Compositionality in Network Theory

Here are the slides of my talk at the workshop on compositionality at the Simons Institute for the Theory of Computing next week. I decided to talk about some new work with Blake Pollard. You can see slides here:

• John Baez, Compositionality in network theory, 6 December 2016.

and a video here:

Abstract. To describe systems composed of interacting parts, scientists and engineers draw diagrams of networks: flow charts, Petri nets, electrical circuit diagrams, signal-flow graphs, chemical reaction networks, Feynman diagrams and the like. In principle all these different diagrams fit into a common framework: the mathematics of symmetric monoidal categories. This has been known for some time. However, the details are more challenging, and ultimately more rewarding, than this basic insight. Two complementary approaches are presentations of symmetric monoidal categories using generators and relations (which are more algebraic in flavor) and decorated cospan categories (which are more geometrical). In this talk we focus on the latter.

This talk assumes considerable familiarity with category theory. For a much gentler talk on the same theme, see:

Monoidal categories of networks.

networks_compositionality

8 Responses to Compositionality in Network Theory

  1. I just spent an hour going through the slides and found them very understandable. The top-prio thing I needed to postpone is working through examples of what exactly the pushouts do when the functions i and o are not bijective.

    • John Baez says:

      Excellent—glad you liked them!

      In the category of sets, when we compose cospans:

      the set \Gamma +_Y \Gamma' is a pushout. (Pardon the weird notation \Gamma for a set; I’m just reusing a diagram I had sitting around.) This pushout turns out to work as follows: first take the disjoint union of the sets \Gamma and \Gamma', and then mod out by the smallest equivalence relation such that o(y) \sim i'(y) for all y \in Y. This is particularly simple when o and i' are injective, but it’s true in general.

      • Thanks for the clarification! Allowing non-injective i and o is the least intuitive part for me. Does this come from a practical need, or does one do this just because one can? Is there perchance a symmetric monoidal subcategory of cospans with injective i and o that would also do the job in realistic cases?

      • John Baez says:

        A good general rule in mathematics is “don’t disallow something unless you need to”. So, demanding that i and o be injective is not something you should do unless (or until) you can prove something interesting that requires this assumption.

        More importantly, non-injective i and o are required to get these cospan categories to be compact closed. I noticed this a long time ago when working on circuits made of resistors.

        Also, to get identity morphisms we need cospans where the range of i is not disjoint from the range of o. So, we definitely don’t want to demand that their ranges be disjoint!

        • Compact closure, of course, that’s a strong argument!

          I’ve reflected why I asked my injectivity question, and I’ve realized the source was my inner software engineer: I have an unusual synesthesia that makes me constantly see analogies between computer code and mathematical theories. When faced with the pushout, I instinctively viewed it as an algorithm in a computer-algebra software I might write and have to maintain. So the question surfaced: am I willing to implement and maintain the more complicated pushout? I find it interesting how the “don’t disallow” you mention turns into an “be careful before you add” when one switches from the mathematical to the algorithmic or software maintenance perspective.

          Anyway, your compact closure argument makes moot any point I might have had.

        • John Baez says:

          Carsten wrote:

          So the question surfaced: am I willing to implement and maintain the more complicated pushout? I find it interesting how the “don’t disallow” you mention turns into an “be careful before you add” when one switches from the mathematical to the algorithmic or software maintenance perspective.

          Yes, interesting! I’m glad to hear that you can explain reason for wanting to disallow noninjective legs on your cospan. If you’re maintaing software this might make sense. What really bugs me is pure mathematicians who throw unnecessary axioms into their theories just because those axioms hold in some paradigmatic example of what they’re thinking about, without carefully investigating what those axioms help us prove. Poorly thought-out axioms that disallow certain objects or morphisms in a category often make that category worse—they actually prevent us from proving interesting results.

          A classic example is demanding that some set be nonempty just because the empty set is a bit spooky. Of course the empty examples aren’t the ones we have in mind, but throwing out the empty set tends to throw out initial objects and thus prevent a category from having all colimits.

  2. John Baez says:

    By the way, I updated the talk slides so that now the penultimate slide shows lots of different symmetric monoidal categories of networks.

  3. Here at the Simons Institute, my talk on compositionality in network theory introduced ‘decorated cospans’ as a general model of open systems. These were invented by Brendan Fong, and are nicely explained in his thesis:

    • Brendan Fong, The Algebra of Open and Interconnected Systems.

    But he went further: to understand the externally observable behavior of an open system we often want to simplify a decorated cospan and get another sort of structure, which he calls a ‘decorated corelation’. His talk here explains decorated corelations and what they’re good for:

You can use Markdown or HTML in your comments. You can also use LaTeX, like this: $latex E = m c^2 $. The word 'latex' comes right after the first dollar sign, with a space after it.

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s