230
$\begingroup$

The title really is the question, but allow me to explain.

I am a pure mathematician working outside of probability theory, but the concepts and techniques of probability theory (in the sense of Kolmogorov, i.e., probability measures) are appealing and potentially useful to me. It seems to me that, perhaps more than most other areas of mathematics, there are many, many nice introductory (as well as not so introductory) texts on this subject.

However, I haven't found any that are written from what it is arguably the dominant school of thought of contemporary mainstream mathematics, i.e., from a structuralist (think Bourbaki) sensibility. E.g., when I started writing notes on the texts I was reading, I soon found that I was asking questions and setting things up in a somewhat different way. Here are some basic questions I couldn't stop from asking myself:

[0) Define a Borel space to be a set $X$ equipped with a $\sigma$-algebra of subsets of $X$. This is already not universally done (explicitly) in standard texts, but from a structuralist approach one should gain some understanding of such spaces before one considers the richer structure of a probability space.]

  1. What is the category of Borel spaces, i.e., what are the morphisms? Does it have products, coproducts, initial/final objects, etc? As a significant example here I found the notion of the product Borel space -- which is exactly what you think if you know about the product topology -- but seemed underemphasized in the standard treatments.

  2. What is the category of probability spaces, or is this not a fruitful concept (and why?)? For instance, a subspace of a probability space is, apparently, not a probability space: is that a problem? Is the right notion of morphism of probability spaces a measure-preserving function?

  3. What are the functorial properties of probability measures? E.g., what are basic results on pushing them forward, pulling them back, passing to products and quotients, etc. Here again I will mention that product of an arbitrary family of probability spaces -- which is a very useful-looking concept! -- seems not to be treated in most texts. Not that it's hard to do: see e.g.

http://alpha.math.uga.edu/~pete/saeki.pdf

I am not a category theorist, and my taste for how much categorical language to use is probably towards the middle of the spectrum: that is, I like to use a very small categorical vocabulary (morphisms, functors, products, coproducts, etc.) as often as seems relevant (which is very often!). It would be a somewhat different question to develop a truly categorical take on probability theory. There is definitely some nice mathematics here, e.g. I recall an arxiv article (unfortunately I cannot put my hands on it at this moment) which discussed independence of events in terms of tensor categories in a very persuasive way. So answers which are more explicitly categorical are also welcome, although I wish to be clear that I'm not asking for a categorification of probability theory per se (at least, not so far as I am aware!).

$\endgroup$
13
  • 6
    $\begingroup$ This is not developed enough to be a (partial) answer rather than a comment, but see perhaps: golem.ph.utexas.edu/category/2007/02/… (and other google/Mathscinet results for "Giry monad") $\endgroup$
    – Yemon Choi
    Apr 8, 2010 at 16:16
  • 12
    $\begingroup$ One thing I thought I'd mention - as a probablist manqué - is a comment at the beginning of Williams' Probability with Martingales, where he says something along lines of "it would be nice if we could think of random variables as equivalence classes of functions rather than functions, so that we don't need to keep inserting 'a.e.' everywhere; but this point of view runs into trouble when dealing with continuous-time stochastic processes". Which implies he is not keen on 'structuralist POV', although it doesn't rule out the possibility. $\endgroup$
    – Yemon Choi
    Apr 8, 2010 at 16:20
  • 26
    $\begingroup$ Something that you may want to consider is the fact that probability spaces are not the essential objects in probability, for at least two reasons. First, it is very common to change the underlying probability space, as long as the distributions of the relevant random variables remain the same. This allows to consider new events along the way. As suggested in Neel's answer, this may have a categorical formulation. But worst than that is the fact that often (every time martingales appear, at least) you want to leave the space unchanged and vary the sigma-algebra. $\endgroup$ Apr 9, 2010 at 16:29
  • 18
    $\begingroup$ Indeed, one of the major differences between measure theory and probability theory (besides the perspective being completely different) is that in measure theory one fixes one sigma algebra, and in probability one considers relationships between multiple sigma algebras. $\endgroup$ Apr 9, 2010 at 16:54
  • 3
    $\begingroup$ A beautiful, structural account of Borel spaces and measurable mappings is given in: ma.utexas.edu/mp_arc/c/02/02-156.pdf The category of Borel spaces and measurable mappings is sometimes denoted by meas (for measurable spaces, the nowadays more common name for Borel spaces), but the only real application I know of is in epistemic game theory, see for example chapter 7 here: scholarworks.iu.edu/dspace/bitstream/handle/2022/7065/… $\endgroup$ Jan 1, 2012 at 0:55

