Non-local Equations

In the world of classical physics, we are accustomed to laws that are beautifully local: the effect is always determined by the immediate cause. Gravity, heat flow, and pressure are typically described by equations where every point only interacts with its infinitesimally close neighbors. This viewpoint has been incredibly successful, but it doesn't capture the whole story. A deeper and more interconnected layer of reality is described by non-local equations, where what happens here can be directly influenced by what is happening over there. These equations challenge our intuition and provide the language to model phenomena that involve long-range forces, collective behavior, and memory effects.

This article peels back the layers of this fascinating topic. It addresses the conceptual leap from a local to a non-local perspective, revealing how many seemingly local laws are simply approximations of a more complex, interconnected world. Across the following sections, you will gain a clear understanding of the core principles of non-locality and see them in action. The "Principles and Mechanisms" section will demystify the mathematics behind these equations, explaining the role of integrals, fractional derivatives, and the indispensable Fourier transform. Following that, the "Applications and Interdisciplinary Connections" section will take you on a tour of the real world, showing how non-local models are crucial for understanding everything from the patterns on a leopard's coat to the fundamental properties of advanced materials.

Most of the laws of physics we first encounter in school are wonderfully, beautifully local. Think about it. The force of gravity on you from the Earth depends on your position right now. The pressure in a fluid at some depth depends on the weight of the column of water directly above it. Fourier's law of heat conduction, which gives us the familiar heat equation, states that the flow of heat at a point is proportional to the temperature gradient at that very same point. The universe, in this view, is a grand machine where every part only interacts with its immediate neighbors. Information and influence ripple outwards, but they don't magically leap across space.

But what if this comforting picture isn't the whole story? What if, in certain situations, what happens here is directly and instantaneously influenced by what is happening over there? This is the fascinating world of ​​non-local equations​​. They challenge our intuition and reveal a deeper, more interconnected layer of physical reality.

Let's revisit the flow of heat. The standard heat equation, $\frac{\partial T}{\partial t} = \alpha \frac{\partial^2 T}{\partial x^2}$, arises from assuming the heat flux $q(x,t)$ at point $x$ depends only on the temperature gradient $\frac{\partial T}{\partial x}$ at that same point. But imagine a peculiar material where the mechanism of heat transfer is different. Suppose the flux at $x$ isn't determined by the infinitesimally close gradient, but by the temperature difference across a small but finite neighborhood of size $L$.

In such a hypothetical material, the flux might look something like this: $q_{nl}(x,t) = -K \frac{T(x+L, t) - T(x-L, t)}{2L}$. Notice the difference! To know the flux at $x$, you need to "peek" at the temperatures at $x+L$ and $x-L$. This is the essence of non-locality.

Now for the magic trick. If this characteristic length $L$ is very, very small, we can use a Taylor series to approximate $T(x \pm L, t)$. When we do this and keep only the most significant terms, our non-local law astonishingly simplifies back into the familiar local Fourier's law, and the resulting evolution equation becomes the standard heat equation. This is a profound revelation: ​​local laws can be excellent approximations of an underlying non-local reality​​. The local world we often perceive might just be the result of "zooming out" so far that the tiny non-local interactions blur into an apparently local effect.

This idea appears in some of the most advanced areas of physics. In the theory of superconductivity, the "true" relationship between the supercurrent and the magnetic field is given by the non-local Pippard equation. It says the current at one point depends on the magnetic vector potential in a whole region around it, defined by a "coherence length" $\xi_0$. The simpler, local London equation, a cornerstone of the subject, is what emerges when we assume the magnetic field varies so slowly that it's essentially constant over this coherence length. Once again, the local is a limit of the non-local.

The finite difference in our heat flow example is just one flavor of non-locality. The most general way for a point to be influenced by all other points is through an ​​integral​​. An equation becomes non-local if, to calculate the rate of change of a quantity at point $x$, you must perform an integral over some (or all) other points $y$.

A beautiful physical picture for this comes from the theory of random walks. Imagine a particle diffusing in a medium. Most of the time, it jiggles around, taking small steps to neighboring positions. This process, in the limit, gives rise to the standard local diffusion equation. But now, suppose there's a small probability that the particle, instead of jiggling, takes a huge leap to a completely different, far-away location. These long-range jumps are known as Lévy flights. To write down the equation for the probability $P(x,t)$ of finding the particle at position $x$, we must account for particles jumping from all other locations $y$ to $x$. This naturally introduces an integral term, resulting in a non-local integro-differential equation.

These kinds of equations often involve special mathematical objects that are inherently non-local. Two of the most important are the Hilbert transform and the fractional Laplacian.

