Title:The factorization of the Giry monad

Abstract:The two categories, measurable and convex spaces, have virtually identical categorical properties with the exception that measurable spaces does not have a coseparator. Noting that the Giry monad factorizes through the full subcategory of separated measurable spaces, which does have a coseparator, the factorization of the Giry monad through the category of convex spaces becomes more transparent, with the Boolean subobjects of a convex space generating the {\sigma}-algebra structure associated with the underlying set of a convex space. This choice is both necessary and sufficient to obtain a separated measurable space.
The duality between these two categories is evident in the tightly knit relationship existing between the symmetric monoidal closed category structure of these two categories. The construction of the tensor product in these two categories are identical in form, both using the constant graph maps to endow a set or space with the required structure such that the constant graph maps become morphisms in the respective categories, necessary to obtain the closed category structure. This aspect is used to show the unit interval, with all its affine endomorphisms, is a codense subcategory in the category of convex spaces, which in turn is used to prove that the category of Giry algebras is equivalent to the category of convex spaces.