Chapter 37: Topological Quantum Computing

37.1 Anyons: Particles That Remember Their Past

In three-dimensional space, all particles fall into exactly two categories: bosons (integer spin, like photons) and fermions (half-integer spin, like electrons). When you swap two identical bosons, the wavefunction stays the same. When you swap two identical fermions, the wavefunction picks up a minus sign. There are no other options in 3D.

But in two dimensions, something remarkable happens. When you exchange two particles by moving them around each other in a plane, the path they trace matters. In 3D, any exchange path can be continuously deformed into any other, so only the start and end positions matter. In 2D, paths that wind around each other cannot be unwound - they are topologically distinct. This opens the door to particles whose exchange statistics are neither bosonic nor fermionic.

These exotic 2D quasiparticles are called anyons, a name coined by physicist Frank Wilczek in 1982. When you exchange two anyons, the wavefunction can pick up an arbitrary phase factor $e^{i\theta}$, where $\theta$ is neither $0$ (bosons) nor $\pi$ (fermions) but somewhere in between. These are called abelian anyons.

Even more exotic are non-abelian anyons. When you exchange two non-abelian anyons, the system does not merely pick up a phase - it undergoes a unitary transformation in a degenerate ground-state subspace. The order of exchanges matters: swapping anyon A around B and then B around C gives a different result than swapping B around C first. The system's state depends on the entire history of how the anyons were moved around each other.

The sequence of exchanges is called a braid, because if you draw the worldlines of the anyons in spacetime (2D space + 1D time), the trajectories look like strands of a braid. The mathematical framework describing these operations is the braid group, and the unitary transformations generated by braiding form the gate set of a topological quantum computer.

Where do anyons live? They do not exist as fundamental particles in nature (as far as we know). Instead, they emerge as quasiparticle excitations in certain exotic phases of matter - particularly in two-dimensional electron systems at very low temperatures. The fractional quantum Hall effect, discovered in the 1980s, is believed to host abelian anyons, and certain fractional quantum Hall states (notably the $\nu = 5/2$ state) may host non-abelian anyons. Engineered systems, such as topological superconductors, provide another route.

37.2 The Toric Code as a Topological Phase

The toric code, introduced by Alexei Kitaev in 1997, is the simplest and most studied model of topological quantum error correction. It provides a concrete example of how topology protects quantum information, and it forms the foundation for many practical error-correcting codes used today (including the surface code from Chapter 19).

The Setup

Imagine a square lattice drawn on the surface of a torus (a donut shape). Place a qubit on every edge of the lattice. For an $L \times L$ lattice on a torus, this gives $2L^2$ physical qubits. The toric code defines two types of stabilizer operators:

• Star operators $A_s$: For each vertex $s$ of the lattice, $A_s$ is the product of Pauli $X$ operators on the four edges meeting at that vertex. $$A_s = \prod_{e \in \text{star}(s)} X_e$$
• Plaquette operators $B_p$: For each face (plaquette) $p$ of the lattice, $B_p$ is the product of Pauli $Z$ operators on the four edges bordering that face. $$B_p = \prod_{e \in \text{boundary}(p)} Z_e$$

All star operators commute with each other, all plaquette operators commute with each other, and every star operator commutes with every plaquette operator. (This is because any star and any plaquette share either zero or two edges, and $XZ = -ZX$ applied twice gives a net factor of $(-1)^2 = 1$.)

The code space is the simultaneous $+1$ eigenspace of all star and plaquette operators. On a torus, this space has dimension 4, encoding exactly 2 logical qubits. The logical operators correspond to non-contractible loops winding around the two "holes" of the torus.

Topological Protection

A single-qubit error (say, an $X$ error on one edge) creates a pair of anyonic excitations - two plaquettes adjacent to that edge will have eigenvalue $-1$ instead of $+1$. These excitations behave like particles: they can be moved by applying further errors, and they annihilate when brought together. The only way to cause a logical error is to move an anyon all the way around a non-contractible loop of the torus, which requires a chain of $O(L)$ single-qubit errors. As the lattice size $L$ grows, logical errors become exponentially suppressed.

Interactive Demo: Toric Code Stabilizers

The sandbox below demonstrates the stabilizer structure of a small toric code. We prepare a minimal 2x2 toric code patch and measure the star and plaquette stabilizers. In the ideal (error-free) case, all stabilizers should return $+1$. Try introducing an error by adding an x gate on one of the data qubits and observe how the stabilizer measurements change.

The $|0000\rangle$ state is a $+1$ eigenstate of the plaquette stabilizer $Z_0 Z_1 Z_2 Z_3$, so c[1] always returns 0. However, it is not an eigenstate of the star stabilizer $X_0 X_1 X_2 X_3$, so c[0] returns 0 or 1 with equal probability - the measurement itself projects the data qubits into a stabilizer eigenstate. Once both measurements have been performed, the data qubits occupy the code space. Try adding x q[0]; before the stabilizer measurements to introduce a bit-flip error, and observe how the plaquette syndrome (c[1]) changes.

