Path Integral
Sum Over Histories
A Particle With Too Many Paths
In classical mechanics, the question "how did this particle get from here to there" has exactly one answer. You solve Newton's equations with initial conditions and you get a single trajectory. The particle followed that path. Done.
Richard Feynman, starting as a graduate student in 1941, proposed a radical alternative. What if a quantum particle did not take one path from here to there? What if it took every path – every wiggling, looping, doubling-back trajectory you could draw – all of them simultaneously, and the observed result emerged from how they add up? What if the smooth, single classical trajectory were not fundamental at all, just the survivor of a massive cancellation among infinitely many quantum alternatives?
This is the path integral formulation of quantum mechanics. It sounds mystical. It is not. It is a precise mathematical procedure that reproduces every prediction of standard quantum mechanics and extends to quantum field theory in ways the standard formulation cannot. Understanding it reframes everything else you think you know about quantum physics. Double-slit interference becomes obvious. The classical limit emerges naturally. Feynman diagrams appear as bookkeeping for this machinery. And tunneling gets an explanation so clean it feels almost unfair.
Sum Over Histories
Here is the rule. You want the probability amplitude for a particle to travel from point A at some initial time to point B at some later time. In standard quantum mechanics, you evolve a wave function and read off the value at point B. In the path integral, you do something else entirely. You consider every possible continuous path from A to B. Every single one. The straight line. The curved one. The path that goes up and comes back down. The path that loops through Jupiter and returns. All of them.
Each path contributes a complex number – an amplitude with a phase attached. The phase of a path is determined by a quantity called the action, computed by integrating the Lagrangian along the path and dividing by a fundamental constant called Planck's constant. Different paths have different actions. Different actions mean different phases. Sum all those phased amplitudes and you get the total amplitude to travel from A to B. Square it, and you get the probability.
The point of the procedure is not that a particle literally takes all these paths through physical space. Feynman himself was careful about the interpretation. The path integral is a calculation method. It produces probabilities that match experiment to extraordinary precision. Whether "all paths" is a literal physical picture or a calculational trick is, like many things in quantum mechanics, a question interpretations disagree on. What everyone agrees on is that the machinery works.
Why the Classical Path Wins
If every path contributes, how does classical physics ever emerge? Why does a baseball follow one clear trajectory instead of dispersing over every conceivable arc?
Phases. The phase of each path is its action divided by Planck's constant. Planck's constant is tiny. For a baseball, the action of any reasonable path is enormous compared to Planck's constant, so dividing by it produces an enormous phase that changes wildly as you vary the path. Two slightly different paths have wildly different phases. Sum enough of those phased contributions and they cancel each other out destructively. Almost all paths annihilate in the sum. The only paths that do not annihilate are the ones where the action is stationary – where nearby paths have nearly the same phase. In mathematics, a stationary point of a function is where small changes in input produce only tiny changes in output. Stationary action means nearby paths contribute constructively to the sum.
This is the classical limit, derived from quantum mechanics without needing to assume it. The principle of stationary action, which 18th-century physicists wrote down to describe classical trajectories, is exactly the condition that a single path in the path integral survives the destructive interference with its neighbors. Newton's laws, Lagrangian mechanics, Hamiltonian mechanics, relativistic classical mechanics – all of them are the stationary-phase limit of the underlying path integral. Classical physics is not an approximation to quantum mechanics in a vague sense. It is the precise mathematical limit where only the stationary-action path contributes to the sum.
Quantum mechanics becomes important when the action of typical paths is comparable to Planck's constant. Then the cancellation is no longer overwhelming. Many paths contribute meaningfully. Interference patterns emerge. Tunneling becomes possible. The classical-to-quantum boundary is the boundary between "action so large that only one path survives" and "action so small that many paths do." There is no sharp line. It is a smooth crossover controlled by the size of the action relative to Planck's constant.
The Double Slit Without Mystery
The double-slit experiment is famously weird in the standard telling. A particle passes through two slits simultaneously and interferes with itself. Insert a detector at one slit and the interference vanishes. The particle seems to know whether it is being watched.
In the path integral picture, it is not mysterious. The amplitude to arrive at a given point on the detection screen is a sum over all paths from the source to that point. Paths through the left slit and paths through the right slit are both included. Their contributions carry phases that depend on path length. At certain points on the screen, the phase difference between left-slit paths and right-slit paths is a multiple of two pi and their contributions add constructively. At other points, the phase difference is half-integer multiples and they cancel. Bright and dark bands appear. There is no wave "going through both slits" that you need to picture. There is a sum over paths.
Put a detector at one slit and the situation changes. Now the quantum state of the particle becomes entangled with the state of the detector, and the paths through the two slits end up in different environmental states. When you sum over the entire system, left-slit paths and right-slit paths no longer interfere: they are associated with different macroscopic configurations of the detector. The sum factorizes. You see no fringes. The mystery dissolves into a straightforward consequence of how the path integral handles environment-entangled systems.
