Symplectic Capacities

In the world of Hamiltonian mechanics, the state of a system is a point in a high-dimensional space known as phase space. While Liouville's theorem guarantees that the volume of any region of states is conserved over time, this concept of "size" is deceptive. Volume preservation allows for extreme distortions, where a ball of states could theoretically be squeezed into an infinitely thin needle. This highlights a critical gap in our understanding: volume is the wrong ruler for measuring size in phase space. A more robust measure is needed to capture the true, unyielding rigidity of Hamiltonian dynamics.

This article introduces the concept of ​​symplectic capacities​​, a powerful tool from symplectic geometry that provides this new ruler. We will explore the fundamental principles that define these capacities and the reasons for their area-like behavior. In the first chapter, "Principles and Mechanisms," we will delve into the axioms of capacities and see how they lead to the astonishing Gromov's Non-Squeezing Theorem, a result that reveals the inherent stiffness of phase space. Subsequently, in "Applications and Interdisciplinary Connections," we will discover how these abstract ideas are applied to concrete physical problems, from proving the existence of periodic orbits in classical systems to quantifying the mysterious phenomenon of quantum entanglement.

How do we measure the "size" of an object? For everyday things, we have intuitive notions: length, area, volume. These are the rulers we use to quantify the world. In the abstract landscape of Hamiltonian mechanics—the phase space where the complete state of a system is represented by a single point—one might think that volume is the ultimate measure of size. After all, the famous Liouville's theorem tells us that as a system evolves, the volume of any region of states in phase space remains perfectly constant.

This, however, is a profound deception. While true, volume preservation is a much "floppier" condition than it sounds. It allows for astonishing distortions. Imagine a perfectly round ball of possible initial states in a four-dimensional phase space. Liouville's theorem says the 4D volume of this ball is conserved as it evolves. But it doesn't prevent the system's dynamics from squeezing this ball in one direction and stretching it in others, transforming it into a long, thin "needle" or a wide, flat "pancake."

To see this is not just a wild fantasy, consider a simple, non-physical transformation in a $2n$-dimensional space. We can construct a map that is perfectly volume-preserving yet dramatically changes a shape's proportions. Imagine a linear map that squishes the first pair of coordinates, $(x_1, y_1)$, by a factor $s 1$, and to compensate, stretches all other coordinate pairs by a factor $t > 1$. By carefully choosing $t = s^{-1/(n-1)}$, the total $2n$-dimensional volume is conserved, with the determinant of the transformation being exactly one. Such a map can take a ball of radius $R$ and squeeze it into a cylinder of radius $r = sR$, which can be as slender as we wish. If Hamiltonian dynamics were only about preserving volume, then a ball of states could indeed be squeezed into a cylinder of arbitrarily small radius.

But Hamiltonian flows are not just any volume-preserving maps. They possess a hidden, unyielding rigidity. Volume is the wrong ruler. We need a new way to measure size in phase space, a measure that is sensitive to this hidden stiffness. This new measure is called ​​symplectic capacity​​.

Rather than defining a specific, complicated formula, let's understand symplectic capacity by its behavior—by the rules of the game it must play. Any function $c$ that we wish to call a symplectic capacity must obey a few fundamental axioms:

1. ​​Monotonicity​​: If a region $U$ can fit inside another region $V$ in a "symplectically appropriate" way (meaning, via a ​​symplectic embedding​​), then its capacity must be less than or equal to that of $V$. That is, if $U \hookrightarrow V$, then $c(U) \le c(V)$. This is the most basic property we'd expect from any notion of "size."

2. ​​Invariance​​: As a direct consequence, if two regions $U$ and $V$ can be transformed into one another by a ​​symplectomorphism​​ (a structure-preserving transformation of the whole space, like the evolution of a Hamiltonian system), then their capacities must be identical: $c(U) = c(V)$. The capacity is an invariant of the symplectic structure.

3. ​​The Conformality Law​​: This is the most crucial and revealing axiom. If we uniformly scale the entire phase space by a factor $\lambda$ (sending every point $z$ to $\lambda z$), the capacity of any set $U$ must scale by a factor of $\lambda^2$. That is, $c(\lambda U) = \lambda^2 c(U)$. This is remarkable. It tells us that symplectic capacity behaves like a two-dimensional ​​area​​, not like a $2n$-dimensional volume (which would scale as $\lambda^{2n}$). This rule originates from the fact that the fundamental object in symplectic geometry, the ​​symplectic form​​ $\omega_0 = \sum_{i=1}^n dq_i \wedge dp_i$, is a 2-form, and its pullback scales by $\lambda^2$ under such a dilation. This is the first clue that the true rigidity of phase space is somehow two-dimensional in nature. For a concrete example, the most famous capacity, the Gromov width, gives the capacity of a ball of radius $R$ as $\pi R^2$. If we dilate by $\lambda$, the new ball has radius $\lambda R$ and its capacity becomes $\pi(\lambda R)^2 = \lambda^2(\pi R^2)$, perfectly obeying the scaling law.

