0

I have no idea how to solve this. It's easy to show that the open ball or the open cube is convex, but how do we show that the open rectangle is? (The same holds for the closed rectangle).

Thanks

Derek O
  • 378

2 Answers2

3

I assume that you are working in a vector space. Let's prove that if $U_1 \subseteq V_1$ and $U_2 \subseteq V_2$ are convex, where $V_i$ are vector spaces, then $U_1 \times U_2 \subseteq V_1 \times V_2$ is also convex. This is sufficient, since every interval in the real line is convex, and rectangles are finite products of intervals.

Let $(x,y),(x,y') \in U_1 \times U_2$. Let $t \in [0,1]$ be arbitrary. Since $x,x' \in U_1$ and $U_1$ is convex, we have $tx+(1-t)x' \in U_1$. Since $y,y' \in U_2$ and $U_2$ is convex, we have $ty+(1-t)y' \in U_2$. Hence $$t(x,y)+(1-t)(x',y') = (tx+ (1-t)x',ty+(1-t)y') \in U_1 \times U_2.$$

Ivo Terek
  • 80,301
1

I am reading "Analysis on Manifolds" by James R. Munkres.
This exercise is the same exercise as Exercise 4(b) on p.40 in this book.

Let $t\in [0,1].$

Let $\mathbf{c},\mathbf{d}\in (a_1,b_1)\times\dots\times(a_n,b_n).$
Then, $$a_i<\min\{c_i,d_i\}\leq (1-t)c_i+td_i\leq\max\{c_i,d_i\}<b_i.$$ So, $\mathbf{b}+(1-t)\mathbf{c}\in (a_1,b_1)\times\dots\times(a_n,b_n).$


Let $\mathbf{c},\mathbf{d}\in[a_1,b_1]\times\dots\times[a_n,b_n].$
Then, $$a_i\leq\min\{c_i,d_i\}\leq (1-t)c_i+td_i\leq\max\{c_i,d_i\}\leq b_i.$$
So, $\mathbf{b}+(1-t)\mathbf{c}\in [a_1,b_1]\times\dots\times [a_n,b_n].$

tchappy ha
  • 9,894
  • i like this method as it's straight from the definition of convexity. also quite intuitive for those with a real analysis background – Derek O Feb 13 '24 at 21:41