Hadamard Three Circles Theorem

In the world of complex numbers, analytic functions exhibit a profound regularity, a kind of enforced smoothness that prevents unpredictable behavior. But how can we quantify this regularity? If we know the maximum size of a function on two concentric circles, what can we say about its size on any circle in between? This question is elegantly answered by the Hadamard Three-Circles Theorem, a cornerstone of complex analysis that acts as a fundamental principle of “no undue sagging” for the magnitude of these functions. This article demystifies this powerful theorem by exploring its intuitive foundations and its surprising impact on other scientific fields. In the following chapters, we will first delve into the core principles of logarithmic convexity and the theorem's deep connection to the physics of harmonic functions. We will then journey beyond pure mathematics to witness its application in constraining physical fields and even in analyzing discrete signals, revealing the theorem's role as a versatile tool for understanding bounded systems.

Imagine stretching a rubber sheet over two circular hoops, one smaller and inside the other, with the outer hoop raised higher than the inner one. What shape does the sheet take? It certainly doesn’t sag in the middle; it curves upwards in a smooth, predictable way. This shape is a "minimal surface," the one that minimizes the total elastic energy. Nature, in its efficiency, finds this shape automatically. In the world of complex numbers, analytic functions exhibit a strikingly similar—and equally profound—regularity. This behavior is captured by a beautiful result known as the ​​Hadamard Three-Circles Theorem​​. It acts as a fundamental principle of "no undue sagging" for the magnitude of these functions.

Let's get a feel for this principle. Suppose we have a function, $f(z)$, that is ​​analytic​​—meaning it's perfectly smooth and well-behaved in the complex plane—within the region between two concentric circles, an annulus. Let's say the inner circle has radius $r_1$ and the outer has radius $r_3$. For any radius $r$ between $r_1$ and $r_3$, we can find the largest value that the modulus (or "size") of our function, $|f(z)|$, takes on the circle of that radius. Let's call this maximum value $M(r)$. So, $M(r) = \max_{|z|=r} |f(z)|$.

You might think that $M(r)$ could behave quite erratically as $r$ increases. Perhaps it dips and rises unpredictably. The Hadamard Three-Circles Theorem tells us this is not the case. It states that the function $\log M(r)$ is a ​​convex function​​ of $\log r$.

What on earth does that mean? A function is convex if, when you draw a straight line connecting any two points on its graph, the graph itself never goes above that line. It always "bows downwards" or is straight. So, if we were to plot the logarithm of the maximum modulus against the logarithm of the radius, we would get a convex curve. The function's growth is constrained; it cannot have "dips" in this logarithmic plot.

Let's make this concrete. Suppose a function is analytic in the annulus $1 \le |z| \le 27$. We find that its maximum size on the inner circle is $M(1) = 2$ and on the outer circle is $M(27) = 162$. What's the highest possible value it could reach on an intermediate circle, say at radius $r=3$? The convexity rule allows us to draw a "limit line" in our log-log plot between the points for $r=1$ and $r=27$. The value at $r=3$ cannot exceed this line. A direct calculation based on the convexity inequality shows that the maximum modulus at $r=3$ can be no larger than $6\sqrt[3]{3}$. The theorem provides a rigid ceiling. The function is not free to grow as it pleases; its nature as an analytic function imposes this beautiful geometric constraint.

This log-log convexity might seem a bit strange and arbitrary. Why this specific relationship? The answer reveals a deep and wonderful connection to another area of physics and mathematics: the study of ​​harmonic functions​​.

A function is harmonic if it satisfies Laplace's equation, $\nabla^2 u = 0$. You may have seen this equation before. It describes things in a state of equilibrium, or "steady state." The temperature distribution across a metal plate that has been left to cool, the electrostatic potential in a region free of charge, or the shape of a stretched soap film are all described by harmonic functions. The most crucial property of harmonic functions is the ​​maximum principle​​: a non-constant harmonic function inside a region must attain its maximum value on the boundary of that region, never in the interior. You can't have a point in the middle of a metal plate that is hotter than every point on its edges.

Here is the magical link: if a function $f(z)$ is analytic and, importantly, has no zeros in a region, then the function $u(z) = \log|f(z)|$ is harmonic! The logarithm of the modulus of an analytic function behaves just like temperature or an electrostatic potential.

With this insight, the Three-Circles Theorem becomes much more intuitive. It is essentially a restatement of the maximum principle for harmonic functions, but applied to an annulus. The rule that $\log M(r)$ is a convex function of $\log r$ is the specific form the maximum principle takes in this circular geometry. In fact, a similar principle holds for any harmonic function $u(z)$, not just those that come from $\log|f(z)|$. The maximum value of a harmonic function on a circle of radius $r$, let's call it $U(r)$, is also a convex function of $\log r$. The Hadamard theorem is a special case of this more general principle for harmonic functions, applied to the harmonic function $\log|f(z)|$.