• The ​​Hilbert transform​​, which appears in the study of water waves in the Benjamin-Ono equation, is defined by an integral over all space: $H(f)(x) = \frac{1}{\pi} \text{ P.V.} \int_{-\infty}^{\infty} \frac{f(y)}{x-y} dy$. The "P.V." means we handle the tricky spot where $y=x$ in a specific way, but the main point is clear: the value of $H(f)$ at $x$ depends on the values of $f(y)$ for all $y$.

• The ​​Fractional Laplacian​​, $(-\Delta)^s$, is an even more curious beast. It's like a derivative of a non-integer order! While its precise definition is quite technical, it represents a specific kind of long-range interaction whose strength falls off with distance as a power law. This operator isn't just a mathematical curiosity; it arises naturally from fundamental principles of physics, such as the principle of least action for certain classical fields.

At first glance, these integro-differential equations look fearsome. How can we possibly solve an equation where derivatives and integrals are tangled together? The answer lies in one of the most powerful tools in all of mathematical physics: the ​​Fourier transform​​.

The Fourier transform is like a pair of magic glasses. When you put them on, you stop seeing a function in terms of its value at each position $x$ and instead see it as a sum of waves, each with a specific wavenumber $k$. The miracle is that in this "Fourier space," complicated operations become simple. A local derivative, like $\frac{d}{dx}$, turns into simple multiplication by $ik$. A local PDE becomes an algebraic equation.

What happens to non-local operators? They also become simple multiplication! This is the key.

• A ​​convolution integral​​, like $\int K(x-y)u(y,t) \, dy$, transforms into a simple product: $\hat{K}(k)\hat{u}(k,t)$, where the hat denotes the Fourier-transformed function.
• The ​​fractional Laplacian​​ $(-\Delta_x)^s$ becomes multiplication by $|k|^{2s}$.
• The ​​Hilbert transform​​ becomes multiplication by $-i \, \text{sgn}(k)$.

Let's see this magic in action. Consider a simple non-local transport equation: $\partial_t u + (-\Delta_x)^{1/2} u = 0$. In real space, this is a tough integro-differential equation. But let's put on our Fourier glasses. The equation transforms into: $\frac{d\hat{u}(k,t)}{dt} + |k| \hat{u}(k,t) = 0$ Look at that! For each wavenumber $k$, this is just a simple first-order ordinary differential equation in time. The solution is trivial: $\hat{u}(k,t) = \hat{u}(k,0) \exp(-|k|t)$. The dynamics in Fourier space are beautifully simple. The hard work is then to transform the solution back to real space. For this particular problem, the result is a famous and elegant function known as the Poisson kernel, $K(x,t) = \frac{t}{\pi(t^2+x^2)}$. The Fourier transform acted as a Rosetta Stone, allowing us to translate the difficult non-local problem into a simple one, solve it, and translate it back. This method is the workhorse for analyzing many non-local systems.

So, non-local equations are a fascinating mathematical and conceptual challenge. But what are the real-world consequences? It turns out that allowing actions at a distance can lead to completely new and unexpected phenomena.

One of the most spectacular examples is ​​spontaneous pattern formation​​. Imagine a system described by a concentration $u(x,t)$. Let's say it has a local diffusion term ($D \frac{\partial^2 u}{\partial x^2}$) that tries to smooth everything out, and a local decay term ($-Au$) that pushes the concentration to zero. This system is boringly stable; any lump or bump will just decay away.

Now, let's add a non-local interaction term, an integral that describes how the concentration at one point can enhance the concentration at other points. Suddenly, the whole character of the system can change. A delicate balance can be struck between the local diffusion that wants to erase patterns and the non-local interaction that wants to create them. When conditions are just right, the uniform zero-state can become unstable, and out of a perfectly homogeneous initial state, intricate, stable, periodic patterns like stripes or spots can spontaneously emerge. This type of process, a "Turing-type instability," is thought to be responsible for patterns we see in nature, from animal coats to chemical reactions. The non-locality is the essential ingredient that allows different parts of the system to "coordinate" over long distances to build a coherent structure.

From the microscopic world of superconductivity to the macroscopic emergence of biological patterns, non-local equations force us to think more globally. They remind us that the simple, local rules we learn first are often just one piece of a much richer, more interconnected puzzle. The universe, it seems, has more than a few ways of letting one part know what another is doing, and exploring these connections reveals a profound unity in the laws of nature.

Now that we have grappled with the mathematical heart of non-local equations, you might be wondering, "Where do these curious beasts actually live?" It's a fair question. In our introductory physics courses, we are often taught a beautifully simple, clockwork vision of the universe governed by local laws. The force on an object depends on the fields at its location. The flow of heat at a point depends on the temperature gradient at that same point. It is a "see no further than your nose" philosophy, and it has carried us an astonishingly long way.

