Solution (source code)

= Solution

For a normalized <computational history state>, the output projector sees only its final clock component. With $p_+(\zeta)$ the <quantum circuit>'s acceptance probability,
$$
\boxed{\langle\operatorname{hist}(\zeta)|H_{\rm out}|\operatorname{hist}(\zeta)\rangle=\frac{1-p_+(\zeta)}{T+1}.}
$$
This proves the hint, including for a <computational basis> <quantum witness>. Moreover $H_{\rm out}=|-\rangle\langle-|_1\otimes|T\rangle\langle T|$ is a <stoquastic Hamiltonian> term, so adding it with positive coefficient preserves stoquasticity.

There are two substantive problems with the stated promise. First, for any basis <quantum witness> the initialized state has nonnegative amplitudes, and every <permutation matrix> preserves them. Write the output as $|0\rangle|r_0\rangle+|1\rangle|r_1\rangle$, with both vectors entrywise nonnegative. Then
$$
p_+=\frac12+\operatorname{Re}\langle r_0|r_1\rangle\geq\frac12.
$$
The <stoquastic acceptance floor> rules out the printed one-third soundness condition. There are no NO instances of that literal promise. Second, the <ground space> includes histories of arbitrary <quantum witnesses>. It is the maximum over those <quantum witnesses>, not the maximum over basis <quantum witnesses>, that determines the lowest output energy. Even the identity <quantum circuit> accepts each basis input with probability $1/2$, but accepts a $|+\rangle$ <quantum witness> with probability one. Thus a basis-witness soundness bound would not control this Hamiltonian, even if its numerical threshold were repaired.

\b[The requested nontrivial hardness statement therefore needs the standard quantum-witness StoqMA promise], with $1/2\leq b<a\leq1$ and inverse-polynomial gap $\gamma=a-b$. The intended reduction can be completed precisely under that corrected promise. Define the <quantum witness> embedding $J|\zeta\rangle=|\zeta\rangle|a\rangle$ and the <witness acceptance operator>
$$
F=J^\dagger U^\dagger(|+\rangle\langle+|_1\otimes I)UJ.
$$
It is a positive contraction, and $p_{\max}=\lambda_{\max}(F)$. The minimum expectation of $H_{\rm out}$ on the history <ground space> is $\nu=(1-p_{\max})/(T+1)$. This is also consistent with a nonnegative optimal <StoqMA> <quantum witness>: $F$ is entrywise nonnegative, so replacing amplitudes by their absolute values cannot decrease its <quadratic form>.

A uniform <ground-space perturbation bound> is needed because the gap shrinks with <quantum circuit> length. Put $g=(T+1)^{-3}$, and choose a positive inverse-polynomial coefficient
$$
\delta=\frac{g\gamma}{16(T+1)}.
$$
For $P$ the history ground projector and any <unit vector> $\psi=p+q$ with $p=P\psi$, the <operator norm> bound $\|H_{\rm out}\|=1$ gives
$$
\langle\psi|(H+\delta H_{\rm out})|\psi\rangle\geq g\|q\|^2+\delta\nu\|p\|^2-2\delta\|p\|\|q\|\geq\delta\nu-\frac{\delta^2}{g-\delta}.
$$
The last inequality completes the square in $\|q\|$ and uses $0\leq\nu\leq1$. Testing a <ground space> minimizing vector gives the upper bound $\lambda_{\min}(H')\leq\delta\nu$. Since $\delta<g/2$, we obtain
$$
\delta\nu-2\delta^2/g\leq\lambda_{\min}(H')\leq\delta\nu.
$$
YES instances of the corrected promise have energy at most $\delta(1-a)/(T+1)$; NO instances have energy at least
$$
\frac{\delta(1-b)}{T+1}-\frac{2\delta^2}{g}\geq\frac\delta{T+1}\left(1-b-\frac\gamma8\right).
$$
The separation is at least $7\delta\gamma/[8(T+1)]$, an inverse polynomial. The <quantum circuit> and all these Hamiltonian terms have polynomial-size descriptions. This proves <StoqMA> hardness for the <local Hamiltonian problem> variant that permits the nonlocal clock, under the corrected <quantum witness> and acceptance promise.

The printed perturbation formula also has $H_{\rm out}$ where a general perturbation $V$ belongs. Its unspecified $O(\delta^2)$ cannot be treated as uniform in a closing gap. Likewise a circuit-length-independent constant $\delta$ cannot in general satisfy the stated small-perturbation requirement for an arbitrary long <quantum circuit>. The explicit bound above avoids both issues; it does not claim the defective literal promise defines standard <StoqMA>.