Block a user
Gaussian::from_ms overflows for tiny sigma, and three constructors still absorb a negative sign silently
Truncation
1 - w cancellation gives NaN skills from plausible parameters
quality() returns 0 then NaN past ~150 groups: a determinant ratio in linear space
expected_information_gain returns 4.67x its own ln k ceiling
predict_outcome returns probabilities greater than 1: the grid clamps with no detection
A tiny non-zero drift returns a posterior variance 12 000x too small, as
Ok
tuple_max erases NaN, so "NaN is never convergence" is a coin flip across processes
ConstantDrift(-x) is bit-identical to ConstantDrift(x) — seal the field
tests/determinism.rs never exercises the parallel sweep it exists to prove
Three production sites use
std transcendentals, and hypot diverges more than the exp case the rule cites
posterior_of is nondeterministic across processes: float sums over a HashMap
Gaussian::from_ms overflows for tiny sigma, and three constructors still absorb a negative sign silently
Truncation
1 - w cancellation gives NaN skills from plausible parameters
quality() returns 0 then NaN past ~150 groups: a determinant ratio in linear space
tuple_max erases NaN, so "NaN is never convergence" is a coin flip across processes
A tiny non-zero drift returns a posterior variance 12 000x too small, as
Ok
expected_information_gain returns 4.67x its own ln k ceiling
predict_outcome returns probabilities greater than 1: the grid clamps with no detection
