The Planet in the Star's Glare: How Radial Fourier Transforms Help Us See Other Worlds

Research Notes — adapted from Albanna, Doelman & Mumtahana (2019), “Radial Fourier transforms in exoplanetary imaging and potential uses at Timau National Observatory,” J. Phys.: Conf. Ser. 1231 012014 — DOI: 10.1088/1742-6596/1231/1/012014.

We have found thousands of planets around other stars, and we have photographed almost none of them.

That sentence sounds like a contradiction, and in a sense it is. Nearly every exoplanet we know about was found indirectly — by watching a star wobble, dim, or shift, and inferring that something invisible was tugging on it. Only a tiny handful have ever been captured as actual light, pixel by pixel, in an image. The rest are shadows of evidence, not pictures.

This note is about the hardest and most rewarding branch of exoplanet science — direct imaging — and about a piece of mathematics that makes it tractable: the Radial Fourier Transform.

The One Picture We Cannot Take

The difficulty is not that exoplanets are faint in some ordinary sense. It is that they are faint next to something blinding.

A typical giant planet might be a million times dimmer than its star — some are ten billion times dimmer. And at the distance of a nearby star, that planet sits only a fraction of an arcsecond from the glare: roughly the width of a coin seen from several kilometres away. The starlight does not politely stay in one spot. It spreads into a bright halo that washes over exactly the region where the planet hides.

The result, in the numbers of the field, is stark. Of the thousands of exoplanet detections to date, direct imaging has managed only a few percent — fewer than a hundred out of more than three and a half thousand. Every other method is cheaper, easier, and less demanding of the telescope.

So why bother? Because indirect methods give you numbers, and direct imaging gives you a photograph — a spectrum of the planet’s own light, its atmosphere, its clouds, its temperature, its orbit. It is the difference between knowing a room exists and being able to look inside it.

To take that photograph, we need to make the star’s glare disappear while the planet’s light survives. That is a job for optics, and behind the optics is a Fourier transform.

A Telescope Is a Fourier Transform

The place to start is a principle from the 17th century. The Huygens–Fresnel principle says that every point on a wavefront acts as a source of secondary spherical wavelets, and the field at any later point is the sum of all of them.

Apply that to a telescope and something remarkable falls out. The light collected at the pupil plane (the aperture) and the light focused at the focal plane (the image) are related by a two-dimensional Fourier transform:

F(u, v) = ∬ f(x, y) e^(−i2π(ux + vy)) dx dy

Read that literally: the image a telescope forms is the Fourier transform of the light entering its aperture. Focus light through a circular hole and the transform hands you back the pattern that hole produces.

This is more than a curiosity. It means that once we normalize the constants away — setting the wavelength and distance to unity, and measuring lengths in units of wavelength — the electric field in the pupil and the field in the image are a Fourier pair, exchangeable without any loss of physical meaning. From that point on, the problem becomes mathematics.

And mathematics is exactly what we need, because “shape the pupil so that the starlight cancels in a ring around the star” is a problem in inverse design — we choose an input, and the transform tells us the output.

Why Circles Simplify Everything

Most telescope apertures are circular. So are most of the things we want to do to them. That symmetry is a gift.

When working with circles, polar coordinates are the natural language. We rewrite positions as:

x = r cos θ,   y = r sin θ,   u = ρ cos ψ,   v = ρ sin ψ

and something important becomes plain: Fourier transforms preserve symmetry. A circularly symmetric pupil cannot produce anything other than a circularly symmetric image. So if we want to suppress starlight across all 360 degrees around a star, we should be looking for circularly symmetric patterns — and we can describe the whole system with a single radial coordinate.

When a pupil function depends only on radius, f(r, θ) = f(r), the angular part of the transform collapses. Every term in the angular Fourier series vanishes except one — the n = 0 term — and the general two-dimensional transform reduces to a single radial integral:

F(ρ) = 2π ∫₀^∞ r f(r) J₀(2πρr) dr

This is the Radial Fourier Transform (RFT), with J₀ the Bessel function of the first kind of order zero. The Bessel function plays the role that a sine wave plays in the ordinary one-dimensional transform: it is the natural oscillation of a circularly symmetric system.

The practical payoff is large. A two-dimensional design problem has been cut down to one dimension, while the pattern itself is still a full two-dimensional disc. For coronagraph design, that is the difference between a tractable search and an intractable one.

There is a useful caution here. We restrict ourselves to the radial mode because it is simpler and because circular symmetry is exactly what a 360-degree dark zone requires. But real telescopes have secondary-mirror spiders — the struts that hold the secondary mirror — which break that symmetry. The full polar-coordinate machinery, with its azimuthal modes, is waiting for those cases.

The Airy Pattern Is the Enemy

With the RFT in hand, the ordinary behaviour of a telescope becomes easy to derive. Put a plain circular aperture — a top-hat that is 1 inside some radius a and 0 outside — into the integral, use a standard Bessel identity, and out comes the famous Airy pattern:

F(ρ) = a · J₁(2πρa) / ρ

That is the bright central disc with concentric rings fading outward that every telescope produces. It looks like a damped sinusoid. If several sources are present, each produces its own scaled copy of the same pattern — which is exactly why a faint planet is lost. Its own faint Airy pattern is buried under the enormous one from the star.

