mersenneforum.org Coin Problem.
 Register FAQ Search Today's Posts Mark Forums Read

2005-12-11, 23:39   #34
drew

Jun 2005

38210 Posts

Quote:
 Originally Posted by nibble4bits Another shape I've considered is interesting... A sphere with three slices taken out of it. If you had tetrahedron circumscribed by the sphere and cut segments out of the sphere in parallel with 3 of the 4 sides of the tetrahedron, then you could argue that it has the effects of a three-sided coin. LOL I'd love to actually see some coins made using these designs just to see how well they work in practice. It would be interesting to have people come up with odd types of dice. BTW Odd numbers are naturally difficult in 3D using a 2D surface/polyhedron/combination. Think about why this is so... and the effects for other numbers of dimensions.
Your tetrahedral shape would be more effective if you had more curvature in the remaining spherical part...sort of if it was a hemisphere. It will be less stable, and tend to topple onto one of the flat surfaces.

As far as arbitrary numbers go, you can have a regular n-gon prism or pyramid, and create any n-sided die. You just need to make it unstable to land on the remaining ends.

Drew

 2005-12-13, 16:32 #35 nibble4bits     Nov 2005 181 Posts I kind of figured to make the "fourth side" curved in a way to make it unlikely to be settled on. An added consequence is that it will not roll as far. There's also the issue of weight and balance but it's possible to tweak one side to make it more balanced. LOL nice tweak of my design. It would be obvious if I had actually used one with a simple prism shape. Adding the curves would be stupid not to add (unless patented). Last fiddled with by nibble4bits on 2005-12-13 at 16:36
 2005-12-13, 17:10 #36 Wacky     Jun 2003 The Texas Hill Country 100001110102 Posts I'm not certain if you are describing the same thing, but here is a design that will work to make any n-sided "dice". Take a regular n-agon (for 3, it is an equilateral triangle) Extrude it along the axis perpendicular to the face for a distance in excess of twice (it may not need to go that far, but the concept will work at that length) the diameter of the circle circumscribing the polygon. Add end caps to this prism by adding an n-sided pyramid to each end. The pyramid needs to be tall enough to assure that the CG of the object would never lie above an end facet if it were to attempt to rest on said facet. Since the ends are designed to be unstable, the object will always come to rest on one of the major faces. By symmetry, they are each equally likely. One problem with using any kind of dice that has an odd number of faces is that it will be difficult to read. The only clearly designated face is the one on the bottom. Its value is obscured by the object itself. With the regular solids such as the cube (and higher), the designated face is the one "on top", opposite the face upon which the object is resting. This provides a ready surface upon which the designation can be easily marked. Last fiddled with by Wacky on 2005-12-13 at 17:12
 2005-12-13, 17:26 #37 Zeta-Flux     May 2003 2×761 Posts Wacky, While it is hard to read the "bottom" face of a die, what I've seen done (with a tetrahedron) is write three numbers on each face, one on each edge, in such a way that when the die lands, the number along the bottom edge on each face next to the floor is the same number, and this is the number rolled.
 2005-12-13, 17:48 #38 Wacky     Jun 2003 The Texas Hill Country 2·541 Posts Zeta, yes, for the tetrahedron, that is a reasonable approach. However, it is significant that the angle between the faces is acute. When you go to more faces than 6, the faces adjacent to the floor are also point "down" and therefore, they, too are going to be difficult to read. If you attempt to use any other face, I fear that it would be difficult to determine which ones to "read" because they are removed from any distinct reference plane. The same can be said for determining which face is the "top" one when the number of faces grows beyond a very small number.
2007-01-10, 13:36   #39
davieddy

"Lucan"
Dec 2006
England

11001001100112 Posts

Quote:
 Originally Posted by drew This is an invalid approach. As you revolve this circle around the coin's axis to create the 3-d sphere, you'll see that the 'edge' region sweeps out a much larger area per projected length than the face, due to it's larger displacement from the axis of revolution. Therefore your result gives more weight to the faces than it should. mfgoode made the same mistake of trying to turn it into a 2-d problem without making the appropriate adjustment. The assumption about circumscribing a sphere is reasonable, though, for the static case. Let's see if someone can corroborate a value of sqrt(2)/4 for the thickness/diameter ratio. Drew
Coming late to this thread, I gather that the argument was
about whether the angles subtended by the three faces of the
cylinder shoud involve the arc length or the area.

