Zulip Chat Archive

Stream: new members

Topic: Piecewise-Linear Functions


Junjie Bai (Mar 23 2024 at 13:31):

(deleted)

Junjie Bai (Mar 23 2024 at 13:44):

I notice that the piecewise-linear function has not been defined in Mathlib yet(or please tell me if I just ignore it), and I have to use some properties of it to do some proofs, I found this but it's not enough.
I'm dealing with a function which is singular at every positive integer, and I want to proof it is monotone , bijective, and equal to the other forms(in math I will proof it by deriv on them), and I think I want to realize the following:

  1. definition of piecewise-linear function(not only on R)
  2. how to show they are continuous or not?
  3. how to show they are surjective or not?
  4. how to compare them with the other piecewise-linear function?
    Does anyone have any ideas of how to realize all of this? please offer some advises, thanks a lot.

Matt Diamond (Mar 23 2024 at 19:16):

What's wrong with using if ... then ... else statements, like in the example you linked to?

Junjie Bai (Mar 24 2024 at 03:44):

I have to dealing with the condition that the function have infinite singular points, so I can't use if ...then ...else statements.

Matt Diamond (Mar 24 2024 at 03:45):

Could you share a mathematical definition of the function?

Junjie Bai (Mar 24 2024 at 04:02):

Sure, it's 0udt/[G0Gt]\int_0^u dt/[G_0:G_t] , where GiG_i is the lower numbering ramification group.

Matt Diamond (Mar 24 2024 at 04:24):

Thanks... unfortunately I can't help with this, but I'm sure there are others here who can point you in the right direction

Matt Diamond (Mar 24 2024 at 04:35):

this recent thread might be relevant: https://leanprover.zulipchat.com/#narrow/stream/116395-maths/topic/Formalizing.20Ramification.20Groups

Junjie Bai (Mar 24 2024 at 05:16):

Thank you!

Kevin Buzzard (Mar 24 2024 at 08:49):

Are you sure you need an API for this? What is a concrete example of a theorem you need to prove about this function? Don't you just need to bound the integrand below by 1/#G_0 and show it's continuous to prove surjectivity, for example?

Junjie Bai (Mar 24 2024 at 12:45):

I think this can solve my problem, thanks a lot!

Junjie Bai (Mar 26 2024 at 11:10):

Kevin Buzzard said:

Are you sure you need an API for this? What is a concrete example of a theorem you need to prove about this function? Don't you just need to bound the integrand below by 1/#G_0 and show it's continuous to prove surjectivity, for example?

I solve most of my problems, but I can't solve the surjectivity since I'm not familar with the functions about continuous and integration, can you be more specific please? Such as where can I found the functions I need to use, or which one should I use? Thanks a lot!

Kevin Buzzard (Mar 26 2024 at 11:26):

I don't know this myself. I'm hoping that the integral will be continuous (you're integrating a step function constant on (n,n+1) for all n) and strictly increasing (because the integrand is always at least 1/#G0>0) and its value at 0 will be 0 and the integrand is always at least 1/#G0 so the integral will be at least u/#G0 so will tend to infinity and now surely the analysts will take over and show you how to prove it's a bijection from [0,infty) to itself :-)

Junjie Bai (Mar 26 2024 at 15:55):

Thanks! I'll try again.

Kevin Buzzard (Mar 26 2024 at 16:52):

If you can't do the basic analysis goals yourself, maybe make some MWE which doesn't mention ramification groups and ask the analysts how to do it :-) (e.g. just have some arbitrary f : Nat -> Real with the hypothesis that a <= b implies f(a) >= f(b) and that there exists some c > 0 such that f(n)>=c for all n, and then integrate x mapsto f(floor x) etc etc) (I should say that I don't have a clue how to prove or even state most of the things in my previous post, I've not used integration at all in mathlib)

Junjie Bai (Mar 27 2024 at 01:28):

That's useful, I'll try to make some MWE first.
I've not used integration either, I'm learning it.


Last updated: May 02 2025 at 03:31 UTC