Von Neumann Algebra

In mathematics there are objects that can be added and multiplied: numbers, for example. But there are also computational rules that are themselves treated as objects. Such a rule takes something as input and returns something else, for instance “rotate this arrow by 90 degrees.” Two such rules can be carried out one after the other, and the result is again a rule. A von Neumann algebra is a collection of such computational rules that is self-contained: if you combine two members of the collection, you stay within the collection. In addition, one requires that the collection also remain stable under limiting processes, that is, when you chain together infinitely many steps. It is named after John von Neumann, who studied it in the 1930s together with Francis Murray.

The Foundation of Quantum Mechanics

Von Neumann did not develop this theory out of pure curiosity. Around 1930, quantum mechanics was physically successful but mathematically untidy. Physicists were working with quantities whose order suddenly mattered. Measuring the position and momentum of a particle in succession yields something different than measuring them in reverse order. Von Neumann found the right language for this: measurable quantities are computational rules, and their multiplication is not commutative.

Out of this language grew an entire field of research. Von Neumann and Murray sorted the algebras into types, called Type I, II, and III. Type I corresponds to what one encounters in simple quantum systems with few particles. Type III long seemed like a mathematical curiosity, but today it describes quantum field theories and even regions of spacetime in work on gravity.

One byproduct is especially remarkable. In Type II algebras there are meaningful dimensions that are not whole numbers. There, one can speak of a space of dimension 1.5. That sounds like a computational error, but it is consistent, and it is one of the reasons the field fascinates mathematicians.

Closed Under Limits

The starting point is a space with infinitely many directions, a so-called Hilbert space. You can picture this as a coordinate system that has not three but infinitely many axes. The computational rules on this space are called operators. They stretch, rotate, and mix the directions.

A von Neumann algebra is now a set of such operators with three properties. First: sums and products of its members again belong to it. Second: for every operator, its mirror image, the adjoint operator, also belongs to it. Third: if a sequence of members approaches a limit, that limit also lies within the set. This third condition distinguishes it from the related C*-algebra, for which a weaker notion of limit suffices.

There is an elegant short form of this definition. Von Neumann showed that such an algebra consists precisely of the operators that commute with all operators of a certain other set. This statement is called the double commutant theorem. It allows a very abstract object to be described purely in terms of commutativity.

From Physics to the Quantum Computer

In everyday life, one does not encounter this term. It surfaces when fundamental research makes the news. Anyone reading reports about quantum computers comes across terms like entanglement or quantum channel. Their precise definitions stem from the theory of operator algebras.

This is concretely useful when it comes to the question of how much information a quantum system contains. An important measure for this is relative entropy, and its mathematically rigorous formulation for infinite systems lives within von Neumann algebras. Companies and institutes working on quantum error correction draw on this framework. Related algebras also play a role in research into so-called topological qubits.

A common mistake is to confuse the name with the von Neumann architecture. That architecture is the blueprint for ordinary computers, in which program and data reside in the same memory. Both go back to the same person, but they have nothing to do with each other in substance. Anyone who encounters “von Neumann” in a text about chips is almost always reading about the architecture, not the algebra.