Extensions of flat modules #
In a short exact sequence 0 → M → N → P → 0 of modules over a commutative ring, if M and
P are flat, then N is flat. This allows flatness of a quotient to be established from
simpler subquotients without requiring the ambient module to be flat. Likewise, if N and P
are flat, then so is M: a submodule of a flat module with flat quotient is flat.
Main results #
Function.Exact.flat_of_injective_of_surjective: the middle term of a short exact sequence with flat outer terms is flat.Module.Flat.of_exact_of_lTensor_quotient_injective: for an exact pairM → N → P → 0withNflat, the cokernelPis flat whenM → Nstays injective after tensoring with everyR ⧸ I.Function.Exact.flat_left_of_injective_of_surjective: the first term of a short exact sequence whose other two terms are flat is flat.
The middle term of a short exact sequence with flat outer terms is flat.
A cokernel is flat when the presentation stays injective modulo ideals. Let
f : M → N and g : N → P be exact with g surjective and N flat over R. If
id ⊗ f : (R ⧸ I) ⊗[R] M → (R ⧸ I) ⊗[R] N is injective for every finitely generated ideal I,
then P is flat over R. In terms of Tor, the hypothesis kills the image of
Tor₁(R ⧸ I, P) in (R ⧸ I) ⊗[R] M, and flatness of N makes Tor₁(R ⧸ I, P) → (R ⧸ I) ⊗[R] M
injective. No flatness of M and no injectivity of f is needed.
The first term of a short exact sequence whose middle and last terms are flat is flat.