13 Answers 13

232
$\begingroup$

$\def\Spec{\mathop{\rm Spec}} \def\R{{\bf R}} \def\Ep{{\rm E}^+} \def\L{{\rm L}} \def\EpL{\Ep\L}$ One can argue that an object of the right category of spaces in measure theory is not a set equipped with a σ-algebra of measurable sets, but rather a set $S$ equipped with a σ-algebra $M$ of measurable sets and a σ-ideal $N$ of negligible sets, i.e., sets of measure 0. The reason for this is that you can hardly state any theorem of measure theory or probability theory without referring to sets of measure 0. However, objects of this category contain less data than the usual measured spaces, because they are not equipped with a measure. Therefore I prefer to call them enhanced measurable spaces, since they are measurable spaces enhanced with a σ-ideal of negligible sets. A morphism of enhanced measurable spaces $(S,M,N)→(T,P,Q)$ is a map $S\to T$ such that the preimage of every element of $P$ is a union of an element of $M$ and a subset of an element of $N$ and the preimage of every element of $Q$ is a subset of an element of $N$.

Irving Segal proved in “Equivalences of measure spaces” (see also Kelley's “Decomposition and representation theorems in measure theory”) that for an enhanced measurable space $(S,M,N)$ that admits a faithful measure (meaning $μ(A)=0$ if and only if $A∈N$) the following properties are equivalent.

  1. The Boolean algebra $M/N$ of equivalence classes of measurable sets is complete;
  2. The space of equivalence classes of all bounded (or unbounded) real-valued functions on $S$ modulo equality almost everywhere is Dedekind-complete;
  3. The Radon-Nikodym theorem is true for $(S,M,N)$;
  4. The Riesz representation theorem is true for $(S,M,N)$ (the dual of $\L^1$ is isomorphic to $\L^∞$);
  5. Equivalence classes of bounded functions on $S$ form a von Neumann algebra (alias W*-algebra).

An enhanced measurable space that satisfies these conditions (including the existence of a faithful measure) is called localizable. This theorem tells us that if we want to prove anything nontrivial about measurable spaces, we better restrict ourselves to localizable enhanced measurable spaces. We also have a nice illustration of the claim I made in the first paragraph: none of these statements would be true without identifying objects that differ on a set of measure 0. For example, take a nonmeasurable set $G$ and a family of singleton subsets of $G$ indexed by themselves. This family of measurable sets does not have a supremum in the Boolean algebra of measurable sets, thus disproving a naive version of (1).

But restricting to localizable enhanced measurable spaces does not eliminate all the pathologies: one must further restrict to the so-called compact and strictly localizable enhanced measurable spaces, and use a coarser equivalence relation on measurable maps: $f$ and $g$ are weakly equal almost everywhere if for any measurable subset $B$ of the codomain the symmetric difference $f^*B⊕g^*B$ of preimages of $B$ under $f$ and $g$ is a negligible subset of the domain. (For codomains like real numbers this equivalence relation coincides with equality almost everywhere.)

An enhanced measurable space is strictly localizable if it splits as a coproduct (disjoint union) of σ-finite (meaning there is a faithful finite measure) enhanced measurable spaces. An enhanced measurable space $(X,M,N)$ is (Marczewski) compact if there is a compact class $K⊂M$ such that for any $m∈M∖N$ there is $k∈K∖N$ such that $k⊂m$. Here a compact class is a collection $K⊂2^X$ of subsets of $X$ such that for any $K'⊂K$ the following finite intersection property holds: if for any finite $K''⊂K'$ we have $⋂K''≠∅$, then also $⋂K'≠∅$.

The best argument for such restrictions is the following Gelfand-type duality theorem for commutative von Neumann algebras.

Theorem. The following 5 categories are equivalent.

  1. The category of compact strictly localizable enhanced measurable spaces with measurable maps modulo weak equality almost everywhere.
  2. The category of hyperstonean topological spaces and open continuous maps.
  3. The category of hyperstonean locales and open maps.
  4. The category of measurable locales (and arbitrary maps of locales).
  5. The opposite category of commutative von Neumann algebras and normal (alias ultraweakly continuous) unital *-homomorphisms.

