Abel Inversion: Reconstructing Reality from Projections

How can we understand the internal structure of an object we cannot touch or slice open, like a distant star or a searingly hot plasma flame? This challenge, common across science and engineering, is known as an inverse problem: we see the effect—a 2D projection or "shadow"—and must deduce the underlying 3D cause. Often, nature provides a crucial clue in the form of symmetry. For countless objects that are axisymmetric (symmetric around a central axis), a powerful mathematical tool known as the Abel inversion allows us to perform this reconstruction from a single viewpoint.

This article provides a comprehensive overview of this elegant technique. It addresses the knowledge gap between observing a flattened image of a symmetric object and knowing its true internal structure. By reading, you will gain a deep understanding of how this mathematical "un-smearing" process works.

The following chapters will guide you through this topic. First, ​​"Principles and Mechanisms"​​ will demystify the mathematics behind the forward and inverse Abel transforms, explore how different physical structures appear in projection, and tackle the critical real-world challenges of noise and imperfect symmetry. Subsequently, ​​"Applications and Interdisciplinary Connections"​​ will take you on a journey through the vast scientific landscape where Abel inversion is indispensable, from probing planetary atmospheres and galaxy clusters to mapping chemical reactions and uncovering fundamental laws of physics.

Imagine you are looking at a beautiful, steady candle flame, a column of shimmering heated air rising from a hot road, or even a picture of a distant, spherical nebula. You cannot reach into the flame or the nebula to measure its properties. How could you possibly figure out its internal structure—say, the temperature or density at its very core—just by looking at it from the side? This is the kind of puzzle that physicists and engineers face all the time. It is a classic ​​inverse problem​​: we have the result of some process (the light that reaches our eyes or detectors) and we want to deduce the underlying cause (the structure of the object).

While in general this can be incredibly difficult, nature often gifts us with a simplifying symmetry. Many of these objects—flames, plasma columns, gas jets—are ​​axisymmetric​​. That is, they look the same if you rotate them around a central axis. A donut is axisymmetric, a car is not. This single, powerful assumption of symmetry is the key that unlocks the problem. Instead of needing a full 3D CT-scan like in a hospital, where pictures are taken from hundreds of angles, we only need to look at the object from one side. The mathematical magic that allows us to reconstruct the internal radial structure from a single 2D projection is called the ​​Abel inversion​​.

Before we can reverse the process, let's understand how it works in the forward direction. Suppose we have a cylindrically symmetric plasma cloud, and we know its emissivity $\epsilon(r)$, which is the amount of light it emits at a given radius $r$ from the center. Now, we place a detector far away to measure the brightness. What does the detector see?

A light ray traveling towards our detector at a certain "height" $y$ from the center (this is often called the ​​impact parameter​​) passes through many different rings of the plasma, like a skewer through a stack of onion rings. The total brightness $I(y)$ that the detector measures for that line of sight is the sum of the emissivity from every point along that path. This is a ​​line integral​​.

Because of the cylindrical symmetry, this line integral can be written in a very specific and elegant form. If the object has a maximum radius $R$, the measured brightness $I(y)$ is given by the ​​Abel transform​​:

This might look a little intimidating, but there’s a simple intuition behind it. The integral sums up the contributions from the emissivity $\epsilon(r)$ at all radii $r$ that the light ray at height $y$ passes through (from $r=y$ out to $r=R$). The peculiar-looking term $\frac{r}{\sqrt{r^2 - y^2}}$ is a purely geometric factor. It accounts for the fact that the light path spends a longer time passing through the larger, outer rings than it does skimming past the inner ones. This integral represents the "forward problem": if you know the radial profile $\epsilon(r)$, you can predict the measurement $I(y)$. This same formula appears whether you are measuring light emission from a plasma, phase shift of a laser through a gas, or the absorption of a beam through a flame.

Now for the real trick. We don't have $\epsilon(r)$; we have the measured data $I(y)$. We need to run the machine backwards. Miraculously, there is an exact formula for this, the ​​inverse Abel transform​​:

This is the central engine of our discussion. Again, let's not get lost in the mathematical derivation. Let's focus on the astonishing insight it provides. The key ingredient is the derivative, $\frac{dI}{dy}$. Why? Think about what happens when you compare the measurements from two very close lines of sight, one at $y$ and the next at $y+dy$. The paths are almost identical, except near the point of closest approach, at radius $r \approx y$. The change in the measurement, $\frac{dI}{dy}$, must therefore be most sensitive to the properties of the object at that specific radius. The inverse Abel transform is the precise mathematical embodiment of this intuition. It uses the rate of change of the projection to reconstruct the value of the local property, radius by radius, from the outside in.

The true beauty of a tool is seen in its application. Let's look at a few "portraits" of physical systems revealed by the Abel inversion.

