Tuesday, March 23, 2021

Cotangent bundle of the projective space

 Here is a sketch of an argument why 

Ω:=ΩPkn  is not decomposable into direct sum of OPkn-modules (this answers one of the questions asked in meeting today):

If we show Hom(Ω,Ω)≅k  then we are done, because suppose Ω=F0⊕F1  for some O -modules F0,F1 . Then dim⁡Hom(Ω,Ω)≥dim⁡Hom(F0,F0)+dim⁡Hom(F1,F1)≥2 , which is impossible. 

To prove the claim, the long exact sequence of cohomology of the Euler's sequence0→Ω→O(−1)n+1→O→0 gives the exact sequence 0→H0(Ω)→H0(O(−1)n+1)→H0(O)→H1(Ω)→H1(O(−1))n+1→⋯ 
But since H0(O(−1))=H1(O(−1))=0  we find that H0(Ω)=0,H1(Ω)≅H0(O)≅k  . (This shows that when n=1 we have  Ω≅O(−2), because this is the only invertible sheaf on Pk1  with these cohomology groups.)

Next, tensor the Euler's sequence by O(1)  and then take cohomology. This gives an exact sequence
0→H0(Ω(1))→H0(O)n+1→H0(O(1))→ H1(Ω(1))→H1(O)n+1→⋯ 
By the construction of the Euler's sequence the third arrow from the left is an isomorphism and H1(O)=0  therefore, H0(Ω(1))=H1(Ω(1))=0 .

Finally, apply Hom(−,Ω)  to the Euler's sequence. We get the exact sequence  0→H0(Ω)→H0(Ω(1))n+1→Hom(Ω,Ω)→H1(Ω)→H1(Ω(1))n+1→⋯   
So by what we proved above, Hom(Ω,Ω)≅H1(Ω)≅k  .





No comments:

Post a Comment