cartesian monoidal category


Monoidal categories

Category theory



A cartesian monoidal category (usually just called a cartesian category), is a monoidal category whose monoidal structure is given by the category-theoretic product (and so whose unit is a terminal object).

A cartesian monoidal category which is also closed is called a cartesian closed category.

A strong monoidal functor between cartesian categories is called a cartesian functor.

Any category with finite products can be considered as a cartesian monoidal category (as long as we have either (1) a specified product for each pair of objects, (2) a global axiom of choice, or (3) we allow the monoidal product to be an anafunctor).

The term cartesian category usually means a category with finite products but can also mean a finitely complete category, so we avoid that term.


Cartesian monoidal categories have a number of special and important properties, such as the existence of diagonal maps Δ x:xxx\Delta_x : x \to x\otimes x and augmentations e x:xIe_x: x \to I for any object xx. In applications to computer science we can think of Δ\Delta as ‘duplicating data’ and ee as ‘deleting’ data. These maps make any object into a comonoid. In fact, any object in a cartesian monoidal category becomes a comonoid in a unique way.

Moreover, one can show (e.g. Heunen-Vicary 12, p. 84) that any symmetric monoidal category equipped with suitably well-behaved diagonals and augmentations must in fact be cartesian monoidal. More precisely: suppose CC is a symmetric monoidal category equipped with monoidal natural transformations

Δ x:xxx \Delta_x : x \to x \otimes x


e x:xI e_x : x \to I

such that the following composites are identity morphisms:

xΔ xxxe x1Ix xx x \stackrel{\Delta_x}{\longrightarrow} x \otimes x \stackrel{e_x \otimes 1}{\longrightarrow} I \otimes x \stackrel{\ell_x}{\longrightarrow} x
xΔ xxx1e xxIr xx x \stackrel{\Delta_x}{\longrightarrow} x \otimes x \stackrel{1 \otimes e_x}{\longrightarrow} x \otimes I \stackrel{r_x}{\longrightarrow} x

where rr, \ell are the right and left unitors. Then CC is cartesian.

Heuristically: a symmetric monoidal category is cartesian if we can duplicate and delete data, and ‘duplicating a piece of data and then deleting one copy is the same as not doing anything’.

A related theorem describes cartesian monoidal categories as monoidal categories satisfying two properties involving the unit object. First, we say a monoidal category CC is semicartesian if the unit for the tensor product is terminal. If this is true, any tensor product of objects xyx \otimes y comes equipped with morphisms

p x:xyx p_x : x \otimes y \to x
p y:xyy p_y : x \otimes y \to y

given by

xy1e yxIr xx x \otimes y \stackrel{1 \otimes e_y}{\longrightarrow} x \otimes I \stackrel{r_x}{\longrightarrow} x


xye x1Iy yy x \otimes y \stackrel{e_x \otimes 1}{\longrightarrow} I \otimes y \stackrel{\ell_y}{\longrightarrow} y

respectively, where ee stands for the unique morphism to the terminal object and rr, \ell are the right and left unitors. We can thus ask whether p xp_x and p yp_y make xyx \otimes y into the product of xx and yy. If so, it is a theorem that CC is a cartesian monoidal category. (This theorem is probably in Eilenberg and Kelly’s paper on closed categories, but they may not have been the first to note it.)


Discussion with an eye towards finite quantum mechanics in terms of dagger-compact categories is in

Revised on April 14, 2017 12:48:55 by Urs Schreiber (