• ​​The Perfect Gaussian:​​ Suppose we are studying a fusion plasma and our instruments tell us that the brightness profile we measure looks like a perfect bell curve, a Gaussian function: $I(y) = I_0 \exp(-y^2/w^2)$. What does the plasma actually look like inside? When we apply the inverse Abel transform, an almost magical result appears: the local emissivity profile $\epsilon(r)$ is also a Gaussian! It's as if the object and its shadow share the same fundamental shape. This self-similarity is a hallmark of elegance in physics.

• ​​The Hollow Shell:​​ Now consider a more peculiar case: a hypothetical hollow plasma column, with all its density concentrated in a shell between an inner radius $a$ and an outer radius $b$. The line-integrated measurement $\Delta\phi(y)$ turns out to be a somewhat complicated function involving square roots. It doesn't look simple at all. But when we turn the crank on the Abel inversion, it reveals the simple truth with perfect clarity: the reconstructed density $n_e(r)$ is zero everywhere except in the region between $a$ and $b$, where it is a constant. This is a powerful lesson: a simple underlying reality can cast a complex-looking shadow, and the inversion allows us to see past the shadow to the object itself.

• ​​The World of Polynomials:​​ It turns out that this linearity runs deep. If our measurements can be described by a simple polynomial function with only even powers, say $I(y) = c_0 + c_2 y^2$, the underlying radial profile we are seeking is also a simple polynomial with only even powers, for example $\epsilon(r) = d_0 + d_2 r^2$. The transform simply remaps the coefficients in a predictable way. This predictable relationship is instrumental in developing fast computational algorithms for the inversion.

This universality extends even to more exotic measurements. Whether we are probing a plasma's edge with microwaves (reflectometry) or analyzing the explosion of a molecule by capturing its fragments on a screen (velocity map imaging), different flavors of the Abel inversion appear as the natural language to translate our measurements into physical reality.

The Abel inversion is powerful, but it's not a silver bullet. Its power is built entirely on the foundation of perfect axisymmetry. What happens when that foundation has a crack?

Let's imagine our "cylindrical" plasma is actually slightly squashed, with an elliptical cross-section. An experimentalist, unaware of this imperfection, takes their data and applies the standard Abel inversion. The formula still produces an answer, but it's a distorted one. A careful analysis shows that for a slight ellipticity $\epsilon$, the reconstructed density at the very center will be wrong by an amount directly proportional to that ellipticity. The reconstructed central density will be approximately $(1 - \epsilon)$ times the true value. This is a crucial lesson: our mathematical tools are only as good as our physical assumptions. A small deviation in reality can lead to a predictable, and sometimes correctable, error in our conclusion.

Systematic errors in our measurement setup can also propagate through the inversion. If we are using reflectometry to measure a plasma density profile that starts at a true edge location $R_0$, but our instrument is miscalibrated and thinks the edge is at $R_0 + \delta R_0$, this error will contaminate our result. The inversion will produce a density profile that is systematically shifted, resulting in an error in the inferred density of $-\alpha\,\delta R_0$, where $\alpha$ is the density gradient. The inversion faithfully processes the information—including our mistakes!

There is one final, and perhaps most important, practical challenge. The inverse Abel formula contains a derivative, $\frac{dI}{dy}$. In the clean world of mathematics, this is no problem. But in the real world, all measurements have noise—small, random fluctuations.

What happens when you take the derivative of a noisy signal? A disaster! A tiny, harmless wiggle in the data can become a huge, wild spike in the derivative. Applying the inversion formula directly to noisy data will almost always produce a reconstructed profile that is completely swamped by garbage oscillations. The problem is what mathematicians call ​​ill-posed​​. It is like trying to balance a needle on its point; the slightest perturbation sends it tumbling.

So, how do we solve a problem that is so exquisitely sensitive to noise? We have to be clever. We must perform ​​regularization​​. Instead of trying to find a solution that fits the noisy data perfectly (which would mean fitting the noise), we look for a solution that is "nice" and "smooth" and fits the data reasonably well.

A very powerful and common method is ​​Tikhonov regularization​​. Imagine we are trying to find the best reconstruction $\mathbf{f}$ from our noisy data $\mathbf{y}$. Instead of just minimizing the misfit $\lVert \mathbf{A}\mathbf{f} - \mathbf{y} \rVert^2$, we minimize a combined goal:

The first term is the "data fidelity" term—it wants the solution to match the measurements. The second term is the "regularization" term. Here, $\mathbf{L}$ is typically a derivative operator, so $\lVert \mathbf{L}\mathbf{f} \rVert^2$ measures how "wiggly" or "rough" the solution is. The parameter $\lambda$ is a knob that controls the trade-off. If $\lambda=0$, we are back to the noisy, unstable solution. If $\lambda$ is very large, we get a very smooth solution that might ignore the data completely.