I actually prefer to work with the opposite category of the category of commutative von Neumann algebras, or with the category of measurable locales. The reason for this is that the point-set definition of a measurable space exhibits immediate connections only (perhaps) to descriptive set theory, and with additional effort to Boolean algebras, whereas the description in terms of operator algebras or locales immediately connects measure theory to other areas of the central core of mathematics (noncommutative geometry, algebraic geometry, complex geometry, differential geometry, topos theory, etc.).

Additionally, note how the fourth category (measurable locales) is a full subcategory of the category of locales. Roughly, the latter can be seen as a slight enlargement of the usual category of topological spaces, for which all the usual theorems of general topology continue to hold (e.g., Tychonoff, Urysohn, Tietze, various results about paracompact and uniform spaces, etc.). In particular, there is a fully faithful functor from sober topological spaces (which includes all Hausdorff spaces) to locales. This functor is not surjective, i.e., there are nonspatial locales that do not come from topological spaces. As it turns out, all measurable locales (excluding discrete ones) are nonspatial. Thus, measure theory is part of (pointfree) general topology, in the strictest sense possible.

The non-point-set languages (2–5) are also easier to use in practice. Let me illustrate this statement with just one example: when we try to define measurable bundles of Hilbert spaces on a compact strictly localizable enhanced measurable space in a point-set way, we run into all sorts of problems if the fibers can be nonseparable, and I do not know how to fix this problem in the point-set framework. On the other hand, in the algebraic framework we can simply say that a bundle of Hilbert spaces is a Hilbert W*-module over the corresponding von Neumann algebra.

Categorical properties of von Neumann algebras (hence of compact strictly localizable enhanced measurable spaces) were investigated by Guichardet in “Sur la catégorie des algèbres de von Neumann”. Let me mention some of his results, translated in the language of enhanced measurable spaces. The category of compact strictly localizable enhanced measurable spaces admits equalizers and coequalizers, arbitrary coproducts, hence also arbitrary colimits. It also admits products (and hence arbitrary limits), although they are quite different from what one might think. For example, the product of two real lines is not $\R^2$ with the two obvious projections. The product contains $\R^2$, but it also has a lot of other stuff, for example, the diagonal of $\R^2$, which is needed to satisfy the universal property for the two identity maps on $\R$. The more intuitive product of measurable spaces ($\R\times\R=\R^2$) corresponds to the spatial tensor product of von Neumann algebras and forms a part of a symmetric monoidal structure on the category of measurable spaces. See Guichardet's paper for other categorical properties (monoidal structures on measurable spaces, flatness, existence of filtered limits, etc.).

Another property worthy of mentioning is that the category of commutative von Neumann algebras is a locally presentable category, which immediately allows one to use the adjoint functor theorem to construct commutative von Neumann algebras (hence enhanced measurable spaces) via their representable functors.

Finally let me mention pushforward and pullback properties of measures on enhanced measurable spaces. I will talk about more general case of $\L^p$-spaces instead of just measures (i.e., $\L^1$-spaces). For the sake of convenience, denote $\L_p(M)=\L^{1/p}(M)$, where $M$ is an enhanced measurable space. Here $p$ can be an arbitrary complex number with a nonnegative real part. We do not need a measure on $M$ to define $\L_p(M)$. For instance, $\L_0$ is the space of all bounded functions (i.e., the commutative von Neumann algebra corresponding to $M$), $\L_1$ is the space of finite complex-valued measures (the dual of $\L_0$ in the ultraweak topology), and $\L_{1/2}$ is the Hilbert space of half-densities. I will also talk about extended positive part $\EpL_p$ of $\L_p$ for real $p$. In particular, $\EpL_1$ is the space of all (not necessarily finite) positive measures on $M$.