These simple rules define a new kind of geometry—a geometry of areas, not volumes, that governs the dynamics in phase space.

Now that we have the rules for our new ruler, let's see what it can do. Its most celebrated achievement is a result so counter-intuitive and powerful that it launched the entire field of modern symplectic topology: ​​Gromov's Non-Squeezing Theorem​​.

Let's return to our two shapes: the $2n$-dimensional ball $B^{2n}(R)$ of radius $R$, and the symplectic cylinder $Z^{2n}(r)$ of radius $r$. The ball is a finite sphere of states. The cylinder is a peculiar object, defined by a constraint on only one pair of conjugate coordinates, say $q_1^2 + p_1^2 \le r^2$, while being infinitely extended in all other $2n-2$ directions.

The question is simple: Can a Hamiltonian flow, a true physical evolution, take the ball $B^{2n}(R)$ and squeeze it entirely inside the cylinder $Z^{2n}(r)$?

As we saw, from a volume perspective, this is trivial. The ball has finite volume, the cylinder has infinite volume; it seems there's plenty of room. But let's apply our new symplectic ruler. The proof is a beautiful piece of physical reasoning, so simple it feels like a magic trick.

Let's use a specific capacity (like the Gromov width) that is "normalized" on these shapes, such that $c(B^{2n}(R)) = \pi R^2$ and $c(Z^{2n}(r)) = \pi r^2$. Now, suppose for the sake of argument that a Hamiltonian flow $\phi$ does squeeze the ball into the cylinder. This means the final state of the ball, $\phi(B^{2n}(R))$, is a subset of $Z^{2n}(r)$. We apply our axioms:

1. From the ​​invariance​​ axiom, the capacity of the ball doesn't change during the evolution: $c(B^{2n}(R)) = c(\phi(B^{2n}(R)))$.
2. From the ​​monotonicity​​ axiom, since the evolved ball is now inside the cylinder, its capacity must be no more than the cylinder's: $c(\phi(B^{2n}(R))) \le c(Z^{2n}(r))$.
3. Combining these, we get a simple inequality: $c(B^{2n}(R)) \le c(Z^{2n}(r))$.
4. Substituting the known values for the capacities gives: $\pi R^2 \le \pi r^2$.

This leads to the astonishing conclusion: $R \le r$. It is fundamentally impossible for any Hamiltonian flow—no matter how clever or complex—to squeeze a ball into a cylinder that is ​​thinner​​ than the ball's original radius. It is as if the "shadow" of the ball projected onto the $(q_1, p_1)$ plane has an irreducible area of $\pi R^2$ that cannot be diminished by any valid physical transformation. In two dimensions ($n=1$), this is simply a statement about preserving area, which is less surprising. But in higher dimensions, it reveals a profound rigidity. A symplectic camel, it turns out, cannot pass through the eye of a smaller symplectic needle.

This deep rigidity begs a question: what makes symplectic geometry so special? Why does it behave so differently from the more familiar geometry of curved surfaces, like a sphere or a saddle, which we study in Riemannian geometry?

The answer lies in another foundational result, ​​Darboux's Theorem​​. In essence, the theorem states that locally, all symplectic manifolds of the same dimension look identical. Around any point, you can always find a set of "canonical coordinates" $(q_1, \dots, p_n)$ in which the symplectic form $\omega$ takes on its standard, universal expression $\omega = \sum dq_i \wedge dp_i$.

This is in stark contrast to Riemannian geometry. The defining feature of a curved surface is its ​​curvature​​, which is a local property. You can measure the curvature of a sphere at a single point, and it will be non-zero. This non-zero curvature is an obstruction that prevents you from finding coordinates that make the sphere look like a flat plane in a neighborhood around that point.

Darboux's theorem tells us that symplectic manifolds have no such local invariants. They are, in a sense, "locally flat" everywhere. This has a monumental consequence: if you want to find an interesting geometric property in a symplectic manifold, something that distinguishes one shape from another, it cannot be a local property. It must be ​​global​​, or at least "semi-global." It must depend on how a shape sits inside the whole space and interacts with its overall structure. Symplectic capacities are precisely such non-local invariants. They are blind to the local, Darboux-trivial structure but keenly sensitive to the global topology and embedding of a region, capturing a rigidity that volume completely misses.

What is the ultimate purpose of this beautiful and abstract machinery? One of the most profound applications of symplectic capacities is in proving the existence of fixed points and periodic orbits in dynamical systems, answering questions at the very heart of classical mechanics.

