Local Hamiltonian

In our quest to understand the universe, we face a daunting challenge: how can we describe systems composed of countless interacting particles? Nature, it turns out, employs a powerful simplifying principle known as locality, where things primarily interact with what is next to them. The mathematical embodiment of this principle is the local Hamiltonian, a concept that describes the total energy of a system not as a hopelessly complex global function, but as a manageable sum of local interactions. This idea is far more than a calculational trick; it is a deep truth that underpins much of modern science.

This article explores the profound significance of the local Hamiltonian, bridging the gap between its simple definition and its complex consequences. By understanding this single principle, you will gain insight into the fundamental rules governing quantum systems and their surprising connections to computation, chemistry, and material science.

The following chapters will guide you through this concept. First, in "Principles and Mechanisms," we will uncover the foundational consequences of locality, exploring how it constrains the spread of quantum information, gives rise to the "Principle of Nearsightedness," and allows macroscopic laws to emerge from microscopic rules. Then, in "Applications and Interdisciplinary Connections," we will see how the local Hamiltonian serves as a Rosetta Stone connecting physics to computer science, enabling powerful quantum simulations, and allowing us to model and engineer new, exotic states of matter.

Imagine trying to understand the workings of a nation's economy by developing a theory that explicitly links the financial decisions of a baker in Paris to those of a fisherman in Hokkaido. The task would be impossible. The world, thankfully, doesn't seem to work that way. The baker is primarily concerned with the price of flour from the local mill, and the fisherman with the demand at the nearby market. Their interactions are, for the most part, ​​local​​.

It is one of the most profound and fortunate facts of our universe that the fundamental laws of physics exhibit this same character. The total energy of a vast system—be it a magnet, a galaxy, or a block of silicon—isn't a horribly complex function of every single particle at once. Instead, it is built up brick-by-brick from simple, local pieces. In physics, we call the function that governs the energy of a system its ​​Hamiltonian​​, and this crucial property of locality gives rise to the ​​local Hamiltonian​​.

A local Hamiltonian is one that can be written as a sum of terms, where each term involves only a small, spatially-contained group of particles that are "neighbors." Perhaps the most famous illustration is the Ising model of a magnet. We can picture a grid of microscopic atomic "spins," each pointing either up ($S=+1$) or down ($S=-1$). The total energy, $E$, is governed by the simple rule:

The magic is in the symbol $\langle i, j \rangle$. It tells us to sum only over pairs of spins that are immediate neighbors. The total energy is just the sum of these local interaction energies. A spin only "talks" to its neighbors, not to a spin on the far side of the material. This is what we mean by a local Hamiltonian. It is the mathematical embodiment of the principle that "things interact with what's next to them."

Now, what we consider a "neighbor" can sometimes be subtle. If we change our description of the system—for instance, mapping spin degrees of freedom to fermion particles via a mathematical tool called the Jordan-Wigner transformation—an interaction between two nearby spins can sometimes look like a much more complicated interaction between the fermions. Yet, remarkably, the underlying locality of the physics often re-emerges, with the complex-looking terms canceling out to restore a simple, local description in the new language. Nature, it seems, conspires to keep its rulebook simple and local.

If the rules are local, what are the consequences for how a system behaves over time? In quantum mechanics, the Hamiltonian dictates evolution. A system's state $|\psi(0)\rangle$ evolves to $|\psi(t)\rangle$ via the time evolution operator $U(t) = \exp(-iHt/\hbar)$. If the Hamiltonian $H$ is local, what does this mean for $U(t)$?

Let's consider one of the most counter-intuitive features of quantum mechanics: ​​entanglement​​, the spooky connection that can exist between two or more particles, no matter how far apart they are. Can we create this non-local connection using only local means?

