A single billiard ball bouncing around a specially designed table may sound like the world’s least impressive computer. Mathematically, however, it can do something far more surprising: simulate a universal Turing machine and, in principle, perform any computation that can be expressed as an algorithm.

That is the conclusion of mathematicians Eva Miranda of the Polytechnic University of Catalonia and Isaac Ramos of ETH Zürich. Their work shows that a carefully constructed two-dimensional billiard with fixed walls can be Turing complete, meaning its dynamics can reproduce the operation of a universal computer. The study was published in the Proceedings of the National Academy of Sciences in 2026.

The trick is not hidden electronics or sophisticated moving parts. It is geometry. The information needed for the computation is encoded in the ball’s position, while the shape of the table determines how that information changes as the ball moves and reflects from the walls.

The result is a striking example of how computation can emerge from an extremely simple physical system.

How can a billiard ball perform a computation?

A normal billiard ball follows a straightforward rule.

It moves in a straight line until it encounters a wall. Under the idealized rules of mathematical billiards, it then reflects so that the angle of incidence equals the angle of reflection.

Miranda and Ramos showed that the walls can be designed so precisely that those simple reflections collectively reproduce the steps of a computation.

In their construction, sections of the billiard correspond to the states of a reversible Turing machine, while corridors connect those states according to the machine’s transition rules.

The ball’s location along a particular segment carries the information about the simulated machine’s tape and reading head. As the particle travels through the table, each movement corresponds to another step in the computation.

In other words, the ball is not “calculating” in the way a processor does. Instead, its trajectory is mathematically equivalent to the operation of a computer.

What is a Turing machine?

A Turing machine is a mathematical model of computation proposed by British mathematician Alan Turing in 1936.

It is deliberately simple. The model contains an imaginary tape divided into cells, a mechanism that reads and writes symbols on those cells, and a set of rules determining what happens next.

At each step, the machine reads a symbol, writes a symbol, moves along the tape and changes its internal state according to its instructions.

The astonishing part is that this minimalist model can capture the logic of general-purpose computation.

A universal Turing machine goes one step further. It can simulate any other Turing machine when given the appropriate instructions and input. Showing that another physical or mathematical system can reproduce a universal Turing machine is therefore a way of establishing that the system is computationally universal.

That is what Miranda and Ramos have demonstrated for their class of two-dimensional billiards.

What makes this billiard different from earlier models?

Billiards have been connected to computation for decades.

Earlier theoretical models showed that collections of billiard balls could imitate logic gates and perform computational operations. Other proposals relied on several interacting particles, three-dimensional arrangements or moving boundaries.

The new result strips away much of that machinery.

The researchers use just one point particle moving in two dimensions inside a fixed table. Nothing has to move except the particle itself. There are no separate computer chips, moving walls or interacting balls carrying out the calculation.

Instead, the complexity has been pushed into the geometry of the boundary.

The researchers describe the construction as a kind of computational labyrinth in which the program is encoded into the shape of the walls.

Where is the information stored?

This is perhaps the most interesting part of the mathematics.

The ball’s position contains the information being processed.

The researchers encode the state of the simulated computation along segments inside the billiard. The ball’s precise location represents information about the simulated tape and the position of its reading head.

The encoding uses a ternary Cantor-set construction, allowing the location of a single particle to represent a potentially unbounded amount of symbolic information in the mathematical idealization.

That sounds almost magical, but there is an important catch.

The information is represented at increasingly fine scales. As the simulated computation becomes more complicated, the required precision becomes extraordinarily demanding.

This is one reason the construction is best understood as a mathematical demonstration rather than a blueprint for a replacement for laptops or CPUs.

Why do the walls have to be so complicated?

The table is not an ordinary rectangular pool table with a few clever pockets.

Its boundary is carefully engineered to perform the logical operations required by the simulated Turing machine.

According to the researchers’ description, different portions of the boundary control how trajectories corresponding to different symbols are routed. Some of the curves contain structure at arbitrarily fine scales.

That complexity is crucial because the billiard has to distinguish between different pieces of information and transform them correctly as the ball moves.

The remarkable part is therefore not that an ordinary pool table can calculate anything. It cannot.

Rather, the mathematical result shows that the geometry of a specially constructed fixed table is sufficient to encode an entire universal computational process.

What does “Turing complete” actually mean?

“Turing complete” does not mean a billiard ball can replace a modern computer.