Pushforward for $\L_p$-spaces. Suppose we have a morphism of enhanced measurable spaces $M\to N$. If $p=1$, then we have a canonical map $\L_1(M)\to\L_1(N)$, which just the dual of $\L_0(N)→\L_0(M)$ in the ultraweak topology. Geometrically, this is the fiberwise integration map. If $p≠1$, then we only have a pushforward map of the extended positive parts, namely, $\EpL_p(M)→\EpL_p(N)$, which is nonadditive unless $p=1$. Geometrically, this is the fiberwise $\L_p$-norm. Thus $\L_1$ is a functor from the category of enhanced measurable spaces to the category of Banach spaces and $\EpL_p$ is a functor to the category of “positive homogeneous $p$-cones”. The pushforward map preserves the trace on $\L_1$ and hence sends a probability measure to a probability measure.

To define pullbacks of $\L_p$-spaces (in particular, $\L_1$-spaces) one needs to pass to a different category of enhanced measurable spaces. In the algebraic language, if we have two commutative von Neumann algebras $A$ and $B$, then a morphism from $A$ to $B$ is a usual morphism of commutative von Neumann algebras $f\colon A\to B$ together with an operator valued weight $T\colon\Ep(B)\to\Ep(A)$ associated to $f$. Here $\Ep(A)$ denotes the extended positive part of $A$. (Think of positive functions on $\Spec A$ that can take infinite values.) Geometrically, this is a morphism $\Spec f\colon\Spec B\to\Spec A$ between the corresponding enhanced measurable spaces and a choice of measure on each fiber of $\Spec f$. Now we have a canonical additive map $\EpL_p(\Spec A)\to\EpL_p(\Spec B)$, which makes $\EpL_p$ into a contravariant functor from the category of enhanced measurable spaces and measurable maps equipped with a fiberwise measure to the category of “positive homogeneous additive cones”.

If we want to have a pullback of $\L_p$-spaces themselves and not just their extended positive parts, we need to replace operator valued weights in the above definition by finite complex-valued operator valued weights $T\colon B\to A$ (think of a fiberwise finite complex-valued measure). Then $\L_p$ becomes a functor from the category of enhanced measurable spaces to the category of Banach spaces (if the real part of $p$ is at most $1$) or quasi-Banach spaces (if the real part of $p$ is greater than $1$). Here $p$ is an arbitrary complex number with a nonnegative real part. Notice that for $p=0$ we get the original map $f\colon A\to B$ and in this (and only this) case we do not need $T$.

Finally, if we restrict ourselves to an even smaller subcategory defined by the additional condition $T(1)=1$ (i.e., $T$ is a conditional expectation; think of a fiberwise probability measure), then the pullback map preserves the trace on $\L_1$ and in this case the pullback of a probability measure is a probability measure.

There is also a smooth analog of the theory described above. The category of enhanced measurable spaces and their morphisms is replaced by the category of smooth manifolds and submersions, $\L_p$-spaces are replaced by bundles of $p$-densities, operator valued weights are replaced by sections of the bundle of relative 1-densities, the integration map on 1-densities is defined via Poincaré duality (to avoid any dependence on measure theory) etc. There is a forgetful functor that sends a smooth manifold to its underlying enhanced measurable space.

Of course, the story does not end here, there are many other interesting topics to consider: products of measurable spaces, the difference between Borel and Lebesgue measurability, conditional expectations, etc. An index of my writings on this topic is available.

$\endgroup$
32
  • 1
    $\begingroup$ This is amazing! Where can I read about these ideas and further developments? For instance i'd really love to see information theory developed with this language. $\endgroup$ Oct 31, 2017 at 13:08
  • 2
    $\begingroup$ @SaalHardali: I am not aware of any exposition of information theory in this language. I have been meaning to write a survey of measure theory along the lines outlined in my answer for a while, but my research leaves me very little time for side projects. That being said, some additional material can be found in my paper arxiv.org/abs/1309.7856. $\endgroup$ Oct 31, 2017 at 17:13
  • 2
    $\begingroup$ @LSpice: It means that the categorical product is a disjoint union of R^2 (with its Lebesgue measurable/negligible sets), R^1, and a lot of other stuff. The diagonal is not a point, it is isomorphic to R and splits off as a disjoint summand. $\endgroup$ Jan 19, 2018 at 23:52
  • 1
    $\begingroup$ @LSpice: Yes, although one of the diagonals is a negligible set, hence invisibile in the category. $\endgroup$ Jan 20, 2018 at 22:08
  • 1
    $\begingroup$ @SergeiAkbarov: I added references to the papers by Segal and Kelley, a remark about the Riesz representation theorem, and a source for the measurable version of Gelfand duality. $\endgroup$ Jun 5, 2018 at 15:44
127
$\begingroup$

In the spirit of this answer to a different question, I'll offer a contrarian answer. How to understand probability theory from a structuralist perspective:

Don't.

