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

Suppose (Ln) is a sequence of loops in a torus S1 × S1 converging in the Hausdorff metric to some set L in the torus. Suppose also that for each loop Ln the projection map p:S1 × S1 -> S1 defined by p(x,y) =x when restricted to Ln is not null-homotopic, then can we conclude that the restriction of the projection map p to L is also not null homotopic? Or are there counter examples?

share|cite|improve this question
up vote 4 down vote accepted

Your intuition is correct: the map p : L → S1 is not nullhomotopic.

If p : L → S1 was nullhomotopic, then it would factor through the universal cover ℝ of S1.
The inclusion ι : L → S1 × S1 would therefore factor through ℝ × S1.
Let ι' : L → ℝ × S1 denote a lift of ι, and let L(0) := ι'(L).
The subset L(0) ⊂ ℝ × S1 is then one of the connected components of the preimage of L in ℝ × S1.
Let us call the other components L(n) for n ∈ ℤ.

Let C0 ⊂ ℝ × S1 be a loop in that separates L(0) from L(1).
We may also assume that C0 is not nullhomotopic in ℝ × S1 [this still needs a small argument...].

The projection CS1 × S1 of C0 is then a loop in the complement of L,
and represents the element (0,1) of ℤ×ℤ = π1(S1 × S1).

Since L lies in the complement of C and Ln → L in the Hausdorff metric,
there exists an n such that Ln lies in the complement of C.
This contradicts your assumtion that p : LnS1 is nullhomotopic.

share|cite|improve this answer
Thanks for the answer. It appears that the notation L<sub>n</sub> is being used in two different contexts. could you please clarify. – arun s May 22 '10 at 17:02
I fixed the mismatch of notations. – André Henriques May 22 '10 at 20:17

EDITED. The following example should not be conisdered as a contre-example, it is just a well known example of a pathology that can happen.

EXAMPLE. $L$ is composed of a union of the vertical circle $(0, S^1)$ and a disjoint $R^1$ that is emdedded in $T^2$ in such a way, that it accumulates to $(0, S^1)$ from both sides. This $R^1$ projects one-to one to the horisontal circle without a point and can be represented a as graph (function of $x$), than if is given by the following formula:

$y= (sin(1/x))$

The point is that such $L$ can be Hausodrff approximated by a sequence of circles $s_n$ that are not null-homotopic ($s_n$ wiggles more an more near the vertical circle $(0, S^1)$ as $n\to \infty$).

The problem with this example, is that the topology on the union of $S^1$ and $R^1$ that we should take (I guess) is the toplogy induced from $T^2$. And for this induced topology, I guess the projection to the horisontal cicle is non null-homotopic...

share|cite|improve this answer
This is not a counterexample. – André Henriques May 22 '10 at 16:16
I dont understand why the inclusion composed with the proyection is nullhomotopic. – rpotrie May 22 '10 at 16:24

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.