Uncountably Many Maxima?

Calculus Level 4

Let f : R R f : \mathbb{R} \to \mathbb{R} . Let S S be the set of all strict local maximum of f f . Can S S be uncountable?


  • A strict local maximum is a value x x such that there exists a neighborhood N x N_x of x x where for all y N x y \in N_x , f ( y ) < f ( x ) f(y) < f(x) . A constant function has no strict local maximum, because f ( y ) f ( x ) f(y) \not< f(x) ; just being equal is not enough.
  • A set is uncountable if there is no injection from it to the natural numbers. In other words, its elements cannot be listed exhaustively in a sequence a 1 , a 2 , a 3 , a_1, a_2, a_3, \ldots .

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

2 solutions

  • Given δ > 0 \delta > 0 , define S δ S_{\delta} to be the points x x such that everything in a δ \delta -neighborhood of x x is strictly smaller than x x .

S δ = { x R y B δ ( x ) : f ( y ) < f ( x ) } S_{\delta} = \left \{ x \in \mathbb{R} \mid \forall y \in B_{\delta}(x) : f(y) < f(x) \right \}

  • Now, notice that given any δ \delta , we could divide up the number line into δ \delta intervals, and there would be at most one such δ \delta -maxima in each interval. To be precise, for each k Z k \in \mathbb{Z} , there exists at most one point x [ k δ 2 , ( k + 1 ) δ 2 ] x \in [k\frac{\delta}{2},(k+1)\frac{\delta}{2}] such that x S δ x \in S_\delta . We conclude that, for any δ > 0 \delta > 0 , S δ S_{\delta} is atmost countable.

  • Notice that if δ 1 > δ 2 \delta_1 > \delta_2 then x S δ 1 x S δ 2 x \in S_{\delta_1} \implies x \in S_{\delta_2} . Hence, one could express S S as a countable union of decreasing δ \delta 's.

n = 1 S 1 n \bigcup_{n=1}^\infty {S_{\frac{1}{n}}}

  • Because a countable union of countable sets is countable, we can say that S S is atmost countable.

Inspired by Stack Exchange

Ivan Koswara
Mar 24, 2017

I don't actually have a complete solution to this yet. My idea is to pick the neighborhoods for each x x in such a way so that no two neighborhoods intersect. If we can do this, then we can map each x S x \in S to a rational number in N x N_x ; the map is injective, so S Q = 0 |S| \le |\mathbb{Q}| = \aleph_0 , proving S S cannot be uncountable. Of course, this approach might be entirely incorrect.

Unfortunately, that does not work.

For example, if we take f ( x ) = sin 1 x , x 0 f(x) = \sin \frac{1}{x} , x \neq 0 and f ( 0 ) = 2 f(0) = 2 , then we have a strict local maxima at x = 2 ( 4 n + 1 ) π x = \frac{2}{(4n + 1 ) \pi} or x = 0 x = 0 . It is clear that any neighbourhood of 0 would contain another stict local maxima.

Calvin Lin Staff - 4 years, 2 months ago

Log in to reply

Ah, that's true. In that case, I picked the correct answer entirely by luck. (And by intuition, but the solution idea I had from that intuition was wrong.)

I can at least prove that for any ϵ > 0 \epsilon > 0 , there exists an interval of length ϵ \epsilon that contains uncountably many strict local maxima. But I don't think this leads to any solution either.

Ivan Koswara - 4 years, 2 months ago

Log in to reply

@Agnishom Chattopadhyay Did you forget the condition of "continuous function"?

Continuity would prevent the counter example that I stated.

Calvin Lin Staff - 4 years, 2 months ago

Log in to reply

@Calvin Lin Since your counterexample only gives countably many strict local maxima, it's still possible that the problem works even with non-continuous f f , only that we need to find a proof that doesn't depend on continuity.

By the way, right, I found a function that has a strict local maximum on each rational number (Thomae's function), so the idea of having a neighborhood around each strict local maximum doesn't work. Although the idea of mapping each strict local maximum to some rational number might still work, maybe...

Ivan Koswara - 4 years, 2 months ago

Log in to reply

@Ivan Koswara That's right. More accurately, I only have a proof for continuous functions.

I do not know if the statement is still false for functions generally. I suspect it is not.

Calvin Lin Staff - 4 years, 2 months ago

@Calvin Lin I think this statement is true regardless of continuity.

Agnishom Chattopadhyay - 4 years, 2 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...