To put it less provocatively, what I really mean is that probabilists don't think about probability theory that way, which is why they don't write their introductory books that way. The reason probabilists don't think that way is that probability theory is not about probability spaces. Probability theory is about families of random variables. Probability spaces are the mathematical formalism used to talk about random variables, but most probabilists keep the probability spaces in the background as much as possible. Doing probability theory while dwelling on probability spaces is a little like doing number theory while dwelling on a definition of 1 as $\{\{\}\}$ etc. (That last sentence is definitely an overstatement, but I can't think of a more apt analogy offhand.)

That said, multiple perspectives are always good to have, so I'm very happy you asked this question and that you've gotten some very nice noncontrarian answers that I hope to digest better myself.

Added: Here is something which is perhaps more similar to dwelling on probability spaces. To set the stage for graph theory carefully one may start by defining a graph as a pair $(V,E)$ in which $V$ is a (finite, nonempty) set and $E$ is a set of cardinality 2 subsets of $V$. You need to start tweaking this in various ways to allow loops, directed graphs, multigraphs, infinite graphs, etc. But worrying about the details of how you do this is a distraction from actually doing graph theory.

Added much later: For a completely different perspective, based on new developments since this question was first asked, see my more recent answer.

$\endgroup$
19
  • 62
    $\begingroup$ Indeed, I saw a quote from somebody famous (if I think of the author I'll edit) to the effect that "one could say that probability theory is the study of measure spaces with measure one, but this is like saying that number theory is the study of finite strings of the digits {0,...,9}." $\endgroup$ May 25, 2010 at 13:36
  • 10
    $\begingroup$ Another great quote along the same lines, from Rudin (Real and Complex Analysis, page 18 in my edition): "For instance, the real line may be described as a quadruple $(R^1, +, \cdot, <)$ where $+$, $\cdot$, and $<$ satisfy the axioms of a complete archimedean ordered field. But it is a safe bet that very few mathematicians think of the real field as an ordered quadruple." $\endgroup$ May 30, 2010 at 14:47
  • 11
    $\begingroup$ That is a great quote, but it doesn't make all the points Nate's quote does. If you think of the reals as a quadruple, you have the formalism necessary to understand and prove theorems about real numbers, although you may lack the intuition needed to appreciate the theorems. But if you think of natural numbers as strings of digits, you're missing not only intuition but also interesting algebraic structure. Likewise, a measure space with measure one is insufficient structure for probability; you need some additional algebraic or geometric structure before you can even talk about expectations. $\endgroup$ May 30, 2010 at 23:35
  • 54
    $\begingroup$ Just for the record, that quote is my own, though the general sentiment that probability is not about measure spaces is certainly very widely held among probabilists. $\endgroup$
    – Terry Tao
    Sep 3, 2010 at 19:08
  • 18
    $\begingroup$ @MartinBrandenburg: Moreover, I would need convincing that probability theory is about the category of probability spaces and measurable maps. That category surely plays a role, but most of what's of interest in the theory seems to take place outside of it. In particular, if you can show me how to view the law of large numbers or the central limit theorem as being about the category of probability spaces and measurable maps, then I'll reconsider. $\endgroup$ Mar 26, 2015 at 10:42
57
$\begingroup$

A few months ago, Terry Tao had a really insightful post about "the probabilistic way of thinking", in which he suggested that a nice category of probability spaces was one in which the objects were probability spaces and the morphisms were extensions (ie, measurable surjections which are probability preserving). By avoiding looking at the details of the sample space, you can elegantly capture the style of probabilistic arguments in which you introduce new sources of randomness as needed.

$\endgroup$
1
  • 1
    $\begingroup$ This is a very interesting (and impressively long!) set of notes; thanks for linking to them. I may make more comments after I've digested them. $\endgroup$ Apr 9, 2010 at 3:11
32
$\begingroup$

I want to post the following as a comment on many of the answers and comments already given.

Several people have said, "Well, watch out -- probability theory is not really the study of probability measures, but rather the study of certain quantities preserved under certain equivalence relations on probability measures, like distribution functions."

