The One-Qubit Wall

Transversal gates are the simplest form of fault-tolerant quantum computation. Each physical qubit in the code block is acted upon independently — no qubit interacts with any other during the gate operation. If one physical qubit fails, the error stays local. It cannot spread. This is the ideal: fault tolerance through isolation.

For a single logical qubit encoded in a stabilizer code, transversal gates can implement a substantial set of operations. The Clifford group — the set of gates that maps stabilizer states to stabilizer states — can be implemented transversally on one logical qubit. Not every gate in the Clifford group, not for every code, but the possibility exists and specific codes achieve it.

The authors prove this cannot extend to two logical qubits. No stabilizer code exists that implements the full Clifford group transversally on more than one logical qubit simultaneously. The wall is at one.

The proof is algebraic. The Clifford group on multiple qubits includes entangling operations — gates that create correlations between the two logical qubits. But transversal gates, by construction, act on each physical qubit independently. An entangling logical gate implemented transversally would require the physical qubits to remain isolated while their encoded logical information becomes correlated. The authors show this is impossible: the algebraic structure of the stabilizer code prevents the transversal gate from inducing the necessary entanglement at the logical level.

The consequence is architectural. Quantum computers cannot achieve fault-tolerant multi-qubit Clifford operations through transversal gates alone. They must use other methods — magic state injection, code switching, lattice surgery — that are more complex, more resource-intensive, and fundamentally different in character.

The simplest fault-tolerance method works for one qubit. It cannot work for two. The wall is not engineering. It is mathematics.