Skip to main content
8 events
when toggle format what by license comment
Nov 9, 2017 at 21:04 comment added Eilon Thanks. In my original question I had $\mathbb{R}^n_+$ instead of $\mathbb{R}^n$ throughout, and I thought that changing this to $\mathbb{R}^n$ would not matter. In fact, it does not, since my son came up with a counterexample to the case where we require $M$ to be a subset of the nonnegative orthant: take $M = (t,x,(t-\frac{1}{2})^2)$, where $t,x \in [0,1]$ and $\frac{x}{3} \leq t \leq \tfrac{2+x}{3}$.
Nov 9, 2017 at 4:25 comment added fedja That is obviously false as written now: take $n=1$ and $M=\{(y-x+0.5)^2\le 0.01\}$. Then at every particular point you have an interval of length $0.2$ of admissible values, so you can choose a value of size at least $0.1$ and extend linearly between the points but your $f$ has to change sign on $[0,1]$, so the intermediate value theorem is a killer. I surmise you meant something else.
Nov 8, 2017 at 14:38 comment added Dirk Oh, I missed that $M\subseteq [0,1]\times \mathbb{R}^n$ (somehow I assumed that $M\subseteq \mathbb{R}^n$…)
Nov 8, 2017 at 12:01 history edited Eilon CC BY-SA 3.0
deleted 2 characters in body
Nov 8, 2017 at 12:00 comment added Eilon Dear Dirk, I would not mind having a constant function, as long as its graph is a subset of M (and the constant is above $\delta_0$).
Nov 8, 2017 at 10:05 comment added Dirk A constant function $x$ is not what you want, right?
Nov 8, 2017 at 9:44 history edited Eilon CC BY-SA 3.0
deleted 2 characters in body
Nov 8, 2017 at 9:09 history asked Eilon CC BY-SA 3.0