I certainly accept this point. In fact, I had more or less accepted it before I asked the question, although I admittedly didn't give much indication of this in the question itself. To be clear, I am aware "rewriting" impulses I have when reading about basic measure-theoretic probability are taking me in a direction away from the material of mainstream probability theory. I have two responses to this:

  1. Okay, let's agree that the definition and study of a category of probability spaces is not the domain of probability theory per se. But this does not mean that it's not useful or worth studying.

1a) If this branch of mathematics is not probability theory, what is it? [User "coudy" gave an answer saying that this is ergodic theory. I was unduly dismissive of this answer at first, and I apologize for that. I still don't think that "ergodic theory" is exactly the answer to my question, for instance because so far as I understand the subject it focuses almost exclusively on the dynamical aspects of iterating a measure-preserving transformation of a probability space. (By way of analogy, the branch of mathematics that studies the category of finite type schemes over a field $K$ is arithmetic geometry, not arithmetic dynamics.)

1b) While I agree that probability theory is at present not concerned with such structuralist questions, is it clear that it shouldn't be? Or, in less polemical terms, is there no advantage or insight to be gained by studying the structural aspects of probability spaces?

  1. I think an outsider to probability theory has a right to ask: "Okay, if probability spaces are really not the point of probability theory, why do they appear so prominently in all (so far as I know) modern foundations of the subject? Wouldn't -- or couldn't -- it be better to isolate exactly the structure that probability theory actually does care about and study this structure explicitly from the outset?"

By way of analogy, consider the notion of a "differentiable atlas" in the study of smooth manifold theory. Gian-Carlo Rota referred to atlases as a polite fiction, meaning (I think) that they are present in the foundations of the subject but do not really exist in the sense that the practitioners of the subject do not think about them and ask questions about them. They don't do any harm so long as you don't take them very seriously, but I have seen students get caught up on this point and "ask too many questions". The more modern approach of a structure sheaf seems like an improvement here -- it does the same work as an atlas but is something that the practitioners of the subject actually care about, so it is not at all a waste of time to "think deeply about structure sheaves". Indeed, the concept of "structure sheaf" is incredibly prevalent in other areas of mathematics, to the extent that if you are founding a new branch of geometry, knowing about structure sheaves will ease the birthing process.

So the dual question to 1) here is "What is the kind of mathematical structure that probability theorists are interested in studying?" (Happily, many of the very nice answers above do in fact address this question.)

$\endgroup$
1
  • 6
    $\begingroup$ Unfortunately I didn't see this nice answer/commentary until just now. A propos of your question 2), you might be interested to look at the classic two-volume probability text by Feller, in which probability spaces play a surprisingly small role and are not even introduced until well into volume 2. $\endgroup$ Sep 4, 2012 at 15:27
25
$\begingroup$

A category consists in a class of objects together with a class of morphisms. Measure theory together with morphisms between measure spaces is the topic of ergodic theory. So if you are interested in a categorical viewpoint at measure theory, you should take a look at advanced books on ergodic theory.

Here are some references. Glasner's book "ergodic theory via joinings" is probably what is close to a full blown categorical account of some basic concepts in ergodic theory. Rudolph's "Fundamentals of measurable dynamics: ergodic theory on Lebesgue spaces" is also pretty geared toward such an account. If you are interested in applications of ergodic theory to Lie group actions and diophantine approximation, you may want to consult the appendices in the books of R. Zimmer "ergodic theory and semisimple Lie groups". These appendices summarize the categorical results relevant to these questions.

Note however that many books on ergodic theory are pretty quick on the categorical stuff. Ergodic theory is a subject which is of interest to group theorists, dynamic people, probabilists, combinatorists, physicists, computer scientists,... So, really, it is not very useful to spend too much time on some fundational material that is irrelevant to these people, and to many applications.

In contrast to algebraic geometry, which is built like a cathedral, and for which category is a very interesting foundational material, ergodic theory is more like of a bazaar. Its structure is definitely transverse to the usual classification of mathematics (algebra, analysis, geometry), and even transverse to the classification of science (math, physics, computer science, biology) you may be accustomed to. Much of the steam in ergodic theory comes from the many interactions between these communities. It is important to keep the entrance level as low as possible to get as much people as possible on the boat. Putting forward a categorical approach in the textbooks or in conferences would do much harm to the field.