Feynman Diagrams Come From Here
The path integral generalizes from quantum mechanics to quantum field theory by promoting "sum over paths of a particle" to "sum over field configurations of a field." In QFT, the thing being summed over is not just position histories of a particle but every possible configuration the field could take across spacetime. Stationary-phase field configurations are the classical field equations. Small quantum deviations around the stationary configuration give you perturbation theory.
Feynman diagrams are the pictorial bookkeeping for this perturbation expansion. Each diagram corresponds to a specific term in the expansion of the path integral around the classical field. Internal lines correspond to integrations over the momenta of virtual field disturbances. Vertices correspond to interaction terms in the action. The rules for translating a diagram into a numerical contribution – the famous Feynman rules – come straight out of the path integral with a bit of combinatorial care. Without the path integral, Feynman diagrams are a collection of tricks. With it, they are a systematic expansion of a well-defined mathematical object.
This is why the path integral is the modern foundation of quantum field theory. The older canonical formulation, based on operators in Hilbert space, still works but gets clumsy in contexts like gauge theories and curved spacetime. The path integral handles these cases naturally. It is also the starting point for non-perturbative techniques like lattice QCD, which compute properties of the strong force by literally sampling field configurations on a discrete spacetime grid. The quark mass inside a proton, the phase diagram of QCD at finite temperature, the confinement of color charge – all come from numerical path integrals.
Tunneling as a Classical Motion in Imaginary Time
Quantum tunneling is one of the most counterintuitive effects in all of physics. A particle passes through an energy barrier it classically cannot cross. In the standard formulation, tunneling is the exponentially decaying tail of a wave function inside the barrier. In the path integral, something beautiful happens.
Rotate the time coordinate by a factor of i. Real time becomes imaginary time. The oscillating phase in the path integral becomes a real exponential. Suddenly the "sum over paths" becomes a weighted sum where paths with smaller imaginary-time action are weighted more heavily – mathematically identical to a statistical mechanics problem. And the potential barrier, which is impossible to cross classically in real time, becomes classically traversable in imaginary time. A particle can roll through the barrier along a specific classical trajectory in this imaginary-time world.
That trajectory is called an instanton. Its existence turns tunneling from a mysterious quantum effect into a concrete calculation: compute the classical action of the instanton, plug it into the formula, and you get the tunneling probability. This technique is used everywhere quantum tunneling matters – nuclear fusion in stars, alpha decay, false vacuum decay in cosmology, magnetic flux tunneling in superconductors. Behind all of them is the same idea: in the path integral, quantum tunneling is classical motion in imaginary time.
This is the kind of insight that makes physicists keep coming back to the path integral. It turns problems that seemed fundamentally quantum into problems that look classical in a different coordinate system, opening entire new calculational toolkits.
Is the Sum Real
Does a particle literally take every possible path? Popular accounts love this phrasing, and it captures something true. But as an interpretation of what is physically happening, it is debated. Some physicists treat the sum over paths as a real ontological statement: all paths exist, and what you observe is the superposed whole. Others treat it as a calculational device that produces the right answers without committing to what is "really" going on between A and B. The interpretation debates around many-worlds, Copenhagen, pilot wave, and relational quantum mechanics all have something to say about which view to take.
The path integral is, in this respect, similar to the wave function. It is an immensely successful mathematical structure that predicts experimental outcomes with unparalleled precision. Whether it describes a literally existing set of histories or is a tool that happens to produce correct probabilities is a question current physics cannot resolve experimentally. What can be said firmly is this: the path integral is the technically most powerful formulation of quantum theory available, and the intuition it builds – that you should think in terms of sums over configurations rather than evolution of single states – has proven fruitful in every branch of physics where quantum effects matter.
The Bigger Picture
The path integral is the hidden engine behind most of modern physics. Quantum field theory is built on it. Feynman diagrams are an expansion of it. Classical mechanics emerges as its stationary-phase limit. Tunneling is captured by its imaginary-time version. Non-perturbative lattice calculations sample it numerically. String theory generalizes it. Quantum gravity research keeps returning to it.
It is also an extraordinary unifying idea. Different areas of physics that seem unrelated – classical trajectories, quantum interference, statistical mechanics near critical points, tunneling through barriers, particle scattering at colliders, black hole evaporation – are all special cases of a single mathematical structure: a weighted sum over configurations. What weight you assign, what configurations you sum over, and which limit you take determine which branch of physics you end up in. Feynman's 1948 insight turned the foundation of quantum theory into something geometric: a space of histories, with amplitudes that add up to produce every physical prediction you care about. A single idea, deep enough to run most of physics, simple enough to sketch on a chalkboard.
