I've just noticed a bug in the Inertia solver: if you call it on an