Worse, the geometry is unkind. Typical hot Jupiters orbit at only about 2 to 5 λ/D from their star — right inside the second or third Airy ring. The planet is not merely dim; it sits in the brightest part of the noise. Direct imaging without extra optics is, in the paper’s phrase, extremely difficult.

Apodization: Sculpting the Pupil

The fix is to stop accepting the aperture as given and modify the input so that the transform changes. This is called apodization — literally, “removing the foot,” from the idea of trimming the base of a profile.

There are two ways to do it, and they are mathematically cousins:

f(r) = A(r)                     amplitude-apodization,
f(r) = A(r) · e^(iφ(r))         phase-apodization.

• In amplitude-apodization, the aperture is combined with a mask that dims some regions and passes others — you reshape the brightness of the pupil.
• In phase-apodization, the aperture shape is kept and a phase-inducing optic φ(r) is added — you reshape the timing of the light, so that some parts arrive slightly out of step and cancel in the dark zone.

The first phase optics were diamond-turned zinc selenide plates with carefully varied height; today, liquid-crystal devices can imprint the needed phase pattern electronically. But the manufacturing is not the point. Both approaches are solving the same design problem, and the goal is stated with unusual clarity:

1. Suppress the Airy rings inside a chosen working angle — drive the halo down to a contrast of about 10⁻⁶ or better.
2. Keep the central brightness high — do not throw away the starlight’s peak, or you lose the ability to locate and study the star itself.

Written as mathematics, the first requirement is a minimization of the mean-squared halo over a set of radii, and the second is a maximization of the on-axis intensity relative to the unmodified aperture. Together, those two conditions are normally enough to produce a working coronagraph design — a pupil pattern that carves a dark hole around the star while leaving the star’s core intact.

A worked design in the paper, for example, opens a clean dark region over the working angle (ρᵢ, ρ_f) = (3, 5) λ/D — that is, it clears the second and third Airy rings exactly where hot Jupiters tend to live.

What Timau Could Do

A mathematical design is only interesting if it can be built and pointed at the sky. So the paper turns to a concrete target: the future 3.8-metre optical telescope at Timau National Observatory in Indonesia.

Direct imaging needs three things at once: extremely high contrast, very high angular resolution, and a field of view that actually contains the working angle. Coupling a RFT-generated phase-apodizing coronagraph to the Timau design, the simulation reports:

Requirement	Simulated performance (700 nm)
Contrast	10⁻⁶ (typical exoplanet brightness is 10⁻⁵–10⁻¹⁰)
Angular resolution	0.04 arcsec
Inner working angle	3.1 λ/D ≈ 0.11 arcsec

Put those numbers next to a real target. A “twin Jupiter” orbiting a star at 10 parsecs would appear about 0.5 arcsec from its host — comfortably outside the 0.11 arcsec inner working angle, and therefore inside the region the instrument can actually search.

That is the headline result: a 3.8-metre telescope in Indonesia, equipped with a phase-apodizing coronagraph designed through radial Fourier transforms, could in principle directly image Jupiter-sized planets at 10 parsecs.

An Honest Ledger

It is worth being precise about what that conclusion is and is not.

• It is a design and simulation study. The result says the configuration can reach the needed contrast and resolution — not that any planet has been detected. The paper is explicit that practical and manufacturing questions are left to a different discussion.
• The physics is exact, but the instrument is not yet built. Timau itself is still on the horizon. The simulations describe what a well-designed coronagraph would do, assuming the optic can be made faithfully.
• Direct imaging will never replace the indirect methods. Radial velocity and transits found the overwhelming majority of known planets because they are efficient. Direct imaging is the specialist’s tool: rarer, harder, and uniquely able to return a photograph.

None of that weakens the point. It sharpens it. Seeing a planet is not a matter of building a bigger mirror. It is a matter of shaping light with mathematics precisely enough that the star’s glare steps aside — letting a firefly, for a moment, outshine the lighthouse.

Why This Matters Beyond the Telescope

There is a larger pattern in this work, and it is the same one this whole site keeps returning to: an instrument is downstream of an idea.

Before anyone grinds a phase plate or grows a liquid-crystal optic, someone has to understand what pattern of light the aperture should contain. That understanding comes from centuries of accumulated mathematics — Huygens’ wavelets, Fourier’s transforms, Bessel’s functions, and the observation that a circular pupil and its image are two views of one object. The telescope is the physical outcome of an abstraction that had to exist first.

So the Radial Fourier Transform is not a footnote to the search for other worlds. It is one of the reasons the search is possible at all. The next time a telescope points at a distant star, it will be carrying a hundred years of mathematics in its optics — and, perhaps, a picture of a planet we have never seen.

---

Reference. H. Albanna, D. Doelman, F. Mumtahana, “Radial Fourier transforms in exoplanetary imaging and potential uses at Timau National Observatory,” Journal of Physics: Conference Series 1231, 012014 (2019). DOI: 10.1088/1742-6596/1231/1/012014.