Drew plumps for area and considers a sphere radius R into
which the cylinder fits. For the areas to be equal, the height
of the cylinder h = 2R/3

If r is the radius of the cylinder then
r2 + h2/4 = R2 = 9h2/4

so h/r = sqr(2)/4 as Drew says.

However, if the "coin" is flipped as usual, it spins about a
horizontal diameter. Freezing this motion at a random instant
suggests the 2D arc length approach giving h=R=2r/sqr(3)

David

Last fiddled with by davieddy on 2007-01-10 at 14:11

2007-01-10, 15:30   #40
fatphil

May 2003

E716 Posts

Quote:
 Originally Posted by Wacky If we circumscribe a sphere about the cylinder, then this means that 1/3 of the total spherical surface must be the projection of one of the faces.
Following this principle, experiments have been made. For example, one so proportioned yielded 32.6% edges in a sample of 3000 tosses, which is in fairly good agreement with the model.

2007-01-10, 16:46   #41
davieddy

"Lucan"
Dec 2006
England

6,451 Posts

Quote:
 Originally Posted by fatphil Following this principle, experiments have been made. For example, one so proportioned yielded 32.6% edges in a sample of 3000 tosses, which is in fairly good agreement with the model.
I find this surprising. Drew (in post #20) points out that the C of G
is lower when the coin lies on a face, which should favour this in
a dynamic situation.
As I said, it also depends on how the coin is spun.
Finally, think about my "stick problem".
The "good agreement" of experiment with naive theory is,
I suggest, fortuitous.

David

2007-01-10, 17:56   #42
xilman
Bamboozled!

"πΊππ·π·π­"
May 2003
Down not across

27DB16 Posts

Quote:
 Originally Posted by davieddy I find this surprising. Drew (in post #20) points out that the C of G is lower when the coin lies on a face, which should favour this in a dynamic situation. As I said, it also depends on how the coin is spun. Finally, think about my "stick problem". The "good agreement" of experiment with naive theory is, I suggest, fortuitous. David
As I said in post #12, this probability question has been known for a very long time as being insoluble without making a number of assumptions explicit. There are many assumptions as to what constitutes "tossing" the coin which agree with common notions of "tossing" and there are many different geometries which meet the constraints so imposed.

Going over to a static model doesn't help much, as I pointed out in a subsequent post.

Paul

2007-01-11, 07:47   #43
davieddy

"Lucan"
Dec 2006
England

11001001100112 Posts

Quote:
 Originally Posted by xilman Going over to a static model doesn't help much, as I pointed out in a subsequent post. Paul
But it at least sets a soluble well defined problem,
apparently tricky enough to confuse certain frequent posters
(who shall remain nameless :)

A sphere with differently coloured areas is a good start
for the design of a general dice.

David

Last fiddled with by davieddy on 2007-01-11 at 07:51

 2007-01-12, 08:55 #44 davieddy     "Lucan" Dec 2006 England 144638 Posts Static Model Everyone is in agreement that the question of what happens to a spinning coin AFTER it hits the surface is too complex for informal discussion. The "static model" consists of choosing an orientation of the coin "at random" and deeming that the coin lands on the surface vertically below the centre of gravity. Choosing a random orientation in 3D is like choosing a random location on earth. If we choose latitude and longitude at random, our choice is biased towards locations near the poles. We should instead choose longitude and the distance from the equatorial plane. However, in the context of a coin spun about a horizontal axis, it is not unreasonable to take the angle at which it strikes the ground as the orientation chosen for the static model. There is a marked tendency for the coin to strike the ground with the leading rim rather than the trailing one, which IMO favours ending on the edge rather than a face in a dynamic scenario. Perhaps this explains the experimental coincidence I noted in an earlier post. David Last fiddled with by davieddy on 2007-01-12 at 08:59

 Similar Threads Thread Thread Starter Forum Replies Last Post petrw1 Puzzles 1 2015-02-06 23:17 Dougal Homework Help 10 2010-12-07 19:18 davieddy Puzzles 62 2010-05-29 01:14 henryzz Puzzles 10 2010-05-14 11:21 eaS Puzzles 20 2003-09-21 02:48

All times are UTC. The time now is 06:46.

Tue Dec 1 06:46:16 UTC 2020 up 82 days, 3:57, 1 user, load averages: 2.58, 2.23, 1.89