Imagine a system of two qubits, A and B, that start off in a simple, unentangled (or "separable") state, like $|\psi(0)\rangle = |\psi_A\rangle \otimes |\psi_B\rangle$. Suppose we have a Hamiltonian that represents only local operations—we can prod and poke qubit A, and we can prod and poke qubit B, but we can't introduce a direct interaction between them. Such a Hamiltonian has the form $H = H_A \otimes I_B + I_A \otimes H_B$, where $H_A$ acts only on A and $H_B$ acts only on B.

Because the two parts of the Hamiltonian, $(H_A \otimes I_B)$ and $(I_A \otimes H_B)$, commute with each other, the exponential neatly factorizes:

The total evolution operator is just a tensor product of the individual, local evolution operators! When we apply this to our initial state, we find:

The final state is still a simple product state. It is not entangled. This is a profound conclusion: ​​local operations cannot create entanglement​​. To generate entanglement between two systems, you need a Hamiltonian that includes a direct interaction term, a piece like $H_{AB}$ that couples them together. Locality fundamentally restricts the kinds of quantum correlations that can be generated. This doesn't mean local Hamiltonians can't affect entangled states—they can, and the evolution can be quite rich—but they are constrained in what they can do, and they cannot, by themselves, pull entanglement out of a hat.

The local nature of the Hamiltonian has an even deeper consequence, one that makes much of modern science possible. In a quantum system, everything is, in principle, connected. An electron in your fingertip is part of a single, gigantic quantum state that includes every other electron in the universe. How can we possibly hope to understand the chemistry of a single molecule without solving for the entire cosmos?

The answer lies in the ​​Principle of Nearsightedness​​. This principle, championed by the physicist Walter Kohn, states that for a huge class of materials—namely, insulators and semiconductors, which have an energy "gap"—the effects of local actions decay incredibly quickly with distance. More formally, the one-particle density matrix, $\gamma(\mathbf{r}, \mathbf{r}')$, which measures the quantum correlation between a point $\mathbf{r}$ and a point $\mathbf{r}'$, decays exponentially as the distance $|\mathbf{r} - \mathbf{r}'|$ grows.

An electron in a diamond crystal is profoundly "nearsighted." It is strongly influenced by the carbon atoms in its immediate vicinity but is blissfully ignorant of what's happening more than a few nanometers away. This exponential decay is a direct consequence of having a local Hamiltonian and an energy gap. The gap effectively "insulates" regions of the material from one another.

This principle is the bedrock of modern computational chemistry. When chemists want to calculate the properties of a large molecule, they don't need to solve for all the Avogadro's number of electrons at once. Because of nearsightedness, they can truncate their problem, focusing on specific "local domains" with the confidence that the errors they introduce will be exponentially small. This is what makes methods like DLPNO-CCSD(T) feasible, allowing us to design new drugs and materials.

Interestingly, nearsightedness is not a universal law. In metals, the absence of an energy gap changes the game. Correlations still decay, but only as a slow power-law. This "farsightedness" is one reason why metals are so much more complex to model than insulators, and it highlights the special role that a local Hamiltonian, combined with a gap, plays in making our world analyzable.

Let's zoom out further. The world we experience is not a maelstrom of quantum particles; it's a world of solid objects, flowing fluids, and sturdy materials. We have entire fields of engineering and continuum mechanics that describe this macroscopic world with stunning accuracy, using concepts like stress, strain, and elasticity, without ever mentioning a Hamiltonian. How do these smooth, continuous laws emerge from the discrete, local rules of the quantum world?

Once again, the local Hamiltonian is the hero of the story. Because the fundamental interactions are short-ranged, it becomes possible to "average over" or "coarse-grain" the microscopic details when looking at the system on a larger scale. This is the central idea behind powerful multiscale simulation techniques like the ​​Quasicontinuum (QC) method​​.

Imagine you want to simulate how a piece of metal deforms under stress. Instead of tracking the trillions of atoms, the QC method allows you to select just a few "representative atoms." The positions of all other atoms are then interpolated from these representatives. This approximation only works if the deformation is smooth over the scale of many atomic spacings. The validity of this entire scheme rests on a separation of scales: the microscopic interaction range of the local Hamiltonian must be much smaller than the scale over which we are coarse-graining, which in turn must be much smaller than the scale over which the material's deformation changes.

