Phase portraits

Grasping the complete behavior of a system governed by differential equations can be a formidable challenge. While equations describe how a system changes from one moment to the next, they don't immediately offer a global picture of every possible outcome. This article introduces the phase portrait, a powerful visual methodology that bridges this gap by transforming abstract equations into a geometric map of a system's destiny. By visualizing the flow of a system in its state space, we gain an intuitive and comprehensive understanding of its long-term dynamics. The following chapters will first delve into the core principles of phase portraits, exploring their fundamental components like vector fields, equilibrium points, and limit cycles. We will then embark on a journey across various scientific disciplines to witness the universal power of these portraits, showcasing their applications in mechanics, ecology, synthetic biology, and even in determining the ultimate fate of the cosmos.

Imagine you want to understand not just one possible future of a system, but every possible future. You want a map, a kind of god's-eye view of a system’s destiny. If a pendulum is swinging, what happens if you push it a little harder? If the population of rabbits and foxes is changing, what are all the possible ways their story could unfold? This map of destiny is precisely what a ​​phase portrait​​ provides. It is one of the most beautiful and powerful ideas in all of science, transforming abstract equations into a tangible, geometric landscape.

Let's begin with a few clear definitions. To describe a system at any instant, we need a set of numbers—its ​​state​​. For a simple pendulum, this could be its angle and its angular velocity. For a planet orbiting the sun, it's its position and momentum. The collection of all possible states is called the ​​phase space​​. For a system with two variables, like the pendulum, this space is a simple two-dimensional plane.

Now, the laws of physics—be it Newton's laws, chemical reaction rates, or ecological models—tell us how the state changes from one moment to the next. At any point $(x, y)$ in the phase space, the governing equations define a velocity vector, $(\dot{x}, \dot{y})$. This collective set of arrows, one for each and every point, is the ​​vector field​​. It's a field of instructions, a silent director telling the system where to go next. An ​​integral curve​​, or a ​​trajectory​​, is simply the path you trace by following these arrows from a given starting point. The ​​phase portrait​​ is the grand tapestry woven from all these individual threads—it is the collection of all possible trajectories, giving us a complete qualitative picture of the system's dynamics.

There is one crucial rule, however. This beautiful, static map only works for ​​autonomous systems​​—systems whose rules do not change over time. Think of a ball rolling on a hill; the shape of the hill is fixed. What if the rules did change? Consider a child on a swing being pushed periodically. The system's equation might look something like $\ddot{x} + x = \sin(t)$. The vector field that governs its motion in the $(x, \dot{x})$ plane is itself changing with time because of that $\sin(t)$ term. The "flow" is not steady but is sloshing back and forth. Trying to draw a single, static phase portrait for this would be like trying to draw a nautical chart on the surface of a stormy sea—the currents are different every moment. Such systems are called ​​non-autonomous​​, and their trajectories can, in fact, cross each other in the 2D plane, a heresy forbidden in the autonomous world. For the rest of our journey, we will stick to the elegant world of autonomous systems.

When you look at a map, what are the first things you notice? The cities, the mountains, the rivers. In a phase portrait, the most important features are the ​​equilibrium points​​—the points where the vector field is zero, meaning all change ceases. These are the locations where the system can, in principle, rest forever.

A wonderfully intuitive way to think about this is to imagine a particle moving in a one-dimensional landscape defined by a potential energy function, $V(x)$. The state of this particle is its position $x$ and its momentum $p$. The phase portrait shows how $(x,p)$ evolves. Where are the equilibria?

• A valley in the potential energy landscape, a local minimum of $V(x)$, corresponds to a ​​stable equilibrium​​. If you place the particle near the bottom of the valley, it will tend to roll back towards it. This often appears in the phase portrait as a ​​center​​ or a ​​stable node​​.
• A hilltop, a local maximum of $V(x)$, corresponds to an ​​unstable equilibrium​​. If the particle is placed perfectly on top, it stays there. But the slightest nudge will send it rolling away, never to return. This appears in the phase portrait as a ​​saddle point​​.

