Not signed in

Want to take part in these discussions? Sign in if you have an account, or apply for one below

• Forgot your password?
• Apply for membership

1. • CommentRowNumber1.
   • CommentAuthorDavidRoberts
   • CommentTimeAug 20th 2010
   • PermaLink
   Author: DavidRoberts
   Format: MarkdownItexCross-posted from MO. Feel free to put any answers there to pointy-licious goodness ---- I'm working out of _Sheaves in geometry and logic_, for reference. There is a characterisation of flat functors $A:C \to Set$ as those such that the Grothendieck construction $\int_C A$ is a filtering category. There are more general versions of this result, in which $Set$ is replaced by a more general topos. One should also be able to characterise those discrete opfibrations that arise from flat functors (up to iso/equiv?). How about if we replace $C$ by an internal category, in a topos $E$ say? Then functors out of $C$ are replaced by discrete opfibrations over $C$ in $E$. My question is this: >What sort of thing should be considered as the analogue of a flat functor in the internal setting?

   Cross-posted from MO. Feel free to put any answers there to pointy-licious goodness

   ---

   I’m working out of Sheaves in geometry and logic, for reference.

   There is a characterisation of flat functors $A:C \to Set$ as those such that the Grothendieck construction $\int_C A$ is a filtering category. There are more general versions of this result, in which $Set$ is replaced by a more general topos. One should also be able to characterise those discrete opfibrations that arise from flat functors (up to iso/equiv?). How about if we replace $C$ by an internal category, in a topos $E$ say? Then functors out of $C$ are replaced by discrete opfibrations over $C$ in $E$.

   My question is this:

   What sort of thing should be considered as the analogue of a flat functor in the internal setting?

2. • CommentRowNumber2.
   • CommentAuthorTobyBartels
   • CommentTimeAug 20th 2010
   • PermaLink
   Author: TobyBartels
   Format: MarkdownItex[MO link](http://mathoverflow.net/questions/36156/internal-version-of-a-flat-functor)

   MO link

3. • CommentRowNumber3.
   • CommentAuthorFinnLawler
   • CommentTimeAug 20th 2010
   • PermaLink
   Author: FinnLawler
   Format: MarkdownItexI've [mumbled something](http://mathoverflow.net/questions/36156/internal-version-of-a-flat-functor/36182#36182) in response over at MO.

   I’ve mumbled something in response over at MO.

4. • CommentRowNumber4.
   • CommentAuthorDavidRoberts
   • CommentTimeAug 21st 2010
   • PermaLink
   Author: DavidRoberts
   Format: MarkdownItexThanks, Finn. I've accepted your answer.

   Thanks, Finn. I’ve accepted your answer.

5. • CommentRowNumber5.
   • CommentAuthorFinnLawler
   • CommentTimeAug 21st 2010
   • PermaLink
   Author: FinnLawler
   Format: MarkdownItexCool, hope it helped.

   Cool, hope it helped.

6. • CommentRowNumber6.
   • CommentAuthorDavidRoberts
   • CommentTimeAug 21st 2010
   • PermaLink
   Author: DavidRoberts
   Format: MarkdownItexIt did, but google books doesn't give me any joy with either of Johnstone's books, nor the handbook of categorical algebra. Could you pretty please copy out the axioms?

   It did, but google books doesn’t give me any joy with either of Johnstone’s books, nor the handbook of categorical algebra.

   Could you pretty please copy out the axioms?

