Computing Library › Quantum Error Correction
Quantum Error Correction

Concatenated Codes

Concatenation nests a code inside itself, encoding each physical qubit of one code as a logical qubit of another to drive down error rates.

Codes within codes

A concatenated code is built by taking each physical qubit of a base code and replacing it with a logical qubit of another code, and possibly repeating. If the base code has distance d, two levels of concatenation give distance d^2, three levels d^3, and so on. The Shor code is the simplest example: a phase-flip code concatenated with a bit-flip code.

Doubly exponential suppression

Kronos motion — error correction

Concatenation is the historical route to the threshold theorem. At each level, a logical error needs two independent lower-level errors, so if the physical error rate p is below threshold p_th, the level-L error rate scales as p_th (p/p_th)^{2^L}. The exponent 2^L doubles with each level, so the logical error rate falls doubly exponentially in the number of levels, while the qubit count grows only exponentially, giving a net win below threshold.

Trade-offs versus topological codes

Concatenated codes are conceptually clean and give the clearest proof of the threshold theorem, and concatenating small codes with good transversal gates supports fault-tolerant logic. But their thresholds are relatively low, around 10^-4 to 10^-3, because errors must be squeezed through many levels, and their checks are not geometrically local, which strains hardware connectivity.

Topological codes such as the surface code achieve the same exponential suppression by increasing a single code's distance, with local checks and a much higher threshold, which is why they dominate current architectures. Concatenation remains important theoretically, in specific high-rate schemes, and in recent hybrid proposals that concatenate a small code with a surface or LDPC code to combine their strengths.