The trajectory that leads to or from a saddle point is special. For our particle on the landscape, if it has exactly the energy of a hilltop, it can roll up, momentarily come to rest at the peak, and then roll down the other side. This critical trajectory acts as a dividing line, a watershed, in the phase portrait. It is called a ​​separatrix​​, separating qualitatively different types of motion—for instance, separating trajectories that are trapped in a valley from those that can travel over the hill and escape.

In general, for a 2D linear system $\dot{\mathbf{x}} = A\mathbf{x}$, the character of the equilibrium at the origin is completely determined by the eigenvalues of the matrix $A$. If the eigenvalues are real and of opposite sign, we get a saddle. If they are real and of the same sign, we get a node (stable if negative, unstable if positive). If they are complex conjugates with non-zero real parts, we get a spiral. And in the very special case where the eigenvalues are purely imaginary, we get a ​​center​​, where the trajectories are a family of nested ellipses, representing perfect, undamped periodic motion.

You might protest: "This is all fine for simple linear systems, but the real world is nonlinear and messy!" You are absolutely right. The equations governing a swinging pendulum or a chemical reaction are not simple straight lines. But here comes a bit of mathematical magic, a profound result called the ​​Hartman-Grobman Theorem​​.

This theorem tells us something remarkable: as long as an equilibrium point is ​​hyperbolic​​ (meaning none of the eigenvalues of its linearization have a zero real part), then in a small neighborhood around that equilibrium, the phase portrait of the complicated nonlinear system looks exactly like the phase portrait of its simple linear approximation. "Looks like" here has a precise mathematical meaning: the portraits are topologically equivalent. You can continuously bend and stretch the linear portrait (without tearing it) to make it identical to the nonlinear one, preserving all the trajectories and their directions.

This is an idea of immense power. It means that by studying the simple "zoo" of behaviors for linear systems (nodes, saddles, spirals), we gain genuine insight into the local behavior of an enormous class of complex real-world systems. It's the mathematical equivalent of realizing that if you zoom in far enough on a curved surface, it looks flat. Our simple linear pictures are the "flat-earth" maps that are incredibly useful for navigating our local neighborhood in phase space.

But what happens if a system doesn't settle down to a dead stop, nor does it oscillate in a delicate, energy-dependent way like a center? Many systems in nature—from the beating of a heart, to the chirp of a cricket, to the daily cycle of genes turning on and off in our cells—exhibit robust, self-sustaining oscillations. They settle into a rhythm.

In a phase portrait, this behavior is represented by a ​​limit cycle​​. A limit cycle is an isolated closed-loop trajectory. What makes it special is that it is an attractor. Trajectories that start near it, both from the inside and the outside, will spiral towards it and eventually merge with it. Unlike the family of ellipses around a center, where the size of the loop depends on the initial energy, a stable limit cycle has a characteristic amplitude and frequency that the system naturally "wants" to adopt, regardless of its specific starting point (within a certain region, its basin of attraction). The existence of limit cycles, first studied by the great Henri Poincaré, is the signature of a natural oscillator.

Let's bring all these ideas together with one of the most beautiful and classic examples in all of physics: the simple pendulum. Its state is given by its angle $\theta$ and its angular velocity $\omega = \dot{\theta}$. When we write down the equation of motion, $\ddot{\theta} + (g/L) \sin(\theta) = 0$, we immediately notice something wonderful. The mass $m$ of the pendulum bob has completely vanished from the equation! This means that the phase portrait, the entire universe of possible motions, is identical for a 1-kilogram bob and a 100-kilogram bob, as long as the length $L$ is the same. The essence of the motion is independent of the mass.

