ProtocolVariants

Basis keying with a third basis, and with a pair announced in place of one

7/7 passed in 1ms

Test	What it does	Result
test_bias_refused	All four basis-keyed branches refuse a declared bias by name rather than dropping it.	✅ pass
test_engine_agreement	Three bases: the rate is decoy_gain into decoy_bounds into sixstate_rate, at sixstate_sift's share, not a literal 1/3.	✅ pass
test_guards	A sampled pulse train, a finite block and detector afterpulsing are each refused by the two variants.	✅ pass
test_pair_agreement	The pair announcement: sarg_gain's and sarg_yield's conclusive rates into the same protocol-blind inversion, and the rate is sarg_rate with no sifting factor outside it.	✅ pass
test_sift_defaults	q.BasisKeying.sift is 1/bases unless declared, and None under a pair announcement.	✅ pass
test_third_basis_pays	At equal sifting the third basis buys a larger rate on the same decoy bounds, and at the uniform 1/3 against 1/2 it costs more on this span than it buys.	✅ pass
test_two_photon_absent	The two-photon term is reported absent (the Ma-Qi-Zhao-Lo inversion bounds Y1, nothing bounds Y2), so the rate sits below the infinite-decoy one.	✅ pass