Converting 2 COD (x,y) into 1 Hilbert curve COD?

  • Context: Undergrad 
  • Thread starter Thread starter greswd
  • Start date Start date
  • Tags Tags
    Curve Hilbert
Click For Summary

Discussion Overview

The discussion revolves around the conversion of two coordinates (x, y) on a unit square into a single coordinate along a Hilbert curve. Participants explore the mathematical implications and potential formulas for this conversion, touching on concepts such as arc length parametrization and recursive approaches.

Discussion Character

  • Exploratory
  • Mathematical reasoning
  • Debate/contested

Main Points Raised

  • Some participants inquire about the formula for converting 2D coordinates into a 1D Hilbert curve coordinate, suggesting that this involves a transformation of the coordinates.
  • One participant proposes a method of interleaving the digits of x and y to create a new coordinate, questioning if this aligns with the concept of converting 2D to 1D.
  • Another participant suggests that the arc length parametrization of the Hilbert curve may provide a useful framework for understanding the conversion.
  • A participant mentions the need for a function of x, y, and n that returns a fraction for a Hilbert curve of order n, indicating a desire to identify patterns through incremental values of n.
  • There is a mention of the limit for n approaching infinity and the distinction between true Hilbert curves and pseudo Hilbert curves for finite n.
  • One participant expresses uncertainty about the problem statement and emphasizes the collaborative nature of the forum, highlighting the challenge of finding a complete solution.
  • Another participant discusses the complexity of deriving functions for x and y in terms of the Hilbert curve length L, noting the existence of infinite equations and the challenge of finding a specific solution.

Areas of Agreement / Disagreement

Participants do not reach a consensus on the formula or method for converting the coordinates, and multiple competing views and approaches remain throughout the discussion.

Contextual Notes

Participants express uncertainty regarding the completeness of their problem statements and the complexity of the mathematical relationships involved, particularly with respect to infinite equations and inverse functions.

greswd
Messages
764
Reaction score
20
COD stands for co-ordinate.

As the title says, you have two co-ordinates of a point, x and y, on a unit square.

What's the formula for converting these two co-ordinates into a single Hilbert curve co-ordinate?

Which represents the percentile along the length of the Hilbert Curve that point is on.
 
Physics news on Phys.org
Inspired by this video

 
greswd said:
What's the formula for converting
Would this be converting 2D to 1D ?

I'm inlcined to think along lines like: write x and y as real numbers. Create a z as follows: first digit is first digit of x, second digit is first digit of y -- third digit is second of x, fourth second of y, etc. etc.

Ever hear of the Hilbert hotel ?
 
Isn't this just the arc length parametrization of the Hilbert curve? Not that it is simple but it seems like a useful way of framing it.
 
Ah, I completely misread this thing o:) -- totally off.
Thanks @WWGD

So @greswd , you want a function of ##\ x, y\ ## and ##n## that returns a fraction for a Hilbert curve of order ##n## starting at the lower left ?

Start with ##n = 1##, then ##2## etc and see if you can find a pattern :rolleyes:
 
  • Like
Likes   Reactions: WWGD
Found some time to watch the video -- cute !
I understand you want the limit for ##n\uparrow \infty## (the ones with finite ##n## are pseudo Hilbert curves). Advice is the same.

Found one value already: ##f(1/2, 1/2) = 1/2 ## :wink:
 
  • Like
Likes   Reactions: greswd
Susspect the time 4:00 -- 5:00 min in the video is important: suppose you work with base 4 numbers and work out the diagonal flip of first and last quadrant ...then make it recursive ..

Just a wake-up thought ...

Intriguing !
 
WWGD said:
Isn't this just the arc length parametrization of the Hilbert curve? Not that it is simple but it seems like a useful way of framing it.
If arc length parametrization is what the video is doing, then yes
 
BvU said:
Ah, I completely misread this thing o:) -- totally off.
Thanks @WWGD

So @greswd , you want a function of ##\ x, y\ ## and ##n## that returns a fraction for a Hilbert curve of order ##n## starting at the lower left ?

Start with ##n = 1##, then ##2## etc and see if you can find a pattern :rolleyes:
thanks, do you know the function for x and y in terms of the fraction and n?
 
Last edited:
  • #10
No idea yet -- first trying to get a good and complete problem statement :wink: in my chaotic mind .

Anyway, PF isn't about doing the work for you, it's about helping you to do the work :biggrin:
(however much we'd like to grab it and do it ourselves o0) )

If I ever have the time, I think I'd start reading this
The heatmap here is nice too

But you already googled those, right :rolleyes: ?
 
  • #11
BvU said:
No idea yet -- first trying to get a good and complete problem statement :wink: in my chaotic mind .

Anyway, PF isn't about doing the work for you, it's about helping you to do the work :biggrin:
(however much we'd like to grab it and do it ourselves o0) )

If I ever have the time, I think I'd start reading this
The heatmap here is nice too

But you already googled those, right :rolleyes: ?

yeah I have haha. The problem is "infinitely" difficult. :-p

Consider the ratio along the length of the HC, L.
From L, you can get a value for X and a value for Y.

So, you get a function for X in terms of L and the same for Y. Next, the inverse functions, L in terms of X and L in terms of Y.
There are an infinite number of inverse functions.

You get two sets of infinite equations, one from X and one from Y.

There can only be one specific value of L which can be the solution to the equations in both infinite sets. And it is the solution to only one equation in each set.

This is way beyond me lol. I just want to know the formula for L in terms of X and Y.
 

Similar threads

  • · Replies 14 ·
Replies
14
Views
3K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 61 ·
3
Replies
61
Views
6K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 9 ·
Replies
9
Views
2K
  • · Replies 18 ·
Replies
18
Views
2K
  • · Replies 14 ·
Replies
14
Views
2K
  • · Replies 3 ·
Replies
3
Views
3K
  • · Replies 2 ·
Replies
2
Views
3K
  • · Replies 4 ·
Replies
4
Views
3K