7. • CommentRowNumber7.
   • CommentAuthorTobyBartels
   • CommentTimeAug 21st 2010
   • PermaLink
   Author: TobyBartels
   Format: MarkdownItexFrom Elephant B2.6.1 (slightly shortened): > We begin by defining a number of objects associated with an internal category $C$. > * $P$, the object of pairs of morphisms with common codomain, is defined as the pullback $C_1 \times_{0,0} C_1$. > * $Q$, the object of pairs with common domain, is $C_1 \times_{1,1} C_1$. > * $R$, the object of parallel pairs, is the intersection of $P$ and $Q$, or equivalently the pullback $C_1 \times_{(0,1),(0,1)} C_1$. > * $S$, the object of commutative squares, is $C_2 \times_{1,1} C_2$. > * $T$, the object of diagrams of type $\bullet \overset{g}\underset{h}\rightrightarrows \bullet \overset{f}\rightarrow \bullet$ with $f g = f h$, is $C_2 \times_{(0,1),(0,1)} C_2$. These are all defined with diagrams which I hope you can write down for yourself, perhaps by my subscript notation. The pulled back maps in these pullbacks are all given by lowercase letters that match the uppercase letter that names the pullback itself. For example, $p_1, p_2\colon P \to C_1$, $d_0 p_1 = d_0 p_2$, and $P$ is universal with this structure. The structure maps $C_1 \rightrightarrows C_0$ themselves are $d_0, d_1$ as usual. From Elephant B2.6.2: > An internal category $C$ in a regular category is __filtered__ if each of the three morphisms $C_0 \to 1$, $(d_1 p_1, d_1 p_2)\colon P \to C_0 \times C_0$, and $(d_2 t_1, d_2 t_2)\colon T \to R$ is a cover. > Note: in this definition we have adopted the custom of naming morhpisms into a pullback by the names of the morphisms into the product of which the pullback is a subobject.

   From Elephant B2.6.1 (slightly shortened):

   We begin by defining a number of objects associated with an internal category $C$.

   • $P$, the object of pairs of morphisms with common codomain, is defined as the pullback $C_1 \times_{0,0} C_1$.
   • $Q$, the object of pairs with common domain, is $C_1 \times_{1,1} C_1$.
   • $R$, the object of parallel pairs, is the intersection of $P$ and $Q$, or equivalently the pullback $C_1 \times_{(0,1),(0,1)} C_1$.
   • $S$, the object of commutative squares, is $C_2 \times_{1,1} C_2$.
   • $T$, the object of diagrams of type $\bullet \overset{g}\underset{h}\rightrightarrows \bullet \overset{f}\rightarrow \bullet$ with $f g = f h$, is $C_2 \times_{(0,1),(0,1)} C_2$.

   These are all defined with diagrams which I hope you can write down for yourself, perhaps by my subscript notation. The pulled back maps in these pullbacks are all given by lowercase letters that match the uppercase letter that names the pullback itself. For example, $p_1, p_2\colon P \to C_1$, $d_0 p_1 = d_0 p_2$, and $P$ is universal with this structure. The structure maps $C_1 \rightrightarrows C_0$ themselves are $d_0, d_1$ as usual.

   From Elephant B2.6.2:

   An internal category $C$ in a regular category is filtered if each of the three morphisms $C_0 \to 1$, $(d_1 p_1, d_1 p_2)\colon P \to C_0 \times C_0$, and $(d_2 t_1, d_2 t_2)\colon T \to R$ is a cover.

   Note: in this definition we have adopted the custom of naming morhpisms into a pullback by the names of the morphisms into the product of which the pullback is a subobject.

8. • CommentRowNumber8.
   • CommentAuthorFinnLawler
   • CommentTimeAug 21st 2010
   • (edited Aug 21st 2010)
   • PermaLink
   Author: FinnLawler
   Format: MarkdownItexEr, yes. What Toby said.

   Er, yes. What Toby said.

9. • CommentRowNumber9.
   • CommentAuthorMike Shulman
   • CommentTimeAug 21st 2010
   • PermaLink
   Author: Mike Shulman
   Format: MarkdownItexFor one whose interests incline in that direction, it would be a nice exercise to work out those axioms by starting with the usual definition of a flat functor and interpreting it in the [[internal logic]] of the topos.

   For one whose interests incline in that direction, it would be a nice exercise to work out those axioms by starting with the usual definition of a flat functor and interpreting it in the internal logic of the topos.

10. • CommentRowNumber10.
    • CommentAuthorDavidRoberts
    • CommentTimeAug 21st 2010
    • PermaLink
    Author: DavidRoberts
    Format: MarkdownItexThanks. Just what I needed. Well, maybe - it is a minor distraction perhaps from the real task - internal saturated anafunctors :)

    Thanks. Just what I needed. Well, maybe - it is a minor distraction perhaps from the real task - internal saturated anafunctors :)