Manuscript Hawk01-B@1 · Hawk01-B.tex · section · cfa976f1

Using leakage to certify a perturbed observer

Regulators, implementation errors, and certificates › Using leakage to certify a perturbed observer · lines 1949–1972

Hawk01-B@1#eq:perturbedleakcertificate, Hawk01-B@1#eq:implementedcertificate

For an approximate complete observer model, suppose a chosen constant state has a certified complementary distance at most $a$ , and the difference between the modeled and physical complete complement is at most $b$ . Then

ϵ_{rec}^{physical} \leq min {1, 2 (a + b)} .

If instead an explicit implemented decoder has modeled error $e$ and the complete accessible-channel allowance is $r$ , then

D_{⋄} (D_{impl} N_{physical}, id_{A}) \leq e + r,

with any additional decoder implementation error included in $e$ or $r$ . These certify different resources: the first proves unrestricted existence, while the second certifies a specified implementation.

Displayed formulas in this passage

Hawk01-B@1#eq:perturbedleakcertificate · line 1955
$ϵ_{rec}^{physical} \leq min {1, 2 (a + b)} .$
Hawk01-B@1#eq:implementedcertificate · line 1963
$D_{⋄} (D_{impl} N_{physical}, id_{A}) \leq e + r,$