But nature, it turns out, is often more subtle and interconnected. The state of a system at one point can, and often does, depend on what is happening far away, or even on what has happened in the past. Non-local equations are not mathematical oddities; they are the precise language we use to describe this deeper layer of reality. They appear whenever we need to account for phenomena involving averages, long-range forces, collective behavior, or memory. Let's take a journey through a few of these domains and see how this idea unfolds.

Imagine a thin metal rod. We know from our basic physics that if you heat one part, the heat will spread out and diffuse along the rod. This is a purely local process, governed by the heat equation we know and love. The rate of temperature change at any point depends only on the curvature of the temperature profile right there.

But now, let’s imagine this rod is embedded in a special "smart" substrate. This substrate has a single goal: to make the rod's temperature as uniform as possible. It has sensors everywhere and a network of tiny heaters and coolers. If it senses a spot on the rod is hotter than the average temperature of the entire rod, it cools that spot down. If a spot is cooler than average, it heats it up. The strength of this heating or cooling is proportional to how far the local temperature deviates from the overall average.

How would we describe the temperature evolution now? We still have the local diffusion, but we must add a new term: a source of heating or cooling that for a point $x$ is proportional to $(\bar{u} - u(x))$, where $\bar{u}$ is the spatial average of the temperature. This average, $\bar{u} = \frac{1}{L}\int_0^L u(y,t) dy$, is the quintessential non-local object. To know how the temperature changes at point $x$, you can't just look at the neighborhood of $x$; you must know the temperature everywhere along the rod to compute the average. Our equation has become non-local.

What is the effect of this non-local control? While standard diffusion slowly washes out any temperature differences, this non-local mechanism acts as a powerful global equalizer. Every part of the system is being told, in unison, how it compares to the whole. The result is a much faster relaxation toward a uniform state. This "mean-field" coupling, where the evolution of a part depends on the average state of the whole, is a recurring theme in physics, from the behavior of magnets to the dynamics of galaxies.

Of course, this extra connectivity comes at a price. When we try to simulate such a system on a computer, the local diffusion part leads to a simple, sparse matrix problem—each point only talks to its immediate neighbors. The non-local integral, however, connects every point to every other point. Our nice, sparse matrix becomes a dense, complicated one, and the computational effort to solve the system skyrockets. This is a practical lesson: interconnectedness can be powerful, but it's also complex to manage.

So far, non-locality seems to be an agent of uniformity, smoothing everything out. But with a slight twist in the ingredients, it can become the engine of creation itself. Some of the most beautiful and complex patterns in nature—the stripes of a zebra, the spots of a leopard, the intricate spirals in a chemical reaction—arise from a competition between local and non-local effects.

Imagine a population of organisms, say, algae in a petri dish. They reproduce locally—where there are algae, more algae will grow. This is "short-range activation." They also diffuse, spreading out slowly. But suppose they also secrete a chemical inhibitor that diffuses very quickly or has a long-range effect, telling other algae "don't grow here!" This is "long-range inhibition."

We can model this with a non-local reaction-diffusion equation. The change in the algae population $u$ at a point $x$ depends on local growth (a term like $u^2$), local diffusion ($D u_{xx}$), and a non-local suppression term that depends on the total amount of inhibitor produced by the whole population, $-\int \phi(x-y) u(y,t) dy$. Here, the kernel $\phi(x-y)$ describes how strongly the inhibitor from point $y$ affects point $x$.

What happens? A tiny, random clump of algae starts to grow. Its local "activator" makes it grow more, but it also starts pumping out the long-range inhibitor. This inhibitor travels further than the algae themselves and prevents other clumps from growing nearby. A bit further away, where the inhibitor is weaker, another clump can start to grow, and it too will enforce its own zone of inhibition. The result of this local activation and non-local inhibition is that a perfectly uniform "soup" of algae becomes unstable. It spontaneously breaks symmetry and forms a stable, repeating pattern of stripes or spots. This mechanism, first envisioned by Alan Turing, shows how non-locality is not a destroyer of structure, but a master architect.

The deeper we look into the structure of matter, the more unavoidable non-locality becomes.

In the quantum world, particles are not tiny billiard balls but blurry waves of probability. The Schrödinger equation tells us how these waves behave. For a single particle in a simple potential (like an electron orbiting a proton), the equation is local. But what about an electron moving through the complex environment of a crystal, or a proton inside a dense atomic nucleus? The potential it feels at one point is a fantastically complicated combination of its interactions with all the other particles.

