I followed this course by, Prof. Benno van den Berg during my 5th Semester at the ILLC Amsterdam for my M.Sc. Logic (UvA). A full description of the course can be found in the MasterMath website at link. I also collect other works I do in Category Theory.
Assignments
I use https://q.uiver.app to draw the diagrams.
Lecture Notes
Category
- A cath. is a tuple of:
- collection of objects of
- of morphisms/arrows in
- hence, for , then
- hence, is a directed-multi-graph
- s.t.
- if
- s.t. and
- for
- epi if
- mono if
- A category is uniquely defined by:
- a class
- a class of the form for
-
- by ex.1 specifying what is the identity of each object, is not necessary, since theyโre unique
- is loc.small if
- is small if and are sets.
- a Monoid is a one-obj. loc.small category
- a small category without parallel arrows is a preorder
- write if .
- by def. of cat. given above, "" is trans. and refl.
- for cat., is iso. if ex. s.t.
- is unique and called the inverse of
- all arrows from to , (homset)
- is split.epi if ex s.t.
- ( is called a section of , and a retraction of )
- for , proj., if
Functor
- A functor , is and s.t.:
- Functors preserve isomorphisms, 2.3
- A Functor preserves a property if , 2.4
- for let :
- if surj. is full
- if inj. if faithful
- if faithful, then reflects epsi and monos. 2.6
- if bij. is fully faithful
- Product Category:
- Opposite Category:
- Duality principle: of any definition, I can always find a dual (co-)
- initial / terminal object
- Terminal objects are unique up to unique isomorphism
- is a mono iff.
- mono are โoftenโ inj.func
- proved for
- in : mono iff. inj.on.obj. and faithful, ex.1
- mono are โoftenโ inj.func
- is epi iff.
- epi are โoftenโ surj.func
- proved for
- in : surj.onobj. and full full., ex.1
- epi are โoftenโ surj.func
- initial / terminal object
- Functorial Structures
- Example: a ring, the matrixes with components in ,
- is , ,
- , if iso, then iso.
- Example: a ring, the matrixes with components in ,
Natural Transformation
- for ex. cat. s.t. , are natural transformations
- for a nat.tr. is a fam. of morph. s.t.
- for , (Naturality Condition)
- : and
- for , (Naturality Condition)
- if in is iso. iff
- , functor composition is the object part of a functor (see diag), 3.3
- for a nat.tr. is a fam. of morph. s.t.
Equivalence of Categories
- if there are functors , s.t.
- it is called equivalence and is the pseudo-inverse
- also is natural isomorphism of categories
- Equivalence preserves and reflects terminal objects (and most other properties)
- is an equivalence iff:, 3.6
- is essentially surjective andโฆ
- that is:
- โฆ is fully faithful
- is full and faithful
- is essentially surjective andโฆ
- sm.cat., , ex.2.2
Limits
- for , a cone is s.t. , nat.tr. (cone), s.t.
- with nat.req., 4.1
- , if and
- for , s.t.
- hence
- , is a map s.t. (map of cones), 4.1
- a lim.cone for is term.obj of (limiting cone). 4.2
- a lim.con for is term.obj. in
- Product Category 4.1.2
- let , a product is the vertex of the limiting cone, called .
- only has elements and only reflexive arrows.
- Equalizer 4.1.3
- for let be a cone, if it is limiting is an equalizer.
- .
- is
- for let be a cone, if it is limiting is an equalizer.
- Pullbacks: 4.1.4
- for , a let be the lim.cone, its diagram is the pullback diagram
- has as objects and arrows and .
- has lim.of.sh. if , 4.3
- if is s.t. it is said complete.
- if only on all s.t. it is fin.compl.
- fin.compl.cat are also called lex, left exact or cartesian.
- has โฆ
- prod. if complete on all
- equal. if complete on all
- pullb. if complete on all
- for , and lim.cone, say it preserved by if lim.con, 4.4
- pres.lim.of.shape. if it preserves every lim.cone of
- it preserves small/finite limits if it preserves limits of shape for each small/finite.
- if only on all s.t. it is fin.compl.
- complete categories:
- sets, top, grps, ring, ab.gr, vs. over a fixed field, cat. (fields are not!)
- Colimits 4.2
- a colimiting cocone for is a limiting cone for .
- s.t. init.obj. in .
- a colimiting cocone for , denoted or
- is cocomplete if all small have colimits of for diagrams .
- coprod. of proj. is proj., ex.3.1.b
Complete Categories
- if has small products and equalizers, then is complete, 5.1
- also for for a cardinal
- also co- version
- if has term.obj and pullb. it has binary prod. and equal, 5.3
- tfae: 5.4
- fin.compl.
- has fin.prod. and equil.
- has pullb. and a term.obj.
