EkertOverlap
What a violation buys a key LENGTH: Lim et al.'s effective overlap
7/7 passed in 1ms
| Test | What it does | Result |
|---|---|---|
test_ceiling | A certified complementarity is never worth more than an assumed one, so the quality a CHSH value buys is at most the one bit pair_length reads off cbar = 1/2, with equality only at a maximal violation. | ✅ pass |
test_charge | The post-selection charge divides the whole excess over 1/2 by the surviving fraction, so it cannot help, and it cannot move a maximal violation. | ✅ pass |
test_crossing | The paper's Fig. 2 curves cross zero at the two surviving fractions its caption orders from left to right, so a coincidence-counted link, whose surviving fraction is its coincidence probability, is far below both. | ✅ pass |
test_endpoints | The certified overlap is exactly 1 at the local bound and exactly 1/2 at Tsirelson's, the value of a mutually unbiased qubit pair. | ✅ pass |
test_monotone | More violation certifies a smaller overlap across the whole domain, and the overlap never falls below the conjugate-pair value of 1/2. | ✅ pass |
test_secret | Feeding the charged overlap through the entropic uncertainty relation reproduces the paper's own Eq. (2) exactly, and at a maximal violation that fraction is BB84's 1 - 2h(Q), which is what the authors claim. | ✅ pass |
test_transcript | Both halves are the paper's two lemmas at every violation and surviving fraction, transcribed from the paper rather than read off the engine. | ✅ pass |