37.3 Fibonacci Anyons and Universal Quantum Computing

Not all anyons are created equal. Abelian anyons (like those in the toric code) can store quantum information topologically, but their braiding operations are limited - they only produce phase gates, which are not sufficient for universal quantum computation. To build a topological quantum computer that can run any quantum algorithm, we need non-abelian anyons whose braiding generates a universal gate set.

The gold standard is the Fibonacci anyon, a theoretical quasiparticle with a remarkable property. Fibonacci anyons have a single nontrivial particle type $\tau$, and when two $\tau$ particles are brought together (fused), the result can be either the vacuum (no particle) or another $\tau$ particle:

This fusion rule is deceptively simple, but its consequences are profound. The number of ways to fuse $n$ Fibonacci anyons grows as the Fibonacci sequence (hence the name): 1, 1, 2, 3, 5, 8, 13, ... More precisely, the dimension of the Hilbert space for $n$ anyons grows as $\phi^n$, where $\phi = (1 + \sqrt{5})/2 \approx 1.618$ is the golden ratio.

A qubit is encoded in the fusion space of a small group of Fibonacci anyons. For example, three $\tau$ anyons have a two-dimensional fusion space (one state where the first pair fuses to $\mathbf{1}$, another where it fuses to $\tau$), which encodes one logical qubit. Quantum gates are performed by braiding the anyons around each other.

Universality from Braiding

The braiding matrices for Fibonacci anyons generate a dense subgroup of the unitary group. This means that by composing enough braids, you can approximate any single-qubit gate to arbitrary precision. Combined with an entangling operation (which also comes from braiding in a multi-qubit encoding), Fibonacci anyons provide a universal gate set purely through braiding.

This is the dream of topological quantum computing: every gate is topologically protected, not just the memory. Unlike the surface code approach, where topological protection helps with storage and error detection but gates still require careful implementation (magic state distillation, etc.), a Fibonacci anyon computer would have fault tolerance built into every operation.

Despite the experimental challenges, Fibonacci anyons remain theoretically important. They prove that in principle, universal quantum computation can be achieved through topology alone. Other non-abelian anyons, such as Ising anyons, are more experimentally accessible but less computationally powerful - their braiding generates only Clifford gates, which must be supplemented by non-topological operations to achieve universality.

37.4 Microsoft's Approach: Majorana-Based Topological Qubits

While Fibonacci anyons remain theoretical, Microsoft has pursued a more experimentally grounded path to topological quantum computing using Majorana zero modes (MZMs) - quasiparticle excitations that emerge at the ends of topological superconducting nanowires. Majorana zero modes are a type of Ising anyon, meaning their braiding alone does not provide universal computation, but they do provide topologically protected storage and certain protected gates.

What Are Majorana Zero Modes?

A Majorana fermion is a particle that is its own antiparticle, predicted by Ettore Majorana in 1937. While no fundamental Majorana fermion has been found in nature, Majorana zero modes can emerge as quasiparticles in engineered condensed matter systems. When a semiconductor nanowire with strong spin-orbit coupling is placed in proximity to a superconductor and subjected to a magnetic field, it can enter a topological superconducting phase. In this phase, Majorana zero modes appear as localized states at the two ends of the wire.

Two Majorana zero modes together encode one fermion, which can be either occupied or unoccupied. This two-state system is the basis for a topological qubit. Crucially, the quantum information is stored nonlocally - it is shared between the two ends of the wire. No local measurement at one end can determine the state, and no local perturbation at one end can flip it. This is the source of topological protection.

The Majorana 1 Processor

In February 2025, Microsoft unveiled Majorana 1, the world's first quantum processing unit powered by a topological core. The chip uses a novel material - an indium arsenide nanowire coated with aluminum to induce superconductivity - and demonstrated the creation and control of Majorana zero modes. Microsoft published results in Nature showing that their devices, when cooled to near absolute zero and tuned with magnetic fields, form topological superconducting nanowires with MZMs at their ends.

The basic qubit design is called a tetron: four Majorana zero modes arranged in a specific geometry, encoding one logical qubit. In July 2025, Microsoft reported that prototype tetron devices could distinguish between two fundamental measurement types tied to MZM behavior, with parity switching times differing by three orders of magnitude - approximately 14.5 microseconds for X-type and 12.4 milliseconds for Z-type measurements.

The Road Ahead

Microsoft's roadmap envisions topological qubits scaling to a million qubits on a single chip - a claim made possible by the inherently small size and digital controllability of topological qubits compared to, say, superconducting transmons. The company is on track to build a fault-tolerant prototype as part of DARPA's Underexplored Systems for Utility-Scale Quantum Computing (US2QC) program.

However, the approach is not without controversy. Independent physicists have raised questions about some of the evidence underlying Microsoft's topological claims, and the history of Majorana research has included high-profile retractions. The key question is whether the observed signatures genuinely reflect topological Majorana zero modes or can be explained by more mundane physics. As of early 2026, the scientific community continues to scrutinize and build upon Microsoft's results.