ProofComplete

Ext and Tor - Key Proof

We prove the balance property for Tor, showing it can be computed by resolving either variable.

Theorem5.22Balance of Tor (Complete Proof)

For RR-modules MM and NN: TornR(M,N)≅TornR(N,M)\text{Tor}_n^R(M, N) \cong \text{Tor}_n^R(N, M)

Moreover, both can be computed using a projective resolution of either variable.

Proof

Step 1: Let P∙→MP_\bullet \to M and Q∙→NQ_\bullet \to N be projective resolutions. We show: Hn(P∙⊗RN)≅Hn(M⊗RQ∙)≅Hn(P∙⊗RQ∙)H_n(P_\bullet \otimes_R N) \cong H_n(M \otimes_R Q_\bullet) \cong H_n(P_\bullet \otimes_R Q_\bullet)

Step 2: Consider the double complex P∙⊗RQ∙P_\bullet \otimes_R Q_\bullet with:

  • Horizontal differential: dh=∂P⊗idQd^h = \partial^P \otimes \text{id}_Q
  • Vertical differential: dv=(−1)pidP⊗∂Qd^v = (-1)^p \text{id}_P \otimes \partial^Q at bidegree (p,q)(p, q)

These anti-commute: dhdv+dvdh=0d^h d^v + d^v d^h = 0.

Step 3: The total complex Tot(P∙⊗RQ∙)\text{Tot}(P_\bullet \otimes_R Q_\bullet) has: Totn=⨁p+q=nPp⊗RQq\text{Tot}_n = \bigoplus_{p+q=n} P_p \otimes_R Q_q

with differential d=dh+dvd = d^h + d^v.

Step 4: We can compute homology of Tot\text{Tot} in two ways via spectral sequences:

First spectral sequence: Fix qq and take homology in the pp direction first. IEp,q1=Hph(P∙⊗RQq){}^I E^1_{p,q} = H_p^h(P_\bullet \otimes_R Q_q)

Since QqQ_q is projective (flat) and P∙→MP_\bullet \to M is a resolution: Hph(P∙⊗RQq)={M⊗RQqp=00p>0H_p^h(P_\bullet \otimes_R Q_q) = \begin{cases} M \otimes_R Q_q & p = 0 \\ 0 & p > 0 \end{cases}

Thus the spectral sequence degenerates at E1E^1: IE0,q2=Hq(M⊗RQ∙){}^I E^2_{0,q} = H_q(M \otimes_R Q_\bullet)

and converges to Hn(Tot)H_n(\text{Tot}).

Second spectral sequence: Fix pp and take homology in the qq direction first. IIEp,q1=Hqv(Pp⊗RQ∙){}^{II} E^1_{p,q} = H_q^v(P_p \otimes_R Q_\bullet)

By the same reasoning (using that PpP_p is projective): IIEp,q1={Pp⊗RNq=00q>0{}^{II} E^1_{p,q} = \begin{cases} P_p \otimes_R N & q = 0 \\ 0 & q > 0 \end{cases}

Thus: IIEp,02=Hp(P∙⊗RN){}^{II} E^2_{p,0} = H_p(P_\bullet \otimes_R N)

and converges to Hn(Tot)H_n(\text{Tot}).

Step 5: Since both spectral sequences converge to the same thing: Hn(P∙⊗RN)≅Hn(Tot)≅Hn(M⊗RQ∙)H_n(P_\bullet \otimes_R N) \cong H_n(\text{Tot}) \cong H_n(M \otimes_R Q_\bullet)

This proves Tor can be computed using a resolution of either variable.

Step 6: Symmetry follows from the symmetry of tensor product M⊗RN≅N⊗RMM \otimes_R N \cong N \otimes_R M, which induces an isomorphism on Tor groups.

■
Remark

This proof uses the spectral sequence of a double complex, a technique that appears throughout homological algebra. The key is that projectivity (flatness) makes one direction of the double complex acyclic, causing the spectral sequence to degenerate.

ExampleApplication

To compute Tor1Z(Z/2Z,Z/3Z)\text{Tor}_1^{\mathbb{Z}}(\mathbb{Z}/2\mathbb{Z}, \mathbb{Z}/3\mathbb{Z}), we can use either:

  • Resolution of Z/2Z\mathbb{Z}/2\mathbb{Z}: 0→Z→×2Z→Z/2Z→00 \to \mathbb{Z} \xrightarrow{\times 2} \mathbb{Z} \to \mathbb{Z}/2\mathbb{Z} \to 0
  • Resolution of Z/3Z\mathbb{Z}/3\mathbb{Z}: 0→Z→×3Z→Z/3Z→00 \to \mathbb{Z} \xrightarrow{\times 3} \mathbb{Z} \to \mathbb{Z}/3\mathbb{Z} \to 0

Both give Tor1=0\text{Tor}_1 = 0 since gcd⁡(2,3)=1\gcd(2,3) = 1.