Definition 1. An R-module P is projective if for any exact sequence ApB→0 and f:P→B, there exists a homomorphism g:P→A s.t. the following diagram commutes:
The homomorphism p:A→B will induce a homomorphism from Hom(P,B) to Hom(P,A), i.e. g is a lifting of f.
First, we have: free module F is projective. Take a basis {fi} of F, from surjectivity of p, ∃ai∈A s.t. p(ai)=f(fi). Then assign fi to ai and generate g from it.
Theorem 1. The functor Hom(P,−) is exact if and only if P projective.
Proof
Proof. For the exact sequence 0→A→B→C→0, we have the sequence 0→Hom(P,A)→Hom(P,B)→Hom(P,C) exact. B→C→0 exact, so Hom(P,B)→Hom(P,C) is surjective if and only if P projective. ◻
The projective module is a direct summand of a free module, i.e. P is projective if and only if ∃M such that P⊕M is free module. We first show the theorem:
Theorem 2. An R-module P is projective if and only if the exact sequence 0→A→B→P→0 splits.
Proof
Proof. Consider the exact sequence $$0\to A\xrightarrow{f}B\xrightarrow{g}P\to 0$$
If P projective, take the exact sequence B→P→0 and idP:P→P, it induces h:P→B with g∘h=idP. So the sequence splits.
Conversely, suppose every exact sequence 0→A→B→P→0 splits. Consider any diagram
take the free presentation 0→Kerπ→FπP→0 of P, by assumption
0→Kerπ→FπP→0
splits. So ∃h:P→F s.t. h∘π=idP. Since F is projective, ∃g:F→B s.t. f∘π=p∘g
Thus f=f∘π∘h=p∘g∘h, P is projective. ◻
Theorem 3. P is projective if and only if it’s a direct summand of a free R-module F.
Proof
Proof. If P is projective, the projective presentation 0→Kerπ→FπP→0 splits, so F=P⊕Kerπ.
Conversely, using similar diagram chasing skill for
◻
A direct corollary is
Theorem 4. Direct summands and sums of projective modules are projective modules.
Proof
Proof. Omitted. ◻
We have shown that for projective module P and any module M, ExtR1(P,M)=0 in the Ext functor and Tor functor
Theorem 5. If ExtR1(P,M)=0 for every R-module M, then P is projective.
Proof
Proof. For a free(projective) resolution
⋯→F1→F0→P→0
Take M=Ker(F0→P), then ExtR1(P,M)=0. Take an element ξ∈ExtR1(P,M) given by the quotient of π:F1→M. Since ξ is zero, there is a map s:F0→M s.t. π=s∘d1, d1:F1→F0. Thus F0=Kers⊕Ker(F0→P)=P⊕Ker(F0→P). ◻
For projective module, it is not necessary to be free. For example, for the ring Z/6Z, the ideal (2) is projective since (2)⊕(3)≅Z/6Z. Actually, any infinitely generated projective module over any Dedekind domain. We proof some weaker result:
Theorem 6. Any finite generated projective module M of a local ring R is free.
Proof
Proof. Take a minimal generating basis m1,…mn of M, here minimal means the minimal number of generators. Let F=i=1⨁nRei, and φ:F→M with φ:∑riei↦∑rimi. Take K=Kerφ. If ∑rimi=0∈M, pass it to the quotient M/mM, ∑rimiˉ=0∈M/mM (Recall R/m⊗RM≅M/mM !). So ri∈m. Hence K⊂mF. Since M is projective, the sequence
0→K→F→M→0
splits. Thus F≅M⊕K. Since F and M have same number of genenrators, dimR/m(R/m⊗RF)=dimR/m(R/m⊗RM)⇒F/mF≅M/mM. We have