Swammerdami
Squadron Leader
Girls just want to have fun is a song written by Robert Hazard and made famous by Cyndi Lauper. But a little-known fact is that the original lyrics were "Mathematicians just want to have fun!" Hazard just couldn't get the meter to scan properly with the original lyrics.
A rather famous example of doubling a set is to take a solid 3-D ball; divide it into five pieces; operate one of 5 simple distance-preserving (and volume-preserving) transforms on each piece and -- Presto! -- end up with TWO solid balls, each with the same size as the original ball!!
Although the ball-doubling is very famous, there was no million-dollar prize for it as there is for proving Poincare's Conjecture or the Riemann Hypothesis. That's because nobody posed the question "How do you double a ball using volume-preserving transforms?" They never posed the question because it's obviously impossible! Yet "Mathematicians just want to have fun!" and sometimes do "impossible" but fun things before breakfast (though seldom before a cup of coffee).
I plan to compose a few posts on the topic of set doubling. If I don't run out of patience, I may get around to a brief summary of the ball-doubling. But participation by others is welcome. Feel free to Google the "Banach-Tarski Paradox" (ball doubling) and post about it before I do.
We will start with a very simple doubling. But first, let's use an example to define a simple notation for manipulating sets. When placing a set where a number or element should go, we're asking that the expression be evaluated for all elements of the set, with the result forming a new set. To avoid confusion, capital letters will always denote sets.
(0,1] denotes the set {x | 0 < x and x ≤ 1} but in keeping with the convention that sets are denoted by capital letters, we write K(0,1] = (0, 1] or more generally
This simple doubling already gives us a simple way to double a ball, except for a nitpick. We just partition the ball into its infinitely-many radii; double each radius; offsetting one copy of each such doubled pair; and presto! Two balls. The very center of the ball is troublesome. Do we include it in a radius or not? If we don't include it, the second ball will be missing the single point at its very center.
More formally: Each point in the ball B can be denoted with three coordinates (α,β,r) where r is the distance from the ball's center and (α,β) are the angular coordinates of radius. Now partition the ball of radius-length 1 into these infinitely many radii:
We've not handled the very center of ball properly -- we had to separate the singleton {(0,0,0)} -- but let's gloss over that detail for now.
Of course this is NOT the Banach-Tarski Paradox. We've doubled the ball but we did NOT do that with distance-preserving transforms let alone with volume-preserving transforms
Hey! How is that even possible?? If the ball has volume v, then the doubled ball will have volume 2⋅v. The Banach-Tarski ball doubling must be impossible, right?
Spoiler alert: Banach and Tarski will partition the ball into pieces for which the volume, like 0÷0, is undefined! But that will be the least of our worries on the Banach-Tarski roller-coaster ride.
Well, this concludes episode 1. I've probably already made mistakes, and certainly sown confusion. Submit your comments and questions. I hope to compose episode 2 tomorrow.
A rather famous example of doubling a set is to take a solid 3-D ball; divide it into five pieces; operate one of 5 simple distance-preserving (and volume-preserving) transforms on each piece and -- Presto! -- end up with TWO solid balls, each with the same size as the original ball!!
Although the ball-doubling is very famous, there was no million-dollar prize for it as there is for proving Poincare's Conjecture or the Riemann Hypothesis. That's because nobody posed the question "How do you double a ball using volume-preserving transforms?" They never posed the question because it's obviously impossible! Yet "Mathematicians just want to have fun!" and sometimes do "impossible" but fun things before breakfast (though seldom before a cup of coffee).
I plan to compose a few posts on the topic of set doubling. If I don't run out of patience, I may get around to a brief summary of the ball-doubling. But participation by others is welcome. Feel free to Google the "Banach-Tarski Paradox" (ball doubling) and post about it before I do.
We will start with a very simple doubling. But first, let's use an example to define a simple notation for manipulating sets. When placing a set where a number or element should go, we're asking that the expression be evaluated for all elements of the set, with the result forming a new set. To avoid confusion, capital letters will always denote sets.
K = {1,2,3,4,5}
10⋅K+7 = {17, 27, 37, 47, 57}
We will also work with partitions. WhenK = G1 ⋃ G2 ⋃ G3 ⋃ G4 ⋃ G5
and ∀i,j Gi ⋂ Gj = Ø
we say that (G1, G2, G3, G4, G5) is a partition of K(0,1] denotes the set {x | 0 < x and x ≤ 1} but in keeping with the convention that sets are denoted by capital letters, we write K(0,1] = (0, 1] or more generally
K(a,b] = {x | a < x and x ≤ b}
Now let's partition K(0,1] into two subsetsK(0,1] = K(0,.5] ⋃ K(.5, 1]
But each of the subsets can be transformed into a copy of the original set!2⋅K(0,.5] = K(0,1]
2⋅K(.5, 1]-1 = K(0,1]
(In practice we might substitute 2⋅K(.5, 1]+99 = K(0,1]+100 so that we finish with two disjoint copies of the original set.)This simple doubling already gives us a simple way to double a ball, except for a nitpick. We just partition the ball into its infinitely-many radii; double each radius; offsetting one copy of each such doubled pair; and presto! Two balls. The very center of the ball is troublesome. Do we include it in a radius or not? If we don't include it, the second ball will be missing the single point at its very center.
More formally: Each point in the ball B can be denoted with three coordinates (α,β,r) where r is the distance from the ball's center and (α,β) are the angular coordinates of radius. Now partition the ball of radius-length 1 into these infinitely many radii:
B = {(0,0,0)} ⋃ B(α,β)1 ⋃ B(α,β)2 ⋃ B(α,β)3 ⋃ ...
where each B(α,β)i has the form (αi, βi, K(0,1]). That is, it is equivalent to the simple K(0,1] = (0, 1] segment we considered above. We showed how to double that above.We've not handled the very center of ball properly -- we had to separate the singleton {(0,0,0)} -- but let's gloss over that detail for now.
Of course this is NOT the Banach-Tarski Paradox. We've doubled the ball but we did NOT do that with distance-preserving transforms let alone with volume-preserving transforms
Hey! How is that even possible?? If the ball has volume v, then the doubled ball will have volume 2⋅v. The Banach-Tarski ball doubling must be impossible, right?
Spoiler alert: Banach and Tarski will partition the ball into pieces for which the volume, like 0÷0, is undefined! But that will be the least of our worries on the Banach-Tarski roller-coaster ride.
Well, this concludes episode 1. I've probably already made mistakes, and certainly sown confusion. Submit your comments and questions. I hope to compose episode 2 tomorrow.