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 seem to remember a proof that if a category $C$ has coproducts and a $0$ object, then necessarily if we had objects of $C$, say $a$ and $-a$, such that $a \oplus -a \simeq 0$, then $a\simeq 0\simeq -a$.

But right now, I can't place this, nor am I 100% sure that that is the correct property. I am able to show that, using the various universal properties at play, that the morphisms in such a category are necessarily quite boring, but not that it completely collapses (i.e. if I assume that all objects have a corresponding 'negative' object).

share|cite|improve this question
up vote 17 down vote accepted

With Yoneda ? For every object $X$, $Mor(a\oplus (-a),X)$ is a singleton since $a \oplus (-a)$ is initial. And $Mor(a\oplus (-a),X) \cong Mor(a,X) \times Mor((-a),X)$ is a singleton as well. So $Mor(a,X) \cong Mor((-a),X) \cong Mor(0,X)$. Hence $a \cong (-a) \cong 0$.

share|cite|improve this answer
So all I need is that $a \oplus -a$ is initial for this to work? – Jacques Carette Jun 20 '12 at 17:57
Yes, the only hypothesis is $a \oplus (-a)$ initial. – Philippe Gaucher Jun 20 '12 at 18:05

There are three proofs:

1) If $C$ has infinite coproducts, you may use the Eilenberg swindle:

$$a \cong a \oplus (-a \oplus a) \oplus (-a \oplus a) \oplus \dotsc \cong (a \oplus -a) \oplus (a \oplus -a) \oplus \dotsc \cong 0.$$

2) Here is my personal summary of Philippe's proof: (a) It is enough to prove the dual statement $a \times a^{-1} \cong 1 \Rightarrow a \cong 1$ in categories with products. (b) This may be reduced, by the Yoneda embedding which preserves products, to the case $\mathsf{Set}$. (c) In the case of $\mathsf{Set}$ it is clear.

3) If $\mathcal{L}$ is an invertible object of a symmetric monoidal category, then $\mathcal{L} \otimes -$ is cocontinuous since $\mathcal{L}^{-1} \otimes - $ is right adjoint to it. In particular $\mathcal{L} \otimes -$ preserves initial objects. Now apply this to $(C,\oplus,0)$. Thus if $a$ is invertible w.r.t. $\oplus$, we have $a \oplus 0 \cong 0$. On the other hand, the left hand side is also $a$.

share|cite|improve this answer
Thank you very much - these are quite enlightening. – Jacques Carette Jun 21 '12 at 11:55

for any object $X\in C$ we have the bijections $C(a, X)\times C(-a, X) \cong C(a\oplus -a, X)\cong C(0, X)=${$0$} then the sets $C(a, X), C(-a, X)$ have one element (the 0-morphism $a\to X=a\to 0\to X$ and $-a\to X=-a\to 0\to X$). THen $a$ is a initial object as $0$ then $a\cong 0$, similary $-a\cong 0$

share|cite|improve this answer
I posted too late the answere.. – Buschi Sergio Jun 20 '12 at 18:09
By a few minutes, yes. – Jacques Carette Jun 20 '12 at 18:11

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.