CHP 1 Cat Idea

You might also like

Download as txt, pdf, or txt
Download as txt, pdf, or txt
You are on page 1of 5

Chapter 1: Categories - The Idea

1.1 Abstraction And Analogies


1.2 Connections And Unification
1.3 Context
1.4 Relationships
1.5 Sameness
1.6 Characterizing Things By The Role They Play
1.7 Zooming In And Out
1.8 Frameworks And Techniques

Detailed Notes

Category Theory is the mathematics of mathematics.


the problem with this == Many widespread misunderstandings of what mathematics is.

another problem == (seems like) you need to understand the whole of mathematics
before you can understand CT.
prevailing wisdom == you need to understand math at least to graduate level, or at
least to upper undergrad level with a solid base in pure mathematics (before
getting into CT)

EC learned CT the traditional way. Many undegrad subjects, then CT. she says this
didn't help her learn category theory. Instead once she learned CT, she started
understanding the parts of pure math she had not fully understood before.

change of direction for EC == start teaching CT directly, to students without a


pure math background, drawing examples from real life (vs other areas of math)

education error == we have to learn math in the order it was developed. counting
numbers -> whole numbers -> negative numbers eventually irrational numbers.

However some branches of math developed because of lack of technology and so are
redundant. It is no longer necessary to learn how to use a slide rule or compass
and ruler constructions, (_ is EC refering to Euclidean geometry here?) but this
does not hinder learning CT, just as poor horse riding skills say nothing about
ability to drive a car.

Use CT to (KEY) learn how to think (_mathematically).

In this chapter, ideas and principles of CT, before delving into details of
implementation in later chapters.

(This chapter is) An informal overview of the entire book, a little vague, but will
become clearer as the book progresses.

1.1 Abstraction And Analogies

Mathematics relies heavily on abstraction to get going (because) its arguments are
all based on rigorous logic, and rigorous logic works properly in abstract
settings. When we try to apply rigorous logic in less abstract settings, we almost
always run into problems of ambiguity - ambiguity of definitions, interpretations,
behavior and so on.
In real life situations, something almost always overpowers rigorous logic.
e.g: rigorous logic says 1 + 1= 2
but if I give you a cookie, and then another, the result maynot be two
cookies. You eat one cookie and at the end 1 + 1 = 1 !
2 flowers in succession. what if one died (or the dog ate one!)

Ambiguity enriches human life, but *in mathematics* one aim is to make arguments
unambigous. We go into detail about abstraction and its advantages and
disadvantages in the next chapter, but now the idea is that (KEY) abstraction has
the potential to be ambigous (_ and so in math, leads to others) and (KEY) CT
provides a secure framework for performing (?) abstractions.

1.2 Connections And Unification

One of the advantages of abstraction is to make connections between things/areas


that might *seem* different.
It is *superficially true* that abstraction takes as away from real life situations
but (in compensation) it enables us to make connections between superficially
different situations so we can deal with all of them at once. (e.g: counting
biscuits, ships, children, all have the same rules, and once we learn counting we
can count anything vs counting children being a different situations from counting
ships)

(step)
One source of abstraction is spotting patterns sometimes in a single situation (_
this is going from 'two apples' to "two Xes" - we spot a pattern in tiles on the
floor - sometimes across different situations - some type of people dominating an
interaction. Union negotiations vs parent child negotiations. Chapter 3 is about
patterns and how this gives us a start at recognizing abstract structures. (KEY)
Category Theory is a unifying theory that can simultaneously (a) encompass a broad
range of topics (b) also work at various scales by zooming in and out.

1.3 Context

One of the starting points on CT is the idea that we study things (KEY) in context
rather than in isolation. It is a bit like always setting a frame preference first.
(KEY) this (specifying the context) is one way to eliminate ambiguity.
e.g : one plus one is always equal to two *in a context similar to the context of
things that behave like natural numbers*.
But (KEY) there are many contexts in which things behave differently as we'll see
in chapter 4. (_ when things fall into/merge, then 1 + 1 may still be one. 1 pile
of grain + 1 pile of grain (merged) is still 1 pile of grain though a larger one)

(KEY) One of the disciplines of CT is that we are always aware of and specific
about the context we are operating in.

1.4 Relationships

(KEY) One of the crucial ways *in which* CT specifies and defines context *is via
relationships* . CT takes the view that *what is important in a given context* is
how objects (in the context) are related to each other.
e.g: age of a group of people relative to each other - this is important in some
contexts (selecting a president?) but not others ( counting them to see how many
buses are required to transport them)
(KEY) There can be *different* types of relationships between the same objects, but
we (as mathematicians, who are 'operating' on these objects in a context) we can
focus only on *specific* relationship types in a specific context. This doesn't
mean that the excluded relationships are useless, just that they are *not relevant*
in the context we examine in them.

e.g: Numbers have different relationships with each other. One is size. We can
arrange numbers sorted by size on a line. But numbers also divide each other (or
not), so we can create a different diagram to put numbers in on *this* basis. In
CT, these are two different ways of putting a categorical structure on the same set
of numbers, by using different types of relationships. We discuss this in chapter
5.

(KEY) the relationships between objects in CT can be *anything* as long as they


satisfy some basic principles that ensure that they can be organized in a mildly
tractable way. This (the basic principles defining what is a relationship relevant
to CT?) to the formal definition of CT.
To build up to this, we look at the notion of *formality* in chapter 6.

