Zulip Chat Archive

Stream: Is there code for X?

Topic: 2^x ≤ 1 + x for x ∈ [0,1]


Geoffrey Irving (Jan 17 2024 at 17:48):

Do we have this?

import Mathlib

lemma two_rpow_le_one_add {x : } (x0 : 0  x) (x1 : x  1) : 2^x  1 + x := by
  sorry

Picture from https://www.wolframalpha.com/input/?i=plot+2%5Ex+and+1+%2B+x:

Screenshot-2024-01-17-at-5.49.40PM.png

Geoffrey Irving (Jan 17 2024 at 17:51):

I suppose it's just convexity, so docs#convexOn_rpow is the answer.

Geoffrey Irving (Jan 17 2024 at 17:53):

Oops, no, that's convexity in the left.

Yaël Dillies (Jan 17 2024 at 19:14):

I don't think we have it

Geoffrey Irving (Jan 17 2024 at 19:16):

Easy to prove from convexity of exp, which I’ll do shortly.

Yaël Dillies (Jan 17 2024 at 19:17):

You could even make it an iff! 2^x ≤ 1 + x ↔ x ∈ Icc 0 1

Geoffrey Irving (Jan 17 2024 at 19:54):

This is odd, but I can't find the lemma that is the normal f (a * x + b * y) <= a * f x + b f y definition of convexity. The closest I can find is convexOn_iff_forall_pos, which is non-ergonomic since the inequalities are strict. There are a ton of other inequalities, but it seems like the very basic one has to also exist.

Ah, I see, it's just .2.


Last updated: May 02 2025 at 03:31 UTC