arXiv:2610.37807
Lower Bound for Region of Non-Distillable Magic States
Akul Sethi
Abstract
Magic states are a leading proposal for realizing fault-tolerant and universal quantum computation. At the present moment, much of the theory of magic states is not understood. Even simple questions such as what single qubit states have magic remain open. In Campbell and Browne's seminal paper on bound magic states, they prove the existence of states outside of the stabilizer octahedron which cannot be used to achieve universal computation. In this paper we expand upon their proof and determine a lower-bound on this region.
AI Summary
The paper addresses how far the obstruction to single-qubit magic-state distillation extends beyond the stabilizer octahedron when reductions use a fixed number $n$ of input copies. It makes quantitative a Campbell–Browne argument establishing an undistillable neighborhood for bounded-size protocols. For a state $\rho$ with Bloch vector $B$, define $F=|B_X|+|B_Y|+|B_Z|$ and $m=\min\{|B_X|,|B_Y|,|B_Z|\}/F$. The stated bound is $1<F<\frac{6^n2^{3n-2}+m^n}{6^n2^{3n-2}-m^n}$. Within this range, the paper states that the output of any $n$-to-one Clifford reduction is a convex combination of a stabilizer state and a Clifford rotation of $\rho$. Thus such a reduction cannot increase the output’s $L_1$ Bloch norm beyond the input value. The proof first treats stabilizer-code projections in the positive Bloch octant. It expands $\rho^{\otimes n}$ as a signed combination of product stabilizer states, bounds nonzero projection probabilities below by $2^{-(3n-2)}$, and uses an existing structural lemma guaranteeing contributions to opposite logical Pauli eigenstates. Bounding the negative coefficients then yields the stated threshold. Existing reduction results are invoked to extend the conclusion to general Clifford protocols, including postselection and classical feedforward. The guarantee depends explicitly on $n$ and the Bloch direction: the certified interval shrinks exponentially with $n$ and is empty when $m=0$. It therefore does not establish a single non-stabilizer region that remains undistillable for arbitrarily large input blocks.
Central claim may not hold: The paper's n-dependent exclusion interval shrinks to the stabilizer boundary as n increases, so it does not establish the abstract's claimed region of states unable to achieve universal computation with unrestricted input size. The second main theorem is false as stated: randomized Clifford operations can produce an output that is not a mixture of a stabilizer state and a single Clifford rotation of the input, even when its numerical hypothesis holds.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.