- has lim.of.sh., then so does any , 5.5
- for an equiv. and a cat., preserves and reflects limits of , 5.6
- if complete has lim.of.sh., does
- if , then is a preorder
Cartesian Closed Categories
- in , for mono:
- are iso. if
- for , there is a coproduct diagram with and
- given the AC, every epi is split.
- Exponentials: for and , and are distinguished
- , s.t. , 6.1
- equivalently is a bij.
- is ccc if it has fin.prod. and every exists.
- in ccc holds with , ,
- (note, nat.tr. are precisely such )
- for with fin.prod., is called exponentiable if all exponentials exists
- 6.6?
- , s.t. , 6.1
- Natural Numbers Object
- in with term.obj. , a nat.num.obj. is s.t.:
- ,
- in with ccc and nat.num.obj. call a num.th.fun. corresponding to if
- PA can be done within this frame
- in with term.obj. , a nat.num.obj. is s.t.:
Presheaves
-
for sm.cat., the presh. is , 7.1
- is complete and cocomplete, limits calculated pointwise
- for with pullb., is mono iff , 7.3
- for with pusho., is epi iff
-
Yoneda Lemma: 7.2
- for loc.sm.cat, , hence
- also, for , s.t. .
- since is an exp.cat., corresponds to
- since , first take , then .
- so, for , we have
- and, for , get , comp.w.
- and, for , get s.t. , com.w.
- , s.t. (Yoneda)
- for , , diag.com. Yoneda_Naturality.jpeg (Naturality)
- , 7.4
- hence, os a nat.iso. s.t.
- is repr. if , 7.5
- for , is a comm.cat. if
- , recall
- is colimiting cocone in , 7.6
- every presheaf is a colimit of representables
- , so
- is inj. and fully faithful, 7.7
- is ccc
- as an example of limit, take the exponentials
- for ,
- for , ?
- as an example of limit, take the exponentials
- , ex.3.1.c
- a presh is proj. iff it is a retract of a copr. of repres.
Presheaves as a Topos
- a topos is a ccc.w.fin.lim with a subobject classifier
- for cat.fin.lim. a subo.cl. is a s.t.:
- is mono
-
- is classifies , or the subobject represented by
- since is mono, there is a unique such too.
- hence, is terminal, 8.3
- call subo.cl. or also just
- for , def. and order
-
- is then unique and mono, 8.4
- for iso, we get eq.classes which we call subo., write
- if is a set, call well-powered, this we assume for .
- in , if inj.,
- for ,
- and monotone
- if small,
- having a subo.cl. is eq. to saying that is repr.
- hence, lec.10.1.5:36
- has subo.cl. if is representable lec.10.1.6:19
- ex. s.t. nat.iso.
-
- for , is pow.obj. of if s.t.
- and corr. to the subo. of , then corr. to the subo. of .
- correspondence is given by the subo.cl.,
- note , hence "" only requires a bij.
- pow.obj. are unique up to iso., usually denoted with
- and corr. to the subo. of , then corr. to the subo. of .
- Subobject Classifiers in Presheaves
- if , a subpresh of is s.t.:
- let be the set of all supresh. on of
- let s.t. , is defined by pulling back along
- if is a sieve on , then
- for ,
- hence ,
-
- is the unique one containing
- for a topos, we have natuarally in .
Adjunctions
- , an adj. is a nat.bij or s.t.:
-
- hence maps , are called transposes of each others
- write , note .
- (Naturality)
- this is eq. to , s.t. and .
-
- for adj. as above, fix , by nat. is repr. by .
- by Yon. , s.t.
- dually: for , s.t.
- and , 9.2
- call the unit and the
- if is inv. of iff and
- and
- for , and s.t.
- then: with unit and counit, 9.3
- eq.cat. with ps.inv, for , , then with unit and one with counit (not nec. at the same time), 9.4
- if then pres.colim., pres.lim.
- pres. init.obj.
- if has right.adj., let , then has right.adj. too, 9.7
Monads and Algebras
- for and , let ,
- note nat.tr.
- also , for
- see diag. 10.1
- for cat, a monad is s.t.:
- nat.tr.
- and s.t. 10.1 holds
- note that it is in fact a monoid ?
- for on a cat. , the cat. -Alg is:
- there is an adj. between and and it is ge. by ., 10.4
- for and , let s.t.
- nat. of proves well-definability.
- call monadic (or that over is) if is an equiv.
Presheaves Revisited
- for sm.cat, let recall , call it category of elements.
- , we can then think of with
- for forg.fun., def.
- is a lim.cocone on , 11.2
- every presheaf is a colimit of representables
- Kan Extensions
- for sm.cat., is s.t.
- , 11.3
- call the left Kan extension
- for for sm.cat., the prec.fun. , , has both adj., 11.5
- for sm.cat., is s.t.