7
$\begingroup$

Consider the semiring $$\mathbb{N}[H,H^{-1}]/(H^p+H^q = H^{p+q}+1)_{p,q \in \mathbb{Z}}.$$ Is it finitely presentable? Is there any simplification of the relations (except for $p \geq q \geq 0$)?

Topological background. Whereas the semiring of vector bundles on the scheme $\mathbb{P}^1_{\mathbb{C}}$ is just $\mathbb{N}[H,H^{-1}]$ (old result by Dedekind-Weber, 1892, which has been rediscovered many times), the semiring of vector bundles on the topological space $\mathbb{C}\mathbb{P}^1$ seems to be the semiring above, where $H$ represents the tautological bundle. The relations follow from Example 1.13 in Hatcher's VBKT. The reason is that $\begin{pmatrix} z^p & 0 \\ 0 & z^q \end{pmatrix}$ is homotopic to $\begin{pmatrix} z^{p+q} & 0 \\ 0 & 1 \end{pmatrix}$ in $\mathrm{Map}(S^1,\mathrm{GL}_2(\mathbb{R}))$. The $K$-theory ring is just $K(\mathbb{C}\mathbb{P}^1)=\mathbb{Z}[H]/(H^2+1=2H)$.

$\endgroup$
4
  • $\begingroup$ Very good question. Any element can be uniquely presented in the form $N$ or $N+MH^p$ with $p>0$. Multiplication and addition in this basis seems tricky though, e.g. $(2+3H^3)(4+5H^2) = 44 + H^{131}$ or something like that. $\endgroup$ – Piotr Achinger Apr 16 '14 at 7:09
  • $\begingroup$ @Piotr: $H^{-1}$ cannot be written like that. I think the normal form is rather $\mathbb{N} \sqcup \{H^p + n : p \in \mathbb{Z} \setminus \{0\}, n \in \mathbb{N}\}$. $\endgroup$ – Martin Brandenburg Apr 16 '14 at 8:12
  • $\begingroup$ right, I forgot I cannot subtract ;) $\endgroup$ – Piotr Achinger Apr 16 '14 at 8:20
  • $\begingroup$ Regarding history, in 1886 del Pezzo classified surfaces of minimal degree, and this was extended to minimal varieties of all dimensions by Bertini. So the "Birkhoff-Grothendieck splitting lemma" goes back further than Dedekind-Weber. $\endgroup$ – Jason Starr Apr 18 '14 at 12:11
3
$\begingroup$

Some observations: Consider the monoid $P = \{ (N, p)\in\mathbb{N}\times\mathbb{Z} \,:\, N = 0 \Rightarrow p = 0\}$ with the semiring structure $(N, p)(M, q) = (NM, Mp+Nq)$. I claim that the map $(N, p)\mapsto N-1+H^p$ is an isomorphism between $P$ and your semiring. In particular, the additive monoid underlying your semiring is not finitely generated.

EDIT. The conclusion was obviously wrong, I deleted it.

$\endgroup$
2
  • $\begingroup$ Isn't part of your conclusion correct - that $H^p$ for $p \in \mathbb Z$ must be monomials in the generators? So the generators are $H^r$ for some set of $r$ that generate the semigroup $\mathbb Z$. Now infinitely many distinct relations involve a sum of at least $2$ of these things, and two such relations cannot combine to form a third, so there is no finite presentation. $\endgroup$ – Will Sawin Apr 16 '14 at 16:35
  • $\begingroup$ The condition in the def. of $P$ should be $p = 0 \Rightarrow N \geq 1$, right? $\endgroup$ – Martin Brandenburg Apr 16 '14 at 17:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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