Let consider a monoidal category $\mathscr C, \otimes$ with unit $I$ and unitors $\lambda_X:I\otimes X\to X$ and $\varrho_X:X\otimes I\to X$.

Each endomorphism of the unit $a:I\to I$ induces, on each object $X$, the following endomorphisms: $$L(a,X)=\lambda_X^{-1}(a\otimes 1_X)\lambda_X:X\to X$$ $$R(a,X)=\varrho_X^{-1}(1_X\otimes a)\varrho_X:X\to X$$

Then both $L(a,X)$ and $R(a,X)$ are natural in $X$, and $L(a,I)=a=R(a,I):I\to I$. Note that such equality holds for all objects if $I$ is a generator for the category, that's $L(a,X)=R(a,X)$. Indeed for each $x:I\to X$, by naturality, we have: $L(a,X)\circ x=x\circ a=R(a,X)\circ x$.

Question: the equality $L(a,X)=R(a,X)$ holds for all objects $X$ (even without assuming $I$ to be a generator for the category)?


By no means. Think of the category of bimodules over a commutative ring $k$, so $k$-vector spaces with two actions which are associative with respect to each other but which may not agree. For instance $\mathbb{R}[\epsilon]$ has a natural extra action on all its modules by quotients go down to $\mathbb{R}$ and using the vector space action. Then $L(a,X)$ corresponds to multiplying on the left, and $R(a,X)$ on the right, with an element of $k$, which in the example, acting on itself, is the difference for instance between $x+y\epsilon\mapsto 0,\mapsto x\epsilon$.


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.