The Setting
A conjugated product projection can connect sectors distinguished by their number of local mismatches. The localization question concerns the transition from no mismatches to a growing mismatch count when the conjugating circuit has fixed depth.
A classical bit belongs to ✦, and a word assigns a bit to each of the wires.✦ A quantum operator on qubits is a complex matrix whose rows and columns are indexed by ✦; a single-qubit operator is a complex matrix whose rows and columns are indexed by .✦ For a family of single-qubit operators , the tensor product has entries (1)✦ The norm throughout is the Euclidean operator norm.
For a depth , the enlarged reflection-circuit class consists of the operators obtained recursively from layers of single-qubit unitaries and layers of reflections by (2)✦ In (2), the rightmost matrix acts first.
Product Projections and Mismatch Sectors
For a one-qubit vector , the outer product is the single-qubit operator defined by (3). (3)✦ A count on qubits is product projection data consisting of a support and vectors for each wire, with (4)✦ For a count on qubits with support and vectors , the zero-count product projection is the following operator, where is the identity on one qubit: (5)✦
For a count on qubits with support and vectors , and a set of wire indices, the mismatch pattern projection is the operator (6)✦ The high spectral projection of a count on qubits, with support and real cutoff , is the sum of the pattern projections in (6) selected by (7)✦
For operators and on the same qubits, the conjugate of by is the operator defined by (8). (8)✦
The Claim
At depth , with real exponents , let be the uniform localization assertion: the estimate (9) holds for all sufficiently large , uniformly over and counts on qubits.✦ The full localization statement requires this assertion at every depth for every pair .✦
Theorem 1. Fix a depth and real exponents . There is a threshold such that, for every , every , and all counts on qubits, (9)✦ For every depth and real exponents , (10)✦
What Is Checked
The localization statement (9) ranges over the recursive circuit classes (2) and normalized product projection data (4), using the zero and high projections (5) and (7) and conjugation (8).