1
$\begingroup$

I was trying to compute singular homology in integer coefficient of the hypersurface $t^2-1=z^{n}+x(xy-1)$ contained in $A^4$. Can anyone help me computing that? Can anyone tell me some reference where singular homology of some singular affine variety is computed?

$\endgroup$
2
  • 2
    $\begingroup$ The tags suggest that you are looking for the singular homology of the space of complex points of the specified algebraic variety, with the classical topology. You should make that explicit. $\endgroup$ Commented Aug 8, 2021 at 20:36
  • $\begingroup$ I have edited now. Thanks @NeilStrickland $\endgroup$
    – rumpi123
    Commented Aug 9, 2021 at 2:35

1 Answer 1

5
$\begingroup$

Denote the hypersurface $\{(x,y,z,t)\in\Bbb{C}^4\mid t^2-1=z^n+x(xy-1)\}$ by $X$. The equation $x=0$ defines a closed subset $Z$ of $X$ that can be identified with $C\times\Bbb{C}$ where $C$ is the curve $\{(t,z)\in\Bbb{C}^2\mid t^2-1=z^n\}$. The complement $U:=X-Z$ may be identified with $\Bbb{C}^2\times\Bbb{C}^*$ because when $x\neq 0$, one can solve for $y$ as $\frac{t^2-1-z^n+x}{x^2}$. Now consider the long exact sequence in the compactly supported cohomology: $$ \dots\rightarrow H^i_c(U)\rightarrow H^i_c(X)\rightarrow H^i_c(Z)\rightarrow H^{i+1}_c(U)\rightarrow\dots $$ Possible non-zero cohomology groups of $Z=C\times\Bbb{C}$ and $U=\Bbb{C}^2\times\Bbb{C}^*$ are given by $$ H^2_c(Z)\cong H_2(C\times\Bbb{C})\cong H_2(C),\, H^3_c(Z)\cong H_1(C\times\Bbb{C})\cong H_1(C),\, H^4_c(Z)\cong H_0(C\times\Bbb{C})\cong H_0(C);\\ H^5_c(U)\cong H_1(\Bbb{C}^2\times\Bbb{C}^*)\cong \Bbb{Z},\, H^6_c(U)\cong H_0(\Bbb{C}^2\times\Bbb{C}^*)\cong \Bbb{Z}. $$ We deduce that for $i\notin\{2,3,4,5,6\}$ $H^i_c(X)$ is trivial; and $$ H^2_c(X)\cong H^2_c(Z)\cong H_2(C),\\ H^3_c(X)\cong H^3_c(Z)\cong H_1(C),\\ H^6_c(X)\cong H^6_c(U)\cong\Bbb{Z}; $$ and finally, there is a long exact sequence $$ 0\rightarrow H^4_c(X)\rightarrow H_0(C)\rightarrow\Bbb{Z}\rightarrow H^5_c(X)\rightarrow 0. $$ The curve $C:t^2-1=z^n$ in $\Bbb{C}^2$ is connected because the polynomial $t^2-1-z^n=0$ is irreducible (I am assuming $n\geq 1$). So we see that $C$ is a non-compact Riemann surface of finite type. To compute $H_1(C)$ (and hence $H^3_c(X)$), one should find the genus of its compactification and the number of punctures (points added at infinity); see the note below. Moreover, $H_2(C)$ is trivial (which gives us $H^2_c(X)=0$), and $H_0(C)$ is of rank $1$. Thus the last exact sequence may be written as
$$ 0\rightarrow H^4_c(X)\rightarrow \Bbb{Z}\rightarrow\Bbb{Z}\rightarrow H^5_c(X)\rightarrow 0. $$ So to compute the remaining two cohomology groups, one should analyze the middle morphism $H^4_c(Z)\cong\Bbb{Z}\rightarrow H^5_c(U)\cong\Bbb{Z}$. If it is injective, then $H^4_c(X)$ is trivial and $H^5_c(X)$ is a finite cyclic group. Otherwise, both groups are isomorphic to $\Bbb{Z}$.

Note: It is easy to check that $X$ is non-singular. So the compactly supported cohomology groups above result in homology groups by applying the Poincaré Duality: $H^i_c(X)\cong H_{6-i}(X)$.

Note. The curve $C:t^2=z^n+1$ is hyperelliptic. The compactification $\bar{C}$ is of genus $\frac{n-2}{2}$ if $n$ is even, and $\frac{n-1}{2}$ if $n$ is odd. In the former case, $\bar{C}\rightarrow\Bbb{CP}^1: (t,z)\mapsto t$ has two points above $\infty$ while in the latter case it has only one. Consequently, $H_1(C)$ is a free abelian group of rank $2(\frac{n-2}{2})+2-1=n-1$ if $2\mid n$, and of rank $2(\frac{n-1}{2})+1-1=n-1$ if $2\nmid n$. We conclude that $H^3_c(X)\cong H_1(C)\cong\Bbb{Z}^{n-1}$.

$\endgroup$
1
  • $\begingroup$ Thanks for your answer @KhasF. Can you give me some reference where techniques to compute singular homology of affine variety is given? $\endgroup$
    – rumpi123
    Commented Aug 9, 2021 at 4:57

Not the answer you're looking for? Browse other questions tagged .