5
$\begingroup$

If I understand correctly, I can obtain the $O$-cobordism group of $$ \Omega^{O}_3(BO(3))=(\mathbb{Z}/2\mathbb{Z})^4, $$ The 3d cobordism invariants have 4 generators of mod 2 classes, are generated by $$g^3,$$ $$g w_2'(V_{SO(3)}),$$ $$w_3'(V_{SO(3)}),$$ $$g w_1(T)^2.$$

Denote:

The $T$ for the spacetime tangent bundle.

The $O(3)=\mathbb{Z}_2 \times SO(3)$.

The $V_{SO(3)}$ for the vector bundle of $SO(3)$.

Here $g$ is related to the $\mathbb{Z}_2$ generator of $g=H^1(B\mathbb{Z}_2,\mathbb{Z}_2)$.

Questions: What are the corresponding 3-manifold generators of O-co/bordism invariants $\Omega_{O}^3(BO(3))$, for $g^3,$ $g w_2'(V_{SO(3)}),$ $w_3'(V_{SO(3)}),$ $g w_1(T)^2.$?

Hint: I think the manifold generator for $g^3$ is $\mathbb{RP}^3$.

The manifold generator for $g w_1(T)^2$ is $S^1 \times \mathbb{RP}^2$?

What are manifold generators of $g w_2'(V_{SO(3)}),$ and $w_3'(V_{SO(3)})$? Can each manifold generator of 4 generators of mod 2 classes be unique and distinct?

$\endgroup$
5
  • $\begingroup$ Why do you write co/bordism? $\endgroup$ Oct 30, 2018 at 17:50
  • $\begingroup$ It means bordism OR cobordism $\endgroup$ Oct 30, 2018 at 17:51
  • $\begingroup$ Is there a difference? $\endgroup$ Oct 30, 2018 at 17:52
  • 1
    $\begingroup$ they are Potryagin dual. Bordism generators are the manifolds. Cobordism generators are the topological terms (characteristic classes, etc). $\endgroup$ Oct 30, 2018 at 17:58
  • $\begingroup$ I see, thanks. Somehow I always assumed that the two terms were basically synonymous. $\endgroup$ Oct 30, 2018 at 17:59

1 Answer 1

7
$\begingroup$

$\newcommand{\RP}{\mathbb{RP}}\newcommand{\ang}[1]{\langle #1\rangle}\newcommand{\R}{\mathbb R}$A representative of a class in $\Omega_3^O(BO_3)$ is a 3-manifold $M$ together with a principal $O_3$-bundle $P\to M$. Principal $O_3$-bundles are equivalent to rank-3 real vector bundles, so I will use 3-manifolds $M$ together with rank-3 real vector bundles $E\to M$. In this setting, your invariants correspond to products of Stiefel-Whitney numbers of $M$ and $E$, evaluated on the mod 2 fundamental class of $M$.

Specifically, I will use the more standard notation $g^3 = \ang{w_1(E)^3, [M]}$, $gw_2'(V_{SO(3)}) = \ang{w_1(E)w_2(E), [M]}$, $w_3'(V_{SO(3)}) = \ang{w_3(E), [M]}$, and $gw_1(T)^2 = \ang{w_1(E)w_1(M)^2, [M]}$. For brevity I'll write $w_1(E)^3$ for $\ang{w_1(E)^3, [M]}$, etc., and let $S(M,E)$ denote the tuple $(w_1(E)^3, w_1(E)w_2(E), w_3(E), w_1(E)w_1(M)^2)$.

Let $\ell_{\RP^n}\to\RP^n$ denote the tautological line bundle (I'll also do this for $\RP^1 = S^1$); if $x\in H^1(\RP^n;\mathbb Z/2)$ denotes the generator, then $w(\ell_{\RP^n}) = 1+x$. Then, a generating set for $\Omega_3^O(BO_3)$ is

$$\{ (\RP^3, \ell_{\RP^3}^{\oplus 3}), (\RP^3, \ell_{\RP^3} \oplus\underline\R^2), (S^1\times\RP^2, \ell_{S^1}\oplus\underline\R^2), (S^1\times\RP^2, \ell_{S^1}\oplus\ell_{\RP^2}\oplus\underline\R)\}. $$

Specifically, using the Whitney sum formula one can calculate that

  • $S(\RP^3, \ell_{\RP^3}^{\oplus 3}) = (1, 1, 1, 0)$,
  • $S(\RP^3, \ell_{\RP^3}\oplus\underline\R^2) = (1, 0, 0, 0)$,
  • $S(S^1\times\RP^2, \ell_{S^1}\oplus\underline\R^2) = (0, 0, 0, 1)$, and
  • $S(S^1\times\RP^2, \ell_{S^1}\oplus\ell_{\RP^2}\oplus\underline\R) = (1, 1, 0, 1)$.

These four vectors are linearly independent in $\mathbb F_2^4$, so these manifolds generate the bordism group.

If you want manifolds equal to $1$ on one invariant and $0$ on the others, you can take disjoint unions of these manifolds: for example, $(0, 1, 0, 0)$ is represented by the disjoint union of the last three generators. It may be possible to find simpler representatives.

$\endgroup$
2
  • 1
    $\begingroup$ I think the second generator should be $(\mathbb{RP}^3,l_{\mathbb{RP}^3}\oplus\underline{\mathbb{R}}^2)$ and $S(\mathbb{RP}^3,l_{\mathbb{RP}^3}\oplus\underline{\mathbb{R}}^2)=(1,0,0,0)$. $\endgroup$
    – Borromean
    Oct 31, 2018 at 8:02
  • $\begingroup$ @ZheyanWan you're absolutely right. Thanks for the correction! I'll fix it. $\endgroup$ Oct 31, 2018 at 12:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.