MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I'm interested in stratifications of smooth maps $\mathbb{R}^n\to\mathbb{R}$ (or more generally of any $n$-manifold $M^n\to\mathbb{R}$). The codimension 0 stratum should be Morse functions, and the codimension 1 stratum should be Morse cancellations, e.g. the $t=0$ value of the following 1-parameter family of maps $$ (x_1,\ldots,x_n) \mapsto tx_1 + x_1^3 \pm x_2^2 \pm\cdots\pm x_n^2 . $$ Is there a good reference for the general codimension $k$ case?

Another way of phrasing the question: given a $k$-parameter family of smooth maps $F: P^k\times \mathbb{R}^n\to\mathbb{R}$, is there a known list of specific singularities such that we may assume that $F(p, \cdot)$ has only these singularities after a small perturbation? I suppose the way to start is to make $F$ Morse as a map from an $(n+k)$-manifold to $\mathbb{R}$, then look at the ways the coordinate axes of $P\times \mathbb{R}$ line up with gradients and the eigenspaces of the hessian of the Morse singularities of $F$. But I would rather cite the details than work them out for myself.

If the general case is messy (instability, cross-ratios, etc.), I would also be interested in an answer for $n=2$.

share|cite|improve this question

It looks to me that what you are really interested in is the Thom-Boardman stratification of the function space. For that I would recommend the well-written, Stable Mappings and Their Singularities by Guillemin and Golubitsky (in the Springer GTM series).

share|cite|improve this answer
Guillemin and Golubitsky don't quite set it up in this setting -- they develop the general Mather machine but they don't apply it to real-valued function spaces, at least, not in any real detail. – Ryan Budney Feb 10 '11 at 23:03
Granted. I guess, the other possible reference might be Cerf, at least when the codimension is small. – John Klein Feb 10 '11 at 23:29
Yup. Cerf starts from the Thom-Mather machine and works out the details of the singularities in the real-valued function case. Guillemin and Golubitsky is probably the most readable account of Thom-Mather theory out there, as far as I know. – Ryan Budney Feb 11 '11 at 1:25
Thanks John, I'll have a look at G&G. – Kevin Walker Feb 11 '11 at 4:15
A good supplement to G&G which provides an alternative understanding of the Thom-Boardman stratification is Porteous' paper – Sergey Melikhov Feb 11 '11 at 15:58

A standard reference is:

F. Sergeraert "Un theoreme de fonctions implicites sur certains espaces de Frechet et quelques applications," Ann. Sci. Ecole Norm. Sup. (4) 5 (1972), 599-660.

This isn't a stratification of the space of maps $M \to \mathbb R$ but it is a stratification of an infinite co-dimension subspace of the space of all smooth maps $M \to \mathbb R$. It's a relatively popular stratification to use among geometric topologists, in that it produces Cerf theory. Rubinstein, Hong and McCullough use it in their work on the homotopy-type of $\operatorname{Diff}(L_{p,q})$. (which is how I learned of it)

Is this roughly what you're looking for?

share|cite|improve this answer
Thanks -- I'll have a look at those references. The paper by J.W. Bruce in the bibliography of Rubinstein et al also looks interesting. – Kevin Walker Feb 10 '11 at 22:00
Ryan: you misspelled Sergeraert's name. – Thierry Zell Feb 11 '11 at 0:09
Ah, thanks for catching that. I made the same mistake on the Cerf Theory Wikipedia page (twice!). – Ryan Budney Feb 11 '11 at 1:05
My knowledge of French is pretty minimal, but I didn't see anything like an explicit list in chapters 8 or 9 of Sergeraert. Is the information necessary to produce an explicit list in there, or is the description of the stratification more abstract than that? – Kevin Walker Feb 11 '11 at 4:14
I think it's fairly explicit at some point though I haven't looked at it in much detail in over a year. I think the best thing to do would be to see how Rubinstein & McCullough cite the work, because they're looking for something in the ballpark of what you want. It's one of those moments in the semester where I rarely get my head out of paperwork so I won't have much more to say for a few weeks. – Ryan Budney Feb 14 '11 at 23:54

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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