A famous question in dynamics is the ​​Arnold Conjecture​​, which posits that any Hamiltonian flow on a closed manifold (like a torus) must have a certain minimum number of fixed points—points that return to their exact starting position after the flow is applied. Think of stirring a cup of coffee; must some particle of coffee end up exactly where it started?

The theory of capacities provides a powerful tool to answer this. The key idea is to relate capacity to ​​displacement energy​​. The displacement energy of a set $U$, denoted $e(U)$, is the minimum "effort" (measured by a quantity called the Hofer norm) required for a Hamiltonian flow to move the set $U$ completely off of itself, so that the final state has no overlap with the initial one. A fundamental result, the ​​energy-capacity inequality​​, states that for any capacity $c$, we have $c(U) \le e(U)$. This means that sets with large capacity are "heavy" and hard to move. In fact, the Hofer-Zehnder capacity is the sharpest possible such measure, as it is equal to the displacement energy, $c_{HZ}(U) = e(U)$.

Now for the grand finale, a beautiful argument that proves the simplest version of the Arnold conjecture. A fixed point of a flow $\phi$ is a point $x$ where $\phi(x) = x$. This is equivalent to saying that the graph of the flow, the set of points $(x, \phi(x))$, intersects the diagonal line $\Delta = \{(x,x)\}$. We can view both the graph and the diagonal as sets within a larger product space, $M \times M$.

A deep theorem of symplectic topology states that the diagonal $\Delta$ is ​​non-displaceable​​. It has an infinite capacity and infinite displacement energy. It is impossible for any Hamiltonian flow on this product space to move the diagonal completely off of itself. The crucial insight is that the graph of the flow $\phi$ is simply the image of the diagonal under a related Hamiltonian flow. Since the diagonal cannot be moved off of itself, its image—the graph of $\phi$—must intersect the original diagonal.

Therefore, $\phi$ must have a fixed point. The unshakeable rigidity of symplectic geometry, captured by the notion of capacity, guarantees that you cannot stir the phase space without leaving something untouched. From a simple rule about how areas scale, we arrive at a profound truth about the inevitable rhythms and fixed points woven into the fabric of nature's laws.

Having acquainted ourselves with the principles of symplectic capacities, we might now be tempted to ask, "What is all this for?" It is a fair question. We have defined a new way of measuring "size," but is it merely a mathematician's curious plaything, or does it tell us something profound about the world? The answer, you will be delighted to find, is that these capacities are not just abstract numbers; they are the very language of a deep rigidity built into the fabric of nature. They govern the rhythms of classical dynamics, reveal the limits of what is possible, and, in a breathtaking leap, even provide a ruler to measure the "spooky action at a distance" of the quantum world.

Let us embark on a journey to see these ideas at work. We will move from the familiar ticking of a clockwork universe to the strange, interconnected realm of quantum information, and find that the thread of symplectic geometry ties them all together.

Imagine a simple pendulum swinging back and forth. Its state at any moment is described by its position and momentum—a single point in its two-dimensional phase space. As it swings, this point traces a closed loop, returning to its starting state after one full period. This loop encloses a certain area. A fundamental discovery of mechanics is that this area, known as the action of the orbit, is a profoundly important quantity. It turns out that for this simple system, the Hofer-Zehnder capacity of the region swept out by the pendulum's motion is precisely this action. The capacity, our abstract measure of size, has found its first physical meaning: it captures the fundamental "quantum" of action for a periodic motion.

This is not a coincidence. Many of the most important symplectic capacities, like the Hofer-Zehnder and Ekeland-Hofer capacities, are intimately tied to the periodic orbits of Hamiltonian systems. For a given region of phase space, these capacities are often defined by the action of the shortest or "fastest" periodic orbit that lives on its boundary. Think of a planet in a complex gravitational field; its boundary in phase space might host a zoo of possible periodic dances. The capacity zeros in on the most fundamental rhythm, the quickest repeating pattern.

For more abstract shapes, like a four-dimensional ellipsoid in phase space, the situation is similar. Its boundary is alive with periodic motions, each corresponding to a rotation in one of the complex planes. The actions of these orbits form a "spectrum" of values, like the discrete energy levels of an atom. The sequence of symplectic capacities of this ellipsoid is nothing more than this action spectrum, sorted from smallest to largest. So, by computing capacities, we are, in a sense, performing a kind of "phase space spectroscopy," uncovering the fundamental frequencies of the system's dynamics.

Perhaps the most startling application of symplectic capacities is their power to tell us what is impossible. This is the essence of symplectic rigidity. The most celebrated example is Gromov's Non-Squeezing Theorem. Imagine you have a ball of radius $R$ in phase space. Its volume is finite. You might think that you could deform it, stretch it long and thin, and squeeze it into a cylinder of any arbitrarily small radius $r$, as long as the cylinder is long enough to accommodate the volume. Our everyday intuition, based on preserving volume, says this should be possible.