The references I provide should answer your four questions. Let me just add a comment. If you define a Borel space as a set endowed with a $\sigma$-algebra, you will soon run into many problems (e.g. a morphism at the level of the algebras not necessarily comes from a map between the sets, also a non-Borel non-Lebesgue measurable subset of $[0,1]$ endowed with the Lebesgue measure is a perfectly well defined measure space, and you definitely don't want it), so that's why people don't usually define it that way. There several choices in use at the moment, for example the Borel standard spaces (Zimmer appendices), and the Lebesgue spaces (Rudolph's book).

$\endgroup$
2
  • 5
    $\begingroup$ Pete, I think you were too quick to dismiss coudy's answer (and frankly, I don't see how you found it disrespectful). Many results and methods in probability theory can be rephrased in terms of ergodic theory, which means this is a perfectly on-topic response to your question. $\endgroup$ May 24, 2010 at 1:14
  • 5
    $\begingroup$ @Tom LaGatta: I agree with you, and awhile ago I deleted these comments and removed my downvote. In some sense coudy's answer is the closest I have received so far to one of the aspects of the question, although I still maintain that it is not quite dead-on. I have explained this in more detail in my CW answer below. $\endgroup$ Dec 19, 2010 at 10:44
20
$\begingroup$

There is an early paper by Victor Bogdan called "A new approach to the theory of probability via algebraic categories" (#54 here or here) which may be of interest.

$\endgroup$
1
  • $\begingroup$ Apologies. I seem to have clicked a down vote by mistake, and now can't undo it. $\endgroup$ Apr 20, 2017 at 13:57
18
$\begingroup$

As already noted, most probabilists identify random variables essentially with their distribution. The problem is that the kind of operations one can do with random variables often depend on the spaces they are defined on. The probabilitys spaces random variables are usually defined on, such as the unit interval with Lebesgue measure, do not allow for all the construction one wants to make (an uncountable family of independent random variables for example). In order to make all the constructions one wants to work with possible, one needs to work with more esoteric tools from measure theory. The problem is even larger when one turns to stochastic processes or adapted stochastic processes.

For this reason, people have worked on probability theory from the model theoretic view, which gives answers to existence questions much closer to the categorial view. A relatively readable introduction to this field is given in the book "Model Theory of Stochastic Processes" by Fajardo and Keisler. Their paper Existence Theorems in Probability might also be of interest.

$\endgroup$
2
  • 5
    $\begingroup$ I should have added a caveat to my answer that working with stochastic processes forces one to grapple with probability spaces more. But I've never heard of anyone actually wanting to consider an uncountable family of independent random variables. $\endgroup$ Apr 9, 2010 at 13:19
  • 6
    $\begingroup$ They actually occur in mathemtical economics. One wants to have a continuum of agents to apply analysis and one wants to be able that their independent actions cancel out in the aggregate. One wants a law of large numbers for such cases. Finding spaces on which one can make this work turned out to be hard but possible. $\endgroup$ Apr 9, 2010 at 13:36
16
$\begingroup$

Misha Gromov, "In a Search for a Structure, Part 1: On Entropy." https://www.ihes.fr/~gromov/wp-content/uploads/2018/08/structre-serch-entropy-july5-2012.pdf provides some interesting category-theoretic musings, among other things. One curious 'other thing' is that the Fisher metric is the flat metric on complex projective space.

He also gave a series of lectures on probability from the category theoretic perspective, "Probability, Symmetry, Linearity":

$\endgroup$
11
$\begingroup$

For a recent approach that looks to provide a better categorical environment for probability theory:

It replaces the category of measurable spaces, which isn't cartesian closed, with the category of quasi-Borel spaces, which is. As they point out in section IX, what they're doing is working with concrete sheaves on an established category of spaces, rather like the move to diffeological spaces.

$\endgroup$
8
$\begingroup$

Last year Voevodsky has given a talk at MIAN about his approach to probability theory; there is online a videorecording in Russian. I do not know if anything is written on this.

There was also an old Russian book (in Russian, afaik not translated, from the 70s) developing a somewhat similar approach but I do not quite remember the reference. I could look for it, though, if there is interest...