We can approximate this complex reality with a non-local potential. The Schrödinger equation gains a term that looks like $\int K(x-x') \psi(x') dx'$, meaning the "potential energy" part of the equation at point $x$ depends on the value of the wavefunction $\psi$ everywhere else, weighted by a kernel $K$. This kernel represents the effective interaction with the surrounding medium. One beautiful consequence of this is that the particle's relationship between energy and momentum—its dispersion relation—is altered. It no longer behaves like a free particle with energy proportional to momentum-squared. The non-local interactions effectively change its inertia, giving it a new "effective mass." The very identity of the particle is reshaped by its non-local connection to its environment.

This idea is the bedrock of modern computational chemistry and materials science. In a method called Density Functional Theory (DFT), scientists try to solve the quantum mechanics of molecules and solids to predict their properties. It turns out that simple, local approximations that treat electrons as if they only care about the electron density at their own location often fail spectacularly. They predict that metals don't hold together and that many insulators should be conductors. The fix? Acknowledge the non-local nature of quantum mechanics. A key interaction between electrons, called the exchange interaction, is profoundly non-local. Modern "hybrid" functionals incorporate this by mixing in a non-local term taken from the Hartree-Fock theory. This term is an integral operator that connects each electron's wavefunction with every other electron's wavefunction. It's computationally brutal, but it's the secret sauce. By embracing non-locality, these methods can now predict chemical bond energies, material band gaps, and magnetic properties with astonishing accuracy.

This principle extends from the quantum realm to the engineering world. Classical theories of elasticity are local: the stress at a point depends only on the strain at that same point. This works wonderfully for bridges and airplane wings. But what about a graphene sheet, a single atom thick? Or a carbon nanotube? At these scales, the discrete atomic lattice becomes important. The idea of a "point" is fuzzy. To bridge the gap between the discrete atomic world and a continuous description, engineers use non-local elasticity. In these theories, the stress at a point is a weighted average of the strain in a small neighborhood. This introduces a characteristic length scale into the material's constitution. A fascinating result is that the material becomes "softer" to very short-wavelength deformations. It's a physically intuitive result: trying to wiggle a material on a scale smaller than its atomic spacing shouldn't be as stiff as the bulk material would suggest. Non-locality provides the perfect mathematical framework to capture this multiscale behavior.

So far, our non-locality has been in space. But an equally important form of non-locality is in time. An object whose behavior today depends on its entire past history is obeying a non-local law. We call this "memory."

Think of a viscoelastic material like silly putty or dough. If you pull it slowly, it flows like a thick liquid. If you hit it sharply, it shatters like a solid. Its response depends on the history of the forces applied to it. The internal stress doesn't just depend on the current strain, but on an integral of the strain over all past times, weighted by a "memory kernel" that fades into the past.

This concept can be incorporated into fundamental equations of fluid dynamics. The famous Burgers' equation, a simple model for the formation of shock waves, can be modified with a non-local viscous term that includes an integral over the history of the velocity gradient. This transforms it into a model for a fluid with memory. The presence of this memory, this non-locality in time, can dramatically change the nature of the shock waves, for instance, by determining whether the shock front is smooth or develops oscillations.

One of the most exciting frontiers in science today is the idea that we can discover physical laws directly from experimental data. Suppose we are observing a complex system—perhaps anomalous heat transport in a novel nanomaterial—and we see it evolving in a peculiar way. We suspect a non-local law is at play, but we don't know the rule. We have the solution, $u(x,t)$, but we don't know the kernel, $K(x)$, in the governing equation $\partial u/\partial t = K * u$.

This is an inverse problem. How can we find the hidden law? The magic key, as we've seen, is the Fourier transform. In Fourier space, the complicated convolution integral becomes a simple multiplication: $\partial \hat{u}/\partial t = \hat{K} \hat{u}$. If we can measure $u(x,t)$, we can compute its Fourier transform $\hat{u}(k,t)$ and its time derivative. Then, we can simply solve for the unknown kernel in Fourier space: $\hat{K}(k) = (\partial \hat{u}/\partial t) / \hat{u}$. Once we have $\hat{K}(k)$, we can take the inverse Fourier transform to find the real-space kernel $K(x)$ that governs the system's interactions.

This is a profoundly powerful idea. It means we can listen to the spatiotemporal "symphony" played by a system and, by analyzing its constituent frequencies, deduce the structure of the "instrument" that is playing it. It allows us to move from just describing phenomena to uncovering the fundamental, and possibly non-local, laws that generate them.

From the way heat spreads in a smart material, to the way leopards get their spots, to the fundamental nature of chemical bonds and the strength of nanomaterials, non-local equations are everywhere. They challenge our simple, local-first intuition but reward us with a deeper and more accurate picture of an intricately connected world.