The story gets even better. What if instead of looking at the maximum value of $\log|f(z)|$ on a circle, we look at its average value? Averaging tends to smooth things out. And in this case, it does so perfectly. If we define $M_{avg}(r)$ as the average value of $\log|f(z)|$ on the circle of radius $r$, it turns out that $M_{avg}(r)$ is not just convex, but is a perfectly ​​linear function​​ of $\log r$. In our log-log plot, the graph of the average value is a perfectly straight line connecting the boundary values! The maximum values can only bow below this line, but the average walks the straight and narrow path.

Now, let's consider the other extreme. We have a ceiling for the maximum modulus; do we have a floor for the minimum modulus? Let $m(r) = \min_{|z|=r} |f(z)|$ be the minimum size of the function on a circle. If our function $f(z)$ is non-zero in the annulus, then its reciprocal, $g(z) = 1/f(z)$, is also analytic. The maximum modulus of $g(z)$ is simply one over the minimum modulus of $f(z)$. That is, $\max|g(z)| = 1/m(r)$.

We can apply the Three-Circles Theorem to $g(z)$: $\log(\max|g(z)|)$ is a convex function of $\log r$. Substituting what we know, this means $\log(1/m(r)) = -\log m(r)$ is convex. If a function is convex, its negative is ​​concave​​—its graph "bows upwards." Therefore, we arrive at a symmetric result: ​​$\log m(r)$ is a concave function of $\log r$​​. This gives us a lower bound. Just as the function can't bulge "up" too much in the middle, it also can't sag "down" too much. It is constrained from both above and below in this elegant, symmetric, logarithmic way.

So far, we have a major restriction: for $\log|f(z)|$ to be harmonic, the function $f(z)$ cannot have any zeros inside our annulus. What happens if it does? What if $f(z)$ has, for instance, a zero at the origin? Then $\log|f(z)|$ shoots off to negative infinity, and our beautiful harmonic picture seems to break.

This is where the true power and elegance of the method shine. A mathematician doesn't give up; they transform the problem. If $f(z)$ has a zero of, say, order two at the origin, it behaves like $z^2$ near that point. We can "divide out" this behavior by defining a new auxiliary function, $g(z) = f(z)/z^2$. The miracle is that if $f(z)$ was analytic, this new function $g(z)$ is also analytic, and it is no longer zero at the origin!

We can now apply the Three-Circles Theorem to our well-behaved function $g(z)$ to find a bound on its maximum modulus, $\max|g(z)|$. Once we have that, it's a simple step to translate this back into a bound for our original function, since $|f(z)| = |z|^2 |g(z)|$. This clever trick of "regularizing" the function allows us to handle seemingly difficult exceptions, extending the reach of the theorem immensely. It is a beautiful example of a common strategy in science and mathematics: if you encounter a problem, change your point of view until the problem disappears.

From a simple geometric constraint emerges a rich tapestry of interconnected ideas, linking the growth of functions to the steady-state laws of physics. The Hadamard Three-Circles Theorem is not just a formula; it is a window into the rigid and elegant structure that underlies the seemingly abstract world of complex analytic functions.

Now that we have grappled with the mechanics of the Hadamard Three-Circles Theorem, you might be tempted to file it away as a neat, but perhaps niche, piece of mathematical machinery. A curious property of functions on a donut-shaped region in the complex plane. But to do so would be to miss the forest for the trees! The true beauty of a deep theorem like this one lies not in its isolation, but in its surprising and powerful connections to a vast landscape of ideas, both within mathematics and far beyond. It is a fundamental statement about how influence propagates, how boundaries constrain the interior, and its echoes can be heard in fields that seem, at first glance, to have nothing to do with complex numbers.

Let’s embark on a journey to see where this principle takes us. We'll see that it's not just a theorem; it's a way of thinking.

At its heart, the Three-Circles Theorem is a principle of regularity. It tells us that an analytic function cannot be pathologically "bumpy" inside its domain. Imagine you have an elastic membrane stretched over an annulus. You know the maximum height of the membrane on the inner boundary and on the outer boundary. The theorem is the mathematical analogue of saying that the maximum height on any circle in between can't have a sudden, unexpected peak or valley. Its growth profile is smooth and predictable in a specific logarithmic sense.

This is more than just a qualitative statement. The theorem gives a sharp bound. Consider a function $f(z)$ that is analytic in the annulus $1 \le |z| \le e^{\pi}$. If we know that $|f(z)|$ is no larger than some value $M_1$ on the inner circle $|z|=1$, and no larger than $M_1 e^2$ on the outer circle $|z|=e^{\pi}$, the theorem doesn't just give us an upper bound for the function on an intermediate circle—it gives us the best possible one. For a point like $z_0 = i e^{\pi/4}$, which lies on the circle $|z| = e^{\pi/4}$, the theorem constrains the magnitude $|f(z_0)|$ with unforgiving precision. The maximum value is locked in by the boundaries, interpolated via that elegant logarithmic convexity we discussed. The function can stretch, but only in this prescribed, orderly way.