The art is in choosing the right $\lambda$. A wonderfully principled way to do this is the ​​Morozov discrepancy principle​​. It says we should choose $\lambda$ such that our solution fits the data just about as well as the known level of noise. In other words, stop trying to make the fit better once the remaining error is about the size of the noise itself. Trying to fit the data any more closely would be a fool's errand of just fitting the random fluctuations. This profound idea—of not taking our data too literally—is the key to extracting meaningful information from the noisy reality of experimental science.

From the heart of a star to the forces between individual atoms, the Abel inversion provides a universal lens. By understanding its principles, its power, and its pitfalls, we gain the ability to look through the shadows of measurement and see the beautiful, hidden structures of the world around us.

Now that we have some feeling for the mathematical machinery of the Abel inversion, we can ask the most important question of all: What is it good for? The answer, you may be delighted to find, is that this clever piece of mathematics is a kind of secret key, unlocking our ability to see inside things we can't possibly slice open. Nature, it turns out, often presents us with objects that are beautifully symmetric—spheres or cylinders—but our view of them is frustratingly flattened. We see a two-dimensional projection, and the details of the internal, three-dimensional structure are all smeared together along our line of sight.

The Abel inversion is our mathematical tool for "un-smearing" that projection. It allows us, under the crucial assumption of symmetry, to reconstruct the true 3D reality from the 2D shadow it casts. It’s a bit like figuring out how a glowing, translucent glass marble is lit from within, just by looking at its face. You might be surprised to learn just how many different kinds of scientists—astronomers, chemists, physicists, engineers—are all, in a sense, trying to figure out what's inside their own particular "glass marbles." Let's take a journey through some of these fields and see this remarkable idea in action.

Perhaps nowhere is our viewpoint more limited than in astronomy. We are stuck on Earth, looking out at objects light-years away. We can't fly a spaceship through a star or a nebula to take a sample. All we get is the light that arrives at our telescopes, a flat picture on the sky. Yet, we speak with confidence about the temperature and density inside these objects. How is this possible?

Consider a beautiful, glowing planetary nebula, the remnant of a dying star. If it's roughly spherical, the image we see is the total brightness from all layers of the gas along our line of sight. The outer, thinner parts contribute, as do the denser, inner parts. An Abel inversion allows us to peel this cosmic onion layer by layer. From the observed surface brightness profile $I(x)$, we can reconstruct the true volume emissivity $\epsilon(r)$—how much light is actually being produced at each radial distance $r$ from the center.

It gets even better. Different chemical elements, or the same elements at different temperatures, emit light at very specific frequencies or "colors." By performing an Abel inversion on images taken in different colors, corresponding to different emission lines, we can map not just the density but also the physical conditions inside the nebula. For example, the ratio of two different emission line emissivities can be highly sensitive to the local temperature. This allows astronomers to construct a radial temperature profile, $T(r)$, of an object they can never hope to visit.

A similar trick is used closer to home, to probe the atmospheres of planets in our own solar system. When a spacecraft sees a distant star pass behind a planet's atmosphere—an event called a stellar occultation—the starlight doesn't just blink out. It fades. The way it fades as a function of the ray's closest approach distance, $h$, gives us a measurement of the total dimming, or optical depth, $\tau(h)$. This projected profile can be inverted to find the local extinction coefficient $\alpha(r)$ at each altitude $r$ in the atmosphere. It is, in effect, a CAT scan of a planet's atmosphere, taken from millions of miles away.

But this powerful tool comes with a warning. Its magic relies on the assumption of perfect spherical (or cylindrical) symmetry. What if Nature isn't quite so neat? Cosmologists face this problem when studying huge clusters of galaxies. These clusters are filled with hot gas, and their total mass is dominated by dark matter. The amount of normal matter (gas) compared to the total matter, the "baryon fraction," is a fundamental parameter of our universe. A common way to measure it is to observe the projected gas density, use Abel inversion to get the 3D gas density, and then use the laws of physics to infer the total mass holding it all together. But what if a cluster is not a sphere, but is instead shaped more like a cigar (a prolate spheroid), and we happen to be looking down its long axis? Our assumption of sphericity is now false. A careful analysis shows that this geometric mistake leads to a systematic error: we infer an incorrect total mass and, as a result, an incorrect baryon fraction. In fact, for a prolate cluster with an axis ratio $q$, the inferred baryon fraction is biased by exactly a factor of $q$. This is a beautiful, if cautionary, example of how understanding the assumptions of a tool is just as important as knowing how to use it.

Let's now turn our gaze from the grandness of the cosmos to the intricate dance of atoms and molecules. Here too, we find ourselves trying to reconstruct a 3D reality from 2D projections.