In essence, the locality of the underlying Hamiltonian allows the material to "forget" its atomic-scale discreteness when responding to large-scale forces. This is how the beautifully smooth laws of continuum mechanics emerge from the granular quantum reality. It is the reason an engineer can design a bridge using calculus, not quantum field theory.

The principle of locality is not just a convenient simplification; it is a deep thread that weaves together some of the most beautiful tapestries in modern physics. When we combine the local Hamiltonian with other fundamental principles, astonishing phenomena emerge.

• ​​Locality and Symmetry:​​ Consider a system with a local Hamiltonian that also possesses a continuous global symmetry—for instance, all spins are free to rotate together in any direction. If the system's ground state "spontaneously" breaks this symmetry by picking one specific direction (like in a ferromagnet), the ​​Nambu-Goldstone theorem​​ demands the existence of system-wide ripples of zero energy: massless Goldstone modes. These modes are the system's way of communicating the broken symmetry over long distances. The theorem's derivation hinges critically on the locality of the Hamiltonian; it's the local nature of the interactions that forces this collective behavior to appear.

• ​​Locality and Thermalization:​​ One of the great mysteries of physics is the arrow of time and why things thermalize. Why does a cup of hot coffee always cool down to room temperature? In the quantum realm, the ​​Eigenstate Thermalization Hypothesis (ETH)​​ provides a startling answer, and locality is at its heart. ETH suggests that for chaotic systems with local Hamiltonians, even a single, high-energy eigenstate of the entire system looks thermal to any local observer. Why? Because any small subsystem is only coupled to its immediate neighbors. The rest of the vast system acts as a featureless heat bath, defined only by its energy density. The local subsystem can't tell that it's part of one perfect, stationary quantum state; all it "sees" are the random-looking thermal fluctuations from its environment. Locality ensures that what happens globally is irrelevant to what is measured locally.

From the structure of a magnet to the feasibility of chemistry, from the strength of materials to the very nature of symmetry and time, the principle of the local Hamiltonian provides the unifying framework. It is Nature's quiet insistence that to understand the whole, we must first start by understanding the parts and how they talk to their neighbors. It is this profound simplicity that makes our fantastically complex universe, in the end, comprehensible.

We have spent some time getting to know the local Hamiltonian, a concept of beautifully deceptive simplicity. The idea that the total energy of a system is just a sum of parts, where each part only involves a few nearby players, seems straightforward enough. One might be tempted to think this is a mere calculational convenience, a physicist's trick to make an impossibly large problem slightly more manageable. But to think that would be to miss the forest for the trees.

The principle of locality is one of nature's most profound truths, and the local Hamiltonian is its mathematical embodiment. What we are about to see is that this single idea is a Rosetta Stone, allowing us to translate questions between seemingly distant realms of thought: the abstract logic of a computer, the tangible substance of a magnet, the intricate dance of electrons in a molecule, and even the ultimate limits of what we can possibly know. Let us now embark on a journey to see how this one key unlocks so many different doors.

Have you ever looked at a flock of birds, a crystal, or a bustling city and wondered how such intricate order arises from simple, local interactions? Physics and computer science have been asking a similar question, and their answers have converged on the local Hamiltonian.

At its heart, finding the ground state of a Hamiltonian is an optimization problem. Nature, in its own inimitable way, "solves" for the configuration of lowest energy. This sounds suspiciously like what a computer does when it searches for the optimal solution to a problem. Could we co-opt nature's problem-solving ability?

Imagine we have a purely logical problem, like the Boolean Satisfiability Problem (SAT), where we need to find an assignment of TRUE or FALSE values to a set of variables to satisfy a list of logical clauses. We can be devilishly clever and translate this into the language of physics. We can design a quantum system, typically of interacting spins (qubits), where each possible assignment of variables corresponds to a particular state of the spins. Then, for each logical clause, we invent a local Hamiltonian term that "penalizes" the system with extra energy if that clause is violated.