In chapter 7, we look at a particular type of relationship called equivalence


relations. Equivalence relations satisfy *many properties* making them very
tractable (?) but they also satisfy (_ from the CT pov) too many properties so they
aren't 'broadly expressive' in the way CT seeks (?)

We will see that CT is a framework that achieves a remarkable tradeoff betwen


1) good behavior (?)
2) expressive possibilities.
CT achieves a balance between these.

Part one of the book will build up to the formal definition of a category.

Then an interlude which is a tour of mathematics, presenting various mathematical


structures as categories. Usually this assumes that the student of CT are already
familiar with these branches of mathematics. Instead, here, we take the ideas of CT
and use it to introduce the mathematical ideas using this.

1.5 Sameness

One aim of CT is to (KEY) have more nuanced ways of describing sameness.


(KEY) Sameness is a key concept in mathematics, and at a basic level, this arises
as (a) the concept of equality (b) equations. (the problem is that people thing
math is *all* about numbers and equations)

For CT ists
numbers *are* things that can be organized into categories, the point is to be
able to study a much wider range of things than numbers
CT does not deal in equations, because in CT, equality is too strong a notion.
consider 5 + 1 = 1 + 5 . the two sides are not identical, but they are 'equal'
in some way. In CT, we pivot on the sense in which things are equal * to make
progress on understading the aspects on which they are different*. (note: the
*result* is equal, but the process maybe different on both sides - consider
computation necessary to count, using a child's mental process of starting from one
number and counting to the other, 1000 + 1 and 1 + 1000)

CT deals with notions of sameness more subtle than equality.


1.6 Characterizing Things By The Role They Play

CT characterises things *not by intrinsic characteristics* but *by the role they
play **in context** *. Once we understand that objects take on different
characteristics in different contexts, the idea of intrinsic characteristics get
shaky.

e.g: we might think the 5 is prime because it is divisible only by 1 and itself.
BUT this is true *in the context of natural numbers*. If we put 5 in the context
of fractions, then 5 is divisible by anything except 0.

One advantage of characterizing things by the roles they play in specific contexts
is that(KEY) we can make comparisons across different contexts by finding
(different) things that play the *same* role in *different* contexts. (_ iow we
look for 'instantiations' of an identified roles across contexts) We discuss this
when we encounter universal properties in chapter 16. This might sound
contradictory because it sounds like now we are saying that there properties that
hold across context, but what we are really saying is that some objects in a
context may be extreme or somehow canonical within a context. This can tell us
about the object having that property within a context, but it can also tell us
something about the context itself. (_ somehow, for a context C, the relationships
that hold within it, are characteristic of the C)

e.g: if we go looking at the highest and lowest paid employees in a company, it


tells us about the employees (the objects) but it also tells us something about the
company (the context). It doesn't tell us as much as a full distribution of
salaries in the company, but it tells us *something*.

1.7 Zooming In And Out

One of the powerful aspects of CT's (treatment of?) level of abstraction is that it
enables us to zoom in and out and look at large and small mathematical objects in
a similar light. This is like a scientific theory that works at the subatomic and
galactic scale, the favorite feature of CT for EC.

If we study birds we might need to build a theory of birds. But a theory of birds
is not a bird.
CT studies mathematical objects. aka CT is a theory of mathematical objects. But
CT is itself a mathematical object. so CT can be used to study itself (_ should be
true for logic also? Logic studies mathematics. But it is also a mathematical
object, so logic can be used to study itself).
This sounds self referential but what is really happening is that we are still in
category theory but in a higher dimension (?) of category theory, where 'dimension'
== 'level of relationship' (??)

In basic category theory, we start by saying that we should study relationship


between objects not just the objects themselves. But what about the relationships
then? If they are mathematical objects (and they are) shouldn't we study the
relationships between the (original) relationships? yes and this gives us one more
dimension.

Then of course why stop there? We can study the relationships between relatioships
between relationships between object? We can keep going like this. This (higher
dimensional CT) is where EC's research focus is. We see a glimpse of this in the
last part of the book.
If category theory is a theory of mathematics, then higher dimensional category
theory is a theory of categories. But this higher dimensional category theory is
still category theory. This is not just about abstraction, but also about subtlety.
HD CT trains us in balancing nuance and rigor.
(EC thinks that) mathematics is a spectacular environment to practise this kind (_
balancing nuance and rigor) kind of thinking.

1.8 Frameworks And Techniques

As described thus far, CT sounds like a philosophy. But it is rigorous technical


mathematics, only *guided* by these philosophical approaches. It sets up a
framework for implementing these (?) philosophies and pursuing these (?) goals
rigorously.

The framework consists of a formal definition of categories as algebraic


structures, and then techniques for studying these structures (_ are presented) and
constructing (?) and investigating features that might arise in them (?).

(and so) a certain amount of formal mathematics to get into the theory itself (vs
exploring the ideas behind it which HtBPi deals with).

In this book the expectation is not that readers become fluent in all aspects of
CT, or readers get to a point where they can do research in CT. The readers should
be able to (a) read and appreciate CT (b) have some buildup into the formality of
it in case you want to go further.

You might also like