Symplectic geometry says, emphatically, no. A ball $B^{2n}(R)$ can be symplectically squeezed into a cylinder $Z^{2n}(r)$ if and only if $R \le r$. The cross-sectional area cannot be compressed, even if the volume is preserved!

Symplectic capacities provide a beautifully simple proof of this astonishing fact. As we've learned, a key rule of the game is monotonicity: if you can symplectically embed one shape into another, the capacity of the first must be less than or equal to the capacity of the second. The other key rule is normalization: the capacity of a ball $B^{2n}(R)$ is defined to be $\pi R^2$, and so is the capacity of a cylinder $Z^{2n}(r)$. If an embedding of the ball into the cylinder existed, monotonicity would demand that $c(B^{2n}(R)) \le c(Z^{2n}(r))$, which means $\pi R^2 \le \pi r^2$. The impossibility of squeezing follows directly!

This "no-squeezing" principle has a wonderful physical interpretation in terms of displacement energy. Suppose you have a ball of particles in phase space. How much "energy" does it take to move the entire ball so that it no longer overlaps with its original position? This minimum energy is called the displacement energy. One might guess it depends on the details of the push. But remarkably, the displacement energy of a ball $B^{2n}(R)$ is exactly $\pi R^2$—its symplectic capacity. The capacity is a direct measure of the ball's "stubbornness" or "inertia" in phase space.

These ideas are not limited to simple balls and cylinders. They provide powerful tools for solving complex "packing problems" in phase space. Suppose you have a symplectic ellipsoid $E(1,2)$, a shape like a squashed four-dimensional football. What is the smallest four-dimensional ball it can fit into? This is not an easy question to answer by just looking. However, we can compute the entire sequence of Embedded Contact Homology (ECH) capacities for both the ellipsoid and the ball. The monotonicity principle must hold for every single capacity in the sequence. By comparing the two sequences, we can find the tightest constraint, revealing the radius of the smallest possible ball that can contain the ellipsoid. For $E(1,2)$, this method proves that it cannot fit into any ball of radius less than $2$, and furthermore, that an embedding into a ball of radius $2$ actually exists. Capacities give us a universal, algorithmic way to discover these hidden geometric obstructions. The same method works for other shapes, like polydisks, allowing us to map out the landscape of what is possible in the world of symplectic embeddings.

So far, our journey has been through the world of classical mechanics. Now, we take a leap into the quantum realm, where a truly beautiful and unexpected connection awaits. One of the defining features of quantum mechanics is entanglement—the strange, nonlocal correlation between two or more quantum systems. How do we measure "how much" entanglement there is between two particles?

Consider a common scenario in quantum optics: a state of two modes of light, such as two laser beams. For a large class of states called Gaussian states, we can represent the quantum state not by a wavefunction, but by a Wigner function, which is a probability-like distribution on the classical phase space. A Gaussian state is then completely described by its mean and its covariance matrix, $\boldsymbol{\sigma}$. This matrix tells us about the fluctuations and correlations of the position and momentum-like variables of the light fields.

Here is the magic. This covariance matrix $\boldsymbol{\sigma}$ is not just any matrix; it is an object that lives in a symplectic vector space. The physical laws that govern the quantum state—the uncertainty principle and the allowed transformations like squeezing or rotating the light—translate directly into symplectic geometry. The invariants of the quantum state, the quantities that remain unchanged under allowed local operations, are precisely the symplectic invariants of the covariance matrix.

The connection becomes even more profound. Measures of entanglement, like the logarithmic negativity or the squashed entanglement, which are designed to quantify the "quantumness" of the connection between the two modes, can be calculated directly from these classical symplectic invariants! The formulas for these entanglement measures look like they are lifted straight out of a textbook on symplectic geometry. For instance, the squashed entanglement for a symmetric two-mode state turns out to be a simple expression involving the von Neumann entropy, which itself is a function of the symplectic eigenvalues of the covariance matrix and its sub-blocks.

This is a stunning example of the unity of physics. The geometric framework developed to understand the orbits of planets and the rigidity of classical phase space provides the exact mathematical tools needed to quantify one of the most counter-intuitive phenomena in the quantum universe. The "size" and "shape" of a quantum state in phase space, as measured by symplectic geometry, determine how entangled it is.

From the action of a pendulum to the non-squeezing of phase space to the quantification of entanglement, symplectic capacities reveal themselves to be far more than a mathematical curiosity. They are a fundamental language for describing structure and rigidity in a vast range of physical systems, weaving a thread of geometric unity through the classical and quantum worlds.