The total Hamiltonian is then the sum of all these penalty terms: $H = \sum_j H_j$. If the logical formula is satisfiable, there exists a spin configuration (a variable assignment) that violates no clauses. This configuration is the ground state of our Hamiltonian, and its energy is exactly zero. If the formula is unsatisfiable, then no matter what we do, at least one clause will be violated. The ground state energy will then be some positive value, corresponding to the minimum number of clauses that must be broken. By preparing this physical system and finding its ground state, we have, in effect, solved the logical problem. The physics of local interactions has become computation.

This connection runs much, much deeper. It forces us to ask: just how hard is it to find the ground state of a general local Hamiltonian? Computer scientists classify problems by their difficulty. Problems solvable by a classical computer in a reasonable (polynomial) amount of time are in the class P. Harder problems, whose solutions can be checked quickly but not necessarily found quickly, are in NP. The quantum analogue of NP is a class called QMA, for Quantum Merlin-Arthur. In this scenario, an all-powerful but untrustworthy wizard (Merlin) gives you a quantum state, claiming it is the ground state. Your job, as a quantum King Arthur, is to perform a measurement on this state to verify, in a reasonable amount of time, whether it truly has low energy.

Here is the astonishing climax of this story: the problem of finding the ground state energy of a local Hamiltonian is not just in QMA; it is ​​$QMA$-complete​​. This means it is among the absolute hardest problems in the entire class. Any problem a quantum computer could hope to verify with the help of a wizard can be rephrased as a local Hamiltonian problem. The local Hamiltonian isn't just an example of a hard quantum problem; it is the archetypal hard quantum problem.

The frontier of this field is a bold idea known as the Quantum PCP Conjecture. It speculates that the local Hamiltonian problem is hard in a particularly robust way: that it's hard even to tell the difference between a system whose ground state energy is nearly zero and one whose ground state energy is a significant fraction of the maximum possible energy. Should this conjecture prove true, it would have profound implications for the stability of quantum information and our understanding of entanglement, revealing an even more intricate link between the fabric of spacetime and the nature of computation.

If finding the ground state of a general local Hamiltonian is so intractably hard, how do physicists and chemists ever hope to understand quantum materials? Is it a lost cause? Not at all! The key is that nature is often kind. The Hamiltonians describing many real materials, while local, have special properties that make their ground states far simpler than the worst-case scenario might suggest.

The central challenge in simulating a quantum system is the sheer vastness of its state space. The number of parameters needed to describe a quantum state of $N$ particles grows exponentially with $N$. However, the ground states of many physically relevant local Hamiltonians do not explore this entire space. They live in a tiny, special corner. Why? Because of locality!

In a one-dimensional system with a local Hamiltonian and an energy gap (a finite energy cost to create an excitation), a remarkable thing happens. The interactions, being local, cannot generate entanglement between far-flung parts of the system. The entanglement in the ground state primarily exists across the "boundary" between any two halves of the system. In 1D, this boundary is just a point! This leads to a principle called the ​​area law​​ of entanglement: the entanglement entropy of a subregion is constant, independent of its size.

This single insight is the foundation for one of the most powerful numerical methods in theoretical physics: the Density Matrix Renormalization Group (DMRG). The area law guarantees that the ground state can be efficiently represented by a structure called a Matrix Product State (MPS). You can think of an MPS as a kind of quantum ZIP file; it's a compressed representation that works exceptionally well for states with limited entanglement, like the ground states of 1D gapped local Hamiltonians. This allows us to simulate systems with thousands of particles, a feat that would be unthinkable otherwise. The locality of the Hamiltonian directly translates into a compressible structure for its ground state.