It means that the system is powerful enough, in theory, to simulate a universal Turing machine.

That distinction matters.

A modern computer performs calculations quickly, reliably and with finite-precision hardware. The billiard construction described by Miranda and Ramos is an idealized mathematical system whose geometry and initial conditions must be specified with extraordinary precision.

So the result concerns computational capability, not practical speed, efficiency or engineering usefulness.

A calculator can multiply two numbers enormously faster than a billiard trajectory could encode a comparable calculation. The mathematical question is whether the system can represent arbitrary computations at all.

The answer, under the researchers’ construction, is yes.

The billiard also inherits the halting problem

Making a physical system Turing complete comes with a famous mathematical price.

That system also inherits the undecidability of the halting problem.

The halting problem asks whether there is a general algorithm capable of determining whether an arbitrary program will eventually stop or continue running forever.

Turing proved that no such universal algorithm exists.

Miranda and Ramos show how a corresponding limitation appears in their billiard system.

They construct the table so that when a simulated computation reaches its halting state, the billiard ball hits a boundary at a 90-degree angle and retraces its path. If the computation does not halt, the trajectory does not return in the corresponding way.

Determining, for every possible case, whether the trajectory will eventually repeat would therefore amount to solving the halting problem, which mathematics tells us is impossible.

Does this mean the ball’s movement cannot be predicted?

Not exactly.

The result does not say that every trajectory is inherently mysterious or impossible to calculate.

Many individual trajectories can be analyzed successfully. What is impossible is a single universal method that can correctly settle every case covered by the system.

This is an important distinction in mathematical physics.

Chaos can make prediction extremely sensitive to tiny differences in starting conditions. Undecidability is different. It places a logical limit on whether a general prediction procedure can exist at all.

As Miranda put it, chaos creates a barrier of precision, while undecidability creates a barrier of logic.

Could we build this billiard table in the real world?

Not in the form required by the mathematical proof.

The construction depends on infinitely fine or increasingly precise geometric structures. A physical table is made from materials with finite dimensions, manufacturing tolerances, and imperfections.

Even the smallest deviation in a wall’s shape could eventually send a trajectory away from the ideal mathematical path.

The same problem affects the particle’s starting position. The theoretical model assumes a degree of precision that physical equipment cannot maintain indefinitely.

That does not weaken the mathematical result. It simply defines what the result is actually saying.

The researchers have shown that universal computation exists within the mathematical dynamics of a particular class of billiard systems. They have not demonstrated a commercially practical mechanical computer built around a pool table.

Why does the result matter beyond billiards?

The broader significance lies in what billiards can teach physicists about computation and dynamical systems.

A billiard is an unusually clean mathematical model. A particle moves freely and interacts with its environment only through reflections from a boundary.

That simplicity makes billiards useful for studying much more complicated physical systems.

The mathematical structures involved can also appear as idealized limits of systems involving particles, gases, steep confining forces and collision dynamics. The researchers describe billiards as a kind of “skeleton” for parts of classical mechanics.

The new result therefore connects two subjects that are often studied separately: the physics of motion and the mathematics of computation.

Could gravity itself become a computer?

The researchers say their work raises a much bigger question.

If a single particle bouncing around a carefully designed table can reproduce universal computation, could naturally occurring gravitational systems possess similar computational complexity?

That question becomes particularly interesting in celestial mechanics, where several interacting bodies can produce complicated and chaotic trajectories.

The researchers are not claiming that the three-body problem has been proved undecidable.

Instead, they are asking whether systems involving a larger number of gravitational bodies might eventually exhibit both the extreme sensitivity associated with chaos and the logical limitations associated with undecidability.

That remains an open research question.

The surprising lesson from one bouncing ball

The billiard-ball result turns an apparently simple physical motion into something much deeper.

A ball moving between walls seems like a problem in elementary mechanics. But once the walls are designed with sufficient mathematical complexity, the trajectory can encode the operation of a universal machine.

The computer, in effect, disappears into the geometry.

That is the most striking lesson from Miranda and Ramos’s work. Computation does not necessarily require silicon, circuits or even multiple moving parts. In the right mathematical system, the rules governing a single moving particle can be rich enough to reproduce the logic of a universal computer.

The catch is equally important: the more computational power the billiard contains, the more demanding its geometry and precision become.

So, no, your local pool hall is not secretly running the world’s next supercomputer.

But in the mathematical universe, one carefully designed ball is enough to make the table itself the machine.