In modern chemistry, a powerful technique to study the dynamics of a chemical reaction is called velocity map imaging. Two molecular beams are crossed, they react, and the newly formed product molecules fly outwards from the collision point. These products are then guided to a 2D detector, which takes a picture. This picture is essentially a snapshot of the flattened-out velocity distribution of the products. For many reactions, the products fly out with equal probability in all directions in the center-of-mass frame. The resulting 3D velocity distribution is spherically symmetric, and the 2D image we capture is its Abel projection. By applying an inverse Abel transform to the radial intensity of the image, chemists can reconstruct the full 3D velocity distribution of the reaction products. This distribution is a direct fingerprint of the forces and energy release during the unimaginably brief moment of chemical transformation.

The same principle helps us to "feel" the world at the atomic scale. The Atomic Force Microscope (AFM) can map surfaces with incredible resolution by tapping a very sharp tip along them. In one of its most sophisticated modes (FM-AFM), the microscope's cantilever is kept oscillating at a fixed amplitude, and the machine measures how the oscillation frequency changes as the tip interacts with the surface. This frequency shift, $\Delta f(z)$, is not a direct measure of the force; it's a complicated average of the tip-sample force over a full swing of the cantilever's oscillation. How can we recover the true, fundamental curve of force versus distance, $F(z)$? The relationship between $\Delta f(z)$ and $F(z)$ is an integral equation that can be inverted. A procedure developed by Sader and Jarvis does precisely this, using a method mathematically equivalent to an Abel inversion, to "de-average" the data and extract the underlying force law. It is a stunning example of wringing a fundamental physical quantity out of a complex, dynamic measurement.

This idea of working backward from a measured outcome to an underlying law is one of the deepest in physics. It's called the inverse scattering problem. Imagine you want to know the force field, or potential, surrounding a particle. The only way to find out is to shoot other particles at it and see how they deflect. The phase shift of the scattered particle's wave function, a measurable quantity, is related to the potential $V(r)$ through an integral. In certain important approximations (like the WKB or eikonal approximations), this integral takes the form of an Abel transform. This means that by measuring the scattering phase shifts at different energies, we can perform an Abel inversion to work backward and reconstruct the potential itself! This technique is a cornerstone of nuclear and particle physics, where it is used to determine the optical potential that describes how neutrons and protons scatter off atomic nuclei. It's remarkable that a specific measured phase shift profile, one involving a modified Bessel function $K_0$, inverts to reveal the famous Yukawa potential, $V(r) \propto \exp(-r/a)/r$, which describes the force mediated by a massive particle.

The reach of Abel inversion extends to cutting-edge technology and even revisits classic problems in mechanics, revealing their hidden depths.

In the quest for clean energy from nuclear fusion, scientists must confine a plasma—a gas of ions and electrons heated to millions of degrees—within a magnetic bottle called a tokamak. To control this searingly hot plasma, they must know its density profile, $n_e(r)$, from the hot center to the cooler edge. One key diagnostic is reflectometry. Microwaves are sent into the plasma, and they reflect from a layer where the wave frequency matches the local "plasma frequency," a value that depends on the density. By sweeping the microwave frequency and measuring the round-trip travel time of the reflected signal, physicists obtain a dataset. This dataset, after some manipulation, can be cast into an Abel-type integral equation, which is then inverted to give the plasma density profile. Similar mathematical structures appear in materials science for deducing the 3D statistical structure of a complex material from a 2D microscopic image. In all these cases, we see inside an object that is inaccessible either because it is too hot, too small, or too complex to dissect.

Let us end with a jewel from classical mechanics, a problem first considered by Christiaan Huygens in the 17th century. We are all familiar with a simple pendulum. Its period, the time for one full swing, depends slightly on how far it swings—the period increases with amplitude. This is an annoying imperfection for a clock. Is it possible, Huygens wondered, to design a frictionless potential well $U(x)$ for a particle of mass $m$ such that the period of oscillation $T$ is exactly the same, regardless of the energy $E$ of the oscillation?

This is called the isochrone problem, and it is a classic inverse problem. The period $T(E)$ can be written as an integral involving the potential $U(x)$. To demand that $T(E)$ be a constant, say $T_0$, is to set up an integral equation for the unknown function $U(x)$. Perhaps you can guess what kind of equation it is. It's an Abel integral equation! Solving it reveals that there is only one shape for the potential that satisfies this remarkable condition: the parabolic well of the simple harmonic oscillator, $U(x) = \frac{2\pi^2m}{T_0^2}x^2$. Isn't that something? The reason the simple harmonic oscillator is so fundamental to physics—to springs, pendulums, circuits, and atoms—is deeply connected to this special property, which can be derived using the very same mathematical key that we use to look inside stars. It is a profound and beautiful demonstration of the unity of physics and mathematics.