Zulip Chat Archive

Stream: Is there code for X?

Topic: Can't find a theorem that works


Timothy Harevy (Feb 22 2024 at 12:00):

Want to rewrite the following

(n)⁻¹ < ε

as

1<(n)*ε

Epsilon is in the Reals and n in the natural numbers tried using inv_lt_iff_one_lt_mul which is defined as

theorem inv_lt_iff_one_lt_mul {α : Type u} [Group α] [LT α] [CovariantClass α α (Function.swap fun x x_1 => x * x_1) fun x x_1 => x < x_1] {a : α} {b : α} :
a⁻¹ < b  1 < b * a

but this doesn't work I'm not sure what I'm missing

Damiano Testa (Feb 22 2024 at 12:11):

The lemma that you mention applies to a Group (under multiplication), which the reals are not. The lemma docs#div_lt_iff is close to what might be able to use, but it applies to a / b, rather than x⁻¹.

Damiano Testa (Feb 22 2024 at 12:12):

If the form of inv_lt_iff does not really exist, then it might also be a lemma that could be added to Mathlib.

Timothy Harevy (Feb 22 2024 at 12:18):

thank you might give it a go then

Kevin Buzzard (Feb 22 2024 at 12:33):

Note that the rewrite is impossible in general because n=0 satisfies the first equation but not the second (at least I think it does, android is not displaying the first equation correctly for me because of a known bug, the point is that 1/0=0 in lean)

Floris van Doorn (Feb 22 2024 at 15:42):

Is this docs#inv_pos_lt_iff_one_lt_mul (or docs#inv_pos_lt_iff_one_lt_mul')?


Last updated: May 02 2025 at 03:31 UTC