The phase portrait of the (undamped) pendulum is a masterpiece:

• ​​Equilibria​​: The point $(\theta=0, \omega=0)$ corresponds to the pendulum hanging at rest. It's a stable equilibrium, a ​​center​​, surrounded by closed loops. The points $(\theta=\pm \pi, \omega=0)$ correspond to the pendulum balanced perfectly upside down—an impossible feat! These are unstable ​​saddle points​​.
• ​​Oscillations​​: The closed loops around the center represent the familiar back-and-forth swinging motion. A single loop traces the angle increasing to a maximum, the velocity dropping to zero, the angle decreasing, and so on.
• ​​A Physical Connection​​: We can connect our physical intuition directly to the geometry. At the moment the swinging pendulum reaches its maximum angle, it momentarily stops before turning back. Its angular velocity $\omega$ is zero. In the phase portrait, this corresponds to the trajectory crossing the $\theta$-axis. What is the tangent to the trajectory at this point? Since $\dot{\theta} = \omega = 0$ but $\dot{\omega} = -(g/L)\sin(\theta) \neq 0$, the velocity vector in the phase plane is purely vertical. Thus, the tangent to the trajectory is vertical at the points of maximum displacement. Physics becomes geometry!
• ​​The Separatrix​​: The trajectory that connects the unstable saddle points is the separatrix. It represents the motion where the pendulum has just enough energy to swing up and come to a perfect stop in the inverted position. It separates the phase space into two fundamentally different regions: the inner region of oscillations (swinging) and the outer region of rotations (whirling).

Finally, there is an even deeper level of beauty that phase portraits reveal. The symmetries of the physical laws are reflected as symmetries in the geometry of the phase portrait. For example, if the equations of motion have a certain algebraic symmetry—say, if flipping the sign of the $y$ variable only flips the sign of its rate of change $\dot{y}$—then the resulting phase portrait must be perfectly symmetric with respect to the x-axis. If a certain path is possible, its mirror image must also be possible. This profound connection between the symmetries of the equations and the symmetries of the solutions is one of the most fundamental principles in all of physics, and in the phase portrait, we see it laid bare.

The phase portrait is more than a tool; it is a new way of seeing. It translates the abstract, time-ordered narrative of differential equations into a single, timeless, and beautiful geometric object. It is a map where every path is a story, and the landscape itself is the law.

In the previous chapter, we learned a wonderful new trick. We discovered that the abstract and often intimidating language of differential equations can be translated into pictures. By plotting the velocity of a system against its position, we create a phase portrait—a map that shows every possible journey the system could ever take. A single glance at this map tells us where the system is going, where it came from, and where it will end up. It’s like having a crystal ball for dynamics.

Now, we will embark on a grand tour across the landscape of science to see what these pictures have to tell us. We will find that this simple idea is astonishingly powerful and universal. It appears everywhere, from the familiar swing of a pendulum to the intricate dance of life inside a cell, and even on the grandest stage of all: the evolution of the entire cosmos. We are about to see that the fundamental patterns of change, whether in a machine, a molecule, or a galaxy, can often be captured in a single, elegant drawing.

Let's begin with something simple and familiar: a pendulum. If you pull it back a little and let it go, it swings back and forth in a predictable rhythm. For these small swings, our equations are simple linear ones, and the phase portrait is an unassuming set of concentric ellipses. Each ellipse represents a swing of a certain energy, a perfect, repeatable oscillation. But what happens if we give the pendulum a much harder push?

The true law of the pendulum involves a $\sin(\theta)$, not just $\theta$, and the phase portrait reveals its secret life. For low energies, we still see closed loops, corresponding to the familiar back-and-forth swinging motion, which we call libration. But if we give the pendulum enough energy to swing up and over the top, its motion changes completely. It goes into a perpetual whirling motion, always moving in the same direction. In the phase portrait, this appears as an undulating, wavy line that continues forever along the angle axis. These are called rotations. Separating these two profoundly different behaviors—swinging versus whirling—is a special trajectory called a ​​separatrix​​. It is the boundary of fate. A system on one side of the separatrix is destined to swing forever; a system on the other side is destined to whirl. This beautiful complexity, this dramatic choice between two fates, was hidden inside the simple sine function all along, and the phase portrait made it visible.