The story changes dramatically when we look at the dynamics of a system. Imagine preparing a simple, unentangled state and then suddenly "switching on" a local Hamiltonian—a process called a global quench. Entangled pairs of quasiparticles are created everywhere and begin to fly apart. The entanglement between two halves of the system will grow, but it cannot do so instantaneously. It grows at a rate limited by the maximum speed at which information can propagate, a speed set by the Hamiltonian itself. The result is that entanglement grows linearly with time.

This has a profound and sobering consequence for simulations. To capture this ever-growing entanglement, the "size" of our MPS quantum ZIP file (its bond dimension) must grow exponentially with time, $D(t) \sim \exp(\kappa t)$. The very locality that makes static ground states easy to handle makes long-time dynamics exponentially difficult. We can see, quite literally, how quantum information, seeded by the Hamiltonian, spreads through the system and eventually overwhelms our classical computers.

What about higher dimensions? The same principles apply, though the technology is more complex. In two dimensions, we use a generalization of MPS called Projected Entangled-Pair States (PEPS). Again, the network-like structure of a PEPS directly mirrors the local connectivity of the Hamiltonian, and they are our best language for describing the complex ground states of 2D quantum matter.

The ultimate application of local Hamiltonians is in a direct dialogue with nature. Physicists write down local Hamiltonians not just to compute with, but to model the world around us and understand how the simple rules of interaction between neighboring particles give rise to the rich, emergent phenomena we observe.

Consider the beautiful case of ​​spin ice​​ materials. In certain crystals with a "pyrochlore" lattice structure—a network of corner-sharing tetrahedra—the magnetic ions behave like tiny spins. The allowed interactions between nearest-neighbor spins are severely constrained by the lattice's symmetries. When a particular local interaction, an Ising-like term, dominates, something amazing happens. The system cannot find a single, perfect ground state. It is geometrically frustrated. The lowest-energy states are a vast collection of configurations that obey a simple local constraint on each tetrahedron: two spins must point in, and two must point out. From a simple, symmetric local Hamiltonian, a complex, macroscopically degenerate manifold of states emerges, complete with its own effective "ice rules" and exotic excitations that behave like magnetic monopoles. Add small quantum mechanical terms to the Hamiltonian, and you enter the realm of "quantum spin ice," a dynamic, fluctuating state of matter that is one of the most exciting frontiers in modern magnetism.

This process can also be run in reverse. Instead of starting with a Hamiltonian and finding its ground state, what if we start with a quantum state that has properties we find interesting—perhaps it describes a novel superconductor or a state with topological order for quantum computing—and ask: what local Hamiltonian would have this state as its ground state? This is the concept of a ​​parent Hamiltonian​​. For a large class of states, including the MPS and PEPS we discussed, it is possible to systematically construct a local, frustration-free Hamiltonian that, by design, has our desired state as its unique ground state. This is an incredibly powerful tool for theoretical physicists, akin to being the lawgiver for a toy universe: you write the local laws of physics to engineer a specific reality.

Finally, this brings us to the bedrock of our physical world: quantum chemistry. The electronic structure Hamiltonian, which governs the behavior of atoms and molecules, is the quintessential local Hamiltonian, with terms describing the kinetic energy of electrons and their pairwise Coulomb interactions. To simulate these systems on a quantum computer, we must first map the fermionic operators of electrons onto the qubit operators of the machine. Here again, locality is king. Mappings like the Jordan-Wigner transformation preserve the spatial locality of a 1D chain of orbitals, which is ideal for simulation methods like DMRG. Other mappings, like the Bravyi-Kitaev transformation, can reduce the total number of qubit operations needed for some terms but may scramble the notion of proximity, turning a physically local interaction into a highly non-local one on the quantum computer. Choosing the right representation is a crucial engineering decision, dictated entirely by how the structure of the local Hamiltonian translates to the architecture of the simulator.

From the abstract heights of computational complexity to the concrete physics of a magnet and the practical challenges of simulating a molecule, the local Hamiltonian is the common thread. It is the language we use to describe our world, to simulate it, and to understand the ultimate limits of that simulation. It is a testament to the profound idea that from local simplicity, global complexity is born.