MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I would like prove that, under the conditions described below, no non-trivial variety exists.

Let $\mathcal{V}$ be a variety of algebras e.g. rings, semigroups, semilattices.

Further suppose that:

  1. The empty algebra exists i.e. $\mathcal{V}$ has no constants.
  2. The dual condition also holds. That is, let $\mathtt{1} \in \mathcal{V}$ be the one element terminal algebra. Then $id_\mathtt{1}$ is the only homomorphism whose domain is $\mathtt{1}$.

Any help much appreciated.

share|cite|improve this question
up vote 6 down vote accepted

It's false. Take the theory generated by unary operations $\zeta$ and $\iota$ and a binary operation $\cdot$, subject to the equations $$ \zeta(x) = \zeta(y), \qquad \iota(x) = \iota(y), \qquad \zeta(x) \cdot y = \zeta(x), \qquad \iota(x) \cdot y = y. $$ An example of an algebra is any ring, with $\zeta(x) = 0$, $\iota(x) = 1$, and the usual $\cdot$. So the theory is not trivial.

Obviously $\emptyset$ is an algebra. Now suppose we have a homomorphism from $1$ to an algebra $A$, picking out an element $a \in A$. The equation $\iota(x) = \zeta(x)$ holds in $1$, so $\iota(a) = \zeta(a)$ too. Hence for all $b \in A$, $$ b = \iota(a) \cdot b = \zeta(a) \cdot b = \zeta(a), $$ giving $A = \{\zeta(a)\} \cong 1$.

The trick is that although the theory has no actual constants, it has two "pseudo-constants", namely, $\zeta$ and $\iota$.

share|cite|improve this answer
Thanks very much for this example. It happens to extend a previous example I had on MO, i.e. with just the first two equations one finds that coproducts don't preserve monos. Actually I am working under further conditions e.g. that the variety is commutative, but this has been very helpful. – Rob Myers Jan 16 '14 at 18:24
You're welcome! – Tom Leinster Jan 16 '14 at 20:55

For varieties that only have unary function symbols, it is necessary to use constant functions as in Tom Leinster's answer.

$\mathbf{Proposition}$ Suppose that $V$ is a variety such that all the function symbols are unary and for all terms $t$, $V$ does not satisfy the identity $t(x)=t(y)$. Then there is some algebra $X\in V$ and multiple homomorphisms $1\rightarrow X$.

$\mathbf{Proof}$ Suppose that $V$ does not satisfy the identity $t(x)=t(y)$ for each term $t$. Then there does not exist terms $s,t$ that satisfy the identity $s(x)=t(y)$. Therefore every identity in $V$ is an identity of the form $s(x)=t(x)$. Let $G$ be the monoid of all terms in $V$ in one variable up to logical equivalence. Then $G$ is a monoid defined by $s(x)t(x)=s(t(x))$. We may therefore think of the variety $V$ as the collection of all actions from the monoid $G$ to some set $X$. On the other hand, let $G$ be the trivial action on a set $X$. In other words, $gx=x$ for all $g\in G$. Then every function $\phi:1\rightarrow G$ is a homomorphism.

share|cite|improve this answer
I am sure I don't understand this. How do we guarantee that there is a big enough set $X$ with such a trivial $G$-action? Gerhard "Thinking Of Single Unary Function" Paseman, 2014.01.13 – Gerhard Paseman Jan 13 '14 at 20:17
Any set $X$ has a trivial $G$-action simply by letting $gx=x$. – Joseph Van Name Jan 13 '14 at 21:20
OK. I guess I'm having trouble with the statement "We may therefore think of the variety V as the collection of all actions from the monoid G to some set X." I am having trouble thinking of V as such. I prefer to take a free algebra A, build X from A, and then show I will have 1 isomorphic to a subalgebra of X. Gerhard "Maybe I Need More Coffee" Paseman, 2014.01.13 – Gerhard Paseman Jan 13 '14 at 21:34

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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