$\endgroup$
5
  • 7
    $\begingroup$ I don't know what is possible after 2 years - but if you're still here, I am interested $\endgroup$
    – SBF
    Mar 31, 2012 at 20:08
  • $\begingroup$ I am also interested. Such a book would be a great way to learn more about probability theory as well as to practice my mathematical Russian. $\endgroup$ Jan 13, 2017 at 12:37
  • $\begingroup$ What does "afaik" mean ? $\endgroup$ Mar 28, 2019 at 2:23
  • 1
    $\begingroup$ afaik it means "as far as I know" $\endgroup$
    – Vincent
    Jul 9, 2019 at 9:50
  • $\begingroup$ For what it's worth, a recording online (title is "Categorical Probability") is youtube.com/watch?v=TIfo1aIeBN4 and a set of notes by Voevodsky in English is item 57 on the page math.ias.edu/Voevodsky/voevodsky-publications.html. Click the .pdf link there. $\endgroup$
    – KConrad
    Jan 15, 2020 at 5:40
6
$\begingroup$

This old question and my old answer continue to get occasional attention, and I believe it's time for a new answer.

Is there an introduction to probability theory from a structuralist/categorical perspective?

I now say: at the time you asked the question, I don't believe there was a well-developed structural/categorical approach to probability theory (as opposed to probability spaces). But by now there is at least one such approach that I find compelling, namely the theory of Markov categories as developed by Tobias Fritz, Paolo Perrone, and their collaborators. This approach is still being developed, and I'm just beginning to really learn about it, but I really like what I've seen so far. I won't attempt an introduction to it here, or an explanation of what I like better about it than other approaches I've seen, but this answer by Tobias Fritz to another question summarizes some highlights.

I will point out, though, that I and many others have argued that probability theory is not well described by any category of measure spaces and measurable maps; the categories that arise here are more subtle beasts. Basically, one works with "the Kleisli category of the Giry monad on the category of measurable spaces and measurable maps", or some variation thereof. Fortunately for us non-category theorists, this thing can be given a concrete description (well-explained in this paper): the objects are measurable spaces, the morphisms are stochastic maps (basically Markov kernels, or functions with random outputs), and there is some additional structure.

One striking aspect of this picture is that the familiar basic things in probability, like probability spaces and random variables, show up not as objects or morphisms but as diagrams in this category. And more complex diagrams can encode more complicated things, like coupling arguments, that probabilists make use of every day.

I particularly like the fact that a probability measure on $\Omega$ is naturally identified with $I \xrightarrow{\varphi} \Omega$, where $I$ is some fixed one-element space and $\varphi$ is a stochastic map. This is reminiscent of the fact that in the category of sets and functions, elements of a set $X$ can be identified with functions $I \xrightarrow{f} X$. (In both cases, $I$ is the identity for a natural monoidal structure on the category in question.) Intuitively: a probability measure on $\Omega$ is the same thing as a random element of the set $\Omega$!

This perspective on elements of sets gives a way to talk about elements in structural approaches to set theory like ETCS, while simultaneously discouraging you from focusing on the action of functions on individual elements. In the same way, this perspective on probability naturally discourages you from focusing on the pointwise behavior of random-variables-as-measurable-functions. Which any probability theorist will tell you is exactly what you should not focus on in probability theory.

$\endgroup$
2
  • 2
    $\begingroup$ I am reminded that someone wiser than me wrote that "good general theory does not search for the maximum generality, but the right generality" $\endgroup$
    – Yemon Choi
    Feb 8, 2023 at 21:09
  • $\begingroup$ Thanks, Mark! It's nice to hear that things are advancing in this subject. $\endgroup$ Feb 8, 2023 at 21:16
5
$\begingroup$

I find this: http://etd.library.pitt.edu/ETD/available/etd-04202006-065320/unrestricted/Matthew_Jackson_Thesis_2006.pdf. (Wayback Machine, A Sheaf Theoretic Approach to Measure Theory by Matthew Jackson)

Anyway I find "Bichteler :Integration, Springer LNM 315"

it is about the foundation of the theory, the style is similar to Bourbaki, and may be adaptable for a categorical view.

$\endgroup$
5
$\begingroup$

Bart Jacobs has a new textbook out now. It is on his webpage. It is called "Structured Probabilistic Reasoning" I believe this is going to be an important reference in the next five years.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.