This idea is more general than it looks. The principle applies not just to the modulus of an analytic function, but to a wider class of "well-behaved" functions known as subharmonic functions. These are functions whose value at a point is always less than or equal to the average of its values on any circle around it. The logarithm of the modulus of an analytic function, $\ln|f(z)|$, is a prime example. By thinking in terms of subharmonic functions, we can see the Three-Circles Theorem as a special case of a more general maximum principle. If a subharmonic function is bounded on the frontiers of an annulus, we can cook up a simple, radially symmetric harmonic function (of the form $a \ln|z| + b$) that matches those boundary values. The maximum principle then guarantees that our subharmonic function must lie below this simple harmonic "ceiling" everywhere inside. This powerful technique allows us to find sharp bounds in more general settings, for instance, for a function mapping an annulus into the unit disk. The theme remains the same: the behavior inside is smoothly and tightly controlled by the behavior on the boundary.

The connection becomes even more profound when we realize that analytic functions are intimately tied to the physical world. The real and imaginary parts of any analytic function are harmonic functions. These are the darlings of physics, describing everything from the steady-state temperature distribution in a metal plate to the electrostatic potential in a region free of charge. They are, in a sense, the "smoothest" possible functions, averaging out all local fluctuations.

So, does the spirit of Hadamard's theorem extend to these physical fields? Absolutely. Imagine a long, hollow pipe, represented by our annulus. Let's say we are interested in the temperature, $u(z)$, which is a harmonic function. We may not know the exact temperature at every point on the boundaries, but perhaps we know the oscillation on each boundary circle—that is, the difference between the hottest and coldest point, $\Omega(r) = \sup_{|z|=r} u(z) - \inf_{|z|=r} u(z)$.

If we know the oscillation on the inner wall is $O_1$ and on the outer wall is $O_2$, what is the largest possible temperature fluctuation we can expect on a circle of pipe somewhere in the middle? Here again, a version of the three-circles principle emerges. The maximum oscillation $\Omega(\rho)$ on an intermediate circle $|z|=\rho$ is bounded by a convex combination of the boundary oscillations, with the weights determined precisely by the logarithms of the radii. It’s the same mathematical tune, just played in a different key. The principle of smooth, logarithmic interpolation governs not just the maximum value of a complex function, but also the maximum temperature swing in a physical system. This reveals a deep unity in the mathematical structure of idealized physical laws.

Now for the most dramatic leap. We journey from the pristine world of continuous functions to the gritty, practical domain of digital signal processing. How could a theorem about complex annuli have anything to say about a discrete audio signal or a radar echo? The bridge is a remarkable mathematical tool called the ​​Z-transform​​.

In signal processing, we often represent a sequence of numbers $x[n]$ (our discrete signal) as a function of a complex variable $z$, called the Z-transform, $X(z) = \sum_{n=-\infty}^{\infty} x[n] z^{-n}$. This is just a Laurent series! The region in the complex plane where this sum converges and defines an analytic function is, you guessed it, an annulus, known as the Region of Convergence (ROC). The properties of the function $X(z)$ in this "frequency domain" hold the key to understanding the behavior of the signal $x[n]$ in the "time domain."

Now, imagine a practical scenario. We have a system, and we can't know its Z-transform $X(z)$ everywhere. But suppose we can perform measurements and find the maximum magnitude of $X(z)$ on two circles within its ROC, say at radii $r_1$ and $r_2$. We have $M(r_1)$ and $M(r_2)$. What does this tell us about the original signal $x[n]$?

This is where Hadamard's theorem enters the stage with stunning effect.

1. First, we know that because $X(z)$ is analytic, the logarithm of its maximum modulus, $\ln M(r)$, must be a convex function of $\ln r$. This allows us to bound $M(r)$ for any radius $r$ between $r_1$ and $r_2$.
2. Next, we use another classic result, the Cauchy Integral Formula, which allows us to recover the original signal samples $x[n]$ by integrating $X(z)$ around a circle in the ROC. A standard estimation technique on this integral gives us a simple inequality: $|x[n]| \le M(r) r^n$ for any $r$ in the ROC.
3. Putting these two pieces together is the masterstroke. We have a bound on $|x[n]|$ that depends on $M(r)$, and we have a bound on $M(r)$ that depends on our measurements at the boundaries. Combining them, we can find the tightest possible bound on any given signal sample, like $|x[10]|$, based only on the two measurements we made!

The result is not just a number; it's a deep insight. It tells us how the constraints in the frequency domain (the magnitude of the transform) translate directly into constraints in the time domain (the magnitude of the signal). For example, this analysis can place hard limits on how quickly the signal must decay for negative times and how quickly it's allowed to grow for positive times. The mere fact that the Z-transform is well-behaved in a certain annulus forces the underlying signal to have a specific character. It connects abstract mathematical properties to tangible physical constraints.

Isn't that something? A century-old theorem, born from pure mathematical curiosity about functions on a plane, provides engineers with a practical tool to deduce the fundamental properties of a time-varying signal from just a handful of measurements. It is in these unexpected, cross-disciplinary applications that we see the true power and inherent beauty of a great mathematical idea. It’s a testament to the interconnectedness of all things mathematical and a beautiful illustration of how the abstract rules governing one domain can provide profound, concrete insights into another.