This connection between an energy landscape and a phase portrait is a deep one. Imagine a particle of unit mass moving in a one-dimensional potential, $U(x)$. The force on it is $F = -dU/dx$. The phase portrait is just another way of looking at the shape of the potential. A valley in the potential energy landscape, a point of stable equilibrium, becomes a center in the phase plane, surrounded by closed orbits representing oscillations within the valley. A hilltop, a point of unstable equilibrium, becomes a saddle point—a precarious perch from which the slightest nudge sends the particle tumbling down one side or the other. The trajectories that lead to or from this saddle carve up the entire phase space, defining the boundaries of different kinds of motion. In a system with a "double-well" potential, like a chemical bond that can exist in two configurations, we find two valleys and a hill in between. This gives rise to two centers and a saddle. The most fascinating trajectories are the ones connected to the saddle itself. A path that leaves the unstable equilibrium and, after a grand tour, loops back to the very same point is called a homoclinic orbit. It's a journey from instability, back to instability.

Of course, the real world isn't always so perfectly smooth. What about the components in a real machine? Consider a mass attached to a spring, but this is an industrial-strength spring: it pulls back linearly for a bit, but then it hits a limit and can't pull any harder. The force saturates. Does our beautiful geometric picture break down? Not at all! It adapts. In the region where the spring is linear, the trajectories are the familiar ellipses of a simple harmonic oscillator. But when the displacement becomes large and the force becomes constant, the acceleration is constant, and the trajectory becomes a parabola. The complete phase portrait for a large oscillation is a beautiful mosaic, with elliptical arcs stitched seamlessly onto parabolic arcs. The picture perfectly captures the piecewise nature of the physics, showing how versatile this tool truly is.

The same geometric principles that govern pendulums and springs also choreograph the dance of life. Consider the age-old struggle between predators and prey—foxes and rabbits, for instance. Their populations don't just find a happy medium; they oscillate in a timeless cycle. The phase portrait, plotting the predator population versus the prey population, makes this cycle plain to see. We often see a single, closed loop. By following a point around this loop, we can read the entire ecological story: an abundance of rabbits (a move to the right on the portrait) allows the fox population to grow (a move upwards). But the growing fox population eats too many rabbits, causing the rabbit population to crash (a move to the left). With less food, the fox population then starves and declines (a move downwards). This allows the rabbit population to recover, and the cycle begins anew. The abstract geometry of the trajectory—for instance, crossing the vertical axis from the second quadrant to the first—translates directly into a concrete ecological event: the prey population, having been below its long-term average, is now increasing and has just reached that average, while the predators, still enjoying a relative abundance of food, remain above their own average.

These cycles are a direct consequence of the predator-prey interaction. But some systems have a preferred, intrinsic rhythm. Many chemical reactions, when held far from thermodynamic equilibrium (by constantly feeding in reactants and removing products), can begin to oscillate spontaneously, with concentrations of intermediate compounds rising and falling in perfect time. The famous Belousov-Zhabotinsky reaction, for instance, cycles through a mesmerizing sequence of colors. The phase portrait of such a system reveals a new and magical object: a ​​limit cycle​​. Unlike the collection of elliptical orbits around a center, each determined by a different initial energy, a limit cycle is a single, isolated, closed loop. It is an attractor. If the system starts anywhere inside the loop, its trajectory spirals outwards towards the cycle. If it starts outside, it spirals inwards. The system wants to be on this cycle. This makes the oscillation incredibly stable and robust to perturbations, a property essential for any reliable clock, whether chemical or biological. The steady beat of our hearts and the circadian rhythms that govern our sleep-wake cycles are biological manifestations of this fundamental dynamical pattern.

