per cent). The duration turns a comb of a hundred and twenty candidates into a ranked list in which a few dozen hold almost all the probability.

A period read from a single dip

A planet whose year is longer than the observations transits once, and there is nothing to fold. Its period is not lost: the duration of the one transit, with the star's density, gives it through Kepler's third law — as the cube of the duration, divided by a factor that depends on where the planet crossed the star. That factor and the orbit's unseen eccentricity spread the answer over a factor of several, and a second transit years later replaces the spread with a comb of a hundred allowed periods that the first transit's duration can rank.

starlight · Periodograms

Named alongside it

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

False-alarm probabilityRadial-velocityBayes factorBox least squaresEccentricityLook elsewhere effectPriorRed noiseSignal-to-noise ratioStellar densityTransit photometryBootstrap

All concepts