5
$\begingroup$

Let $(M,+,e)$ be a commutative monoid with unit $e$. An element $a\in M$ is called cancellative element if

for any $b,c \in M$ such that $a+b=a+c$ implies that $b=c$.

Let $(\mathbf{N},+,0)$ the commutative monoid of natural numbers.

suppose that

  1. we have two morphisms of monoids $f:(\mathbf{N},+,0)\rightarrow (M,+,e)$ and $g:(M,+,e)\rightarrow (\mathbf{N},+,0) $ such that $g\circ f= id$.
  2. The monoid $(M,+,e)$ is torsion-free.

My question is the following: is the element $a=f(1)$ automatically a cancellative element in $(M,+,e)$ ?

Edit: By torsion-free I do mean that there does not exist a natural number $n>0$ and some element $x\in M-\{e\}$ such that $n x=e$.

$\endgroup$
2
  • $\begingroup$ By torsion free you mean that there does not exist a natural number $n$ and some element $x\in M$ such that $n\cdot x=e$? $\endgroup$ Jan 2, 2021 at 1:14
  • 2
    $\begingroup$ The usual definition of torsion free for commutative monoids is na=nb implies a=b. $\endgroup$ Jan 2, 2021 at 1:30

2 Answers 2

7
$\begingroup$

The answer is no. Let $U=\{0,1\}$ under multiplication. Let $P$ be the semigroup of positive integers under $+$. Consider $S=P\times U$, the direct product and let $M=S\cup \{I\}$ where $I$ is an adjoined identity. Then $M$ is torsion-free, there is a homomorphism $f\colon \mathbf N\to M$ given by $f(0)=I$ and $f(n)=(n,0)$ for $n>0$ and $g\colon M\to \mathbf N$ with $g(I)=0$ and $g(n,x)=n$ for $n>0$ and $x\in \{0,1\}$ and clearly $gf=1$ but $(1,0)$ is not cancellable.

$\endgroup$
5
$\begingroup$

Consider the monoid $M=\mathbb{N}\times \{0,1\}$ where $$(n,a)*(m,b):=(n+m, a\cdot b).$$ The unit element is $e:=(0,1)$. Note that this monoid is torsion free. Now consider the maps $$g:(M,*,e)\rightarrow (\mathbb{N}, +,0), g(n,a)=n$$ and $f: (\mathbb{N}, +,0) \rightarrow (M,*,e)$ such that $f(0)=e$ and $f(n)=(n,0)$. Then we have $g\circ f=id$, but $f(1)=(1,0)$ is not cancellative as $$ (1,0)*(0,0)=(1,0)=(1,0)*(0,1).$$

$\endgroup$
6
  • 1
    $\begingroup$ You answered with essentially the same construction as me while I was typing it. I just removed $(0,0)$ to make it more torsion-free in that there are no elements generating a finite subsemigroup except the identity. $\endgroup$ Jan 2, 2021 at 1:40
  • $\begingroup$ @BenjaminSteinberg: I essentially know nothing about semigroups, thus I also didn't know about idempotents not being considered torsion-free. Surely, you should keep your version :) $\endgroup$ Jan 2, 2021 at 1:47
  • 1
    $\begingroup$ @BenjaminSteinberg Thanks, I keep forgetting about idempotens. Do you know a good introductory textbook (or survey paper) on semigroup theory? $\endgroup$ Jan 2, 2021 at 13:24
  • 1
    $\begingroup$ It's a pretty big subject and commutative is vastly different than noncommutative. John Howie has a fairly basic book. Well two of them but I think the newer one is better. Peter Higgins has a nice book too. $\endgroup$ Jan 2, 2021 at 17:59
  • $\begingroup$ @BenjaminSteinberg Thanks for the reference! $\endgroup$ Jan 3, 2021 at 13:31

Your Answer

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

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