Taking this idea of design further, we can ask: can we engineer biological systems to have specific behaviors? This is the domain of synthetic biology. A classic achievement is the genetic toggle switch, a synthetic gene circuit designed to act like a memory bit in a computer. The circuit consists of two genes that mutually repress each other. The phase portrait shows us exactly how this memory works. Under the right conditions, the system is bistable: it has two stable fixed points. One corresponds to a state where gene A is highly expressed and gene B is shut off; the other corresponds to gene B being expressed and gene A being shut off. These two stable states are separated by an unstable saddle point. The stable manifold of this saddle is a separatrix that divides the entire phase space into two basins of attraction. If the cell's state starts on one side of this boundary, it will inevitably flow to the "A-ON" state. If it starts on the other side, it flows to the "B-ON" state. The cell "remembers" its initial condition. This bistability is also the origin of hysteresis, the phenomenon where the system's response depends on its history. The simple geometry of two stable nodes and one saddle point provides a complete blueprint for engineering a biological memory device.

Phase portraits are not just for modeling systems whose laws we already know; they are also powerful tools for inference. In modern biology, RNA velocity is a cutting-edge technique used to predict the future state of individual cells from a single snapshot of their gene expression. For each gene, a phase portrait is created by plotting the amount of mature, spliced mRNA ($s$) versus its precursor, unspliced mRNA ($u$). A simple kinetic model states that new RNA is transcribed as $u$, which is then spliced into $s$, which is finally degraded. This defines a flow in the $(u, s)$ plane. A cell where the amount of $u$ is high relative to the amount of $s$ is likely in the process of "turning on" that gene; its trajectory in the phase portrait points towards higher $s$. A cell where $s$ is high but $u$ is low has likely finished its burst of expression and is "turning off" the gene. By measuring the $(u, s)$ coordinates for a cell, we can infer its "velocity" vector in this abstract space, and thus predict its developmental trajectory. This remarkable method, which helps us understand how a stem cell decides to become a neuron or a skin cell, is built entirely upon the logic of phase portraits.

Finally, let us turn from the smallest scales of life to the largest scale imaginable: the universe itself. The expansion of our universe is governed by Albert Einstein's theory of general relativity, distilled into a set of relations known as the Friedmann equations. These equations describe the evolution of the universe's scale factor, $a(t)$. We can create a phase portrait for the universe in the $(\dot{a}, a)$ plane. What does this cosmic map tell us? It tells us our ultimate fate. The shape of the trajectories depends critically on a parameter, $k$, which represents the overall curvature of space.

If the universe is spatially "closed" ($k=+1$), with a high enough density of matter and energy, the trajectory in the phase portrait is a closed loop. It starts at the Big Bang ($a=0$ with infinite expansion speed), expands to a maximum size (where $\dot{a}=0$), and then gravity wins. The expansion halts and reverses, sending the universe hurtling back towards a final "Big Crunch."

If the universe is spatially "open" ($k=-1$), the trajectory is an open curve that expands forever. Gravity is not strong enough to halt the expansion, and the universe is destined to grow, cool, and darken for eternity.

And what of the critical case, a "flat" universe ($k=0$)? Its trajectory is the ​​separatrix​​ poised exactly between these two fates. It expands forever, but the expansion rate continuously slows, approaching zero as time approaches infinity. The universe coasts to a stop, but only after an infinite amount of time.

The entire destiny of the cosmos—its fiery birth and its ultimate fate, whether a final collapse or an endless expansion—is laid bare as a simple curve in a two-dimensional plane. There is perhaps no more profound or breathtaking demonstration of the power of the phase portrait. It is a tool that allows us to hold the universe in a drawing, and to read its story. From the tiny oscillations of atoms to the grand sweep of cosmic evolution, the language of phase portraits unifies our understanding of dynamics, revealing the deep and beautiful geometric order that governs the flow of time.