Sarah Sarah
Posted on Wednesday, 25 February, 2004 - 12:41 pm:

Can you hang a picture with 1 piece of string and 2 pegs so that if you remove any one peg, the picture will fall?

(I can think of a daft answer but I think it misses the point)
roko mijic
Posted on Wednesday, 25 February, 2004 - 03:23 pm:

looks like it might involve twisting the rope in some way ... presumably the rope is already attached to the picture in 2 places?
Vicky Neale
Posted on Wednesday, 25 February, 2004 - 04:12 pm:

How about something like this? I'm not quite sure whether I've understood the problem correctly. I think the picture will get lower down the wall if either peg is removed, but not fall off completely in one case.

Picture hanging on wall

(Please excuse the quality of the graphic - the pegs are supposed to be vertically above each other)

Vicky
Nicola Coles
Posted on Wednesday, 25 February, 2004 - 04:26 pm:

Sarah, what do you mean by 'fall'; fall lower or fall to the floor?
James Cranch
Posted on Wednesday, 25 February, 2004 - 04:34 pm:

I don't mean to put words in Sarah's mouth, but if this is the same problem as I've seen before, then it means "fall to the floor" (and the problem's a fine poser).
David Loeffler
Posted on Wednesday, 25 February, 2004 - 04:38 pm:

The best I can do is
foo.gif
(sorry, this is a truly dreadful graphic).

In this case, if you remove the top peg the picture will fall off; if you remove the bottom one, the picture just falls down slightly (as the string pulls tight around the top peg)

David
James
Posted on Wednesday, 25 February, 2004 - 04:39 pm:

I think the answer is no because the rope must be looped over at least one of the pegs, so removing one peg will leave the other which has the rope looped over it.

Unless this sort of thing is allowed.
Put the pegs in half way, so the picture can only be supported if both pegs are in

James
Nicola Coles
Posted on Wednesday, 25 February, 2004 - 04:53 pm:

Maybe this is allowed...
picture
...but maybe not
Dan Goodman
Posted on Wednesday, 25 February, 2004 - 05:44 pm:

I think I have a solution (it's a great puzzle by the way).

Unfortunately, the website where Dan put his solution no longer exists - but if you keep reading, you'll find a solution here.
Ian Short
Posted on Wednesday, 25 February, 2004 - 06:26 pm:

There seems to be a link with winding numbers, for anyone who knows of them. Perhaps a reasonable way to approach it would be to look for curves, strings with 0 winding number about both pegs.

Fantastic picture Dan!
Dan Goodman
Posted on Wednesday, 25 February, 2004 - 07:52 pm:

Ian, that's exactly right it's just winding numbers I think. And thanks everyone for your compliments on my pictures! For an encore, here are two solutions my housemates came up with:

Cheating solutions
Dan Goodman
Posted on Wednesday, 25 February, 2004 - 07:53 pm:

The left hand one is supposed to be a thick knot between two close together pegs.
Olof Sisask
Posted on Wednesday, 25 February, 2004 - 11:06 pm:

Here's another solution:

[Woops, I should mention that if you don't want to see a solution, look away now!]

It's gonna fall!

Obtained by the winding number principle.

Regards,
Olof
Dan Goodman
Posted on Wednesday, 25 February, 2004 - 11:29 pm:

Excellent solution Olof, very symmetrical.
Dan Goodman
Posted on Thursday, 26 February, 2004 - 12:58 am:

An update for advanced readers. Stop reading this post now if you haven't done a university level topology course of some sort.

Write fi:R2-{x1,x2}® R2-{xi} for the inclusion of the plane minus two points into the plane minus a single point. These induce maps fi*:p1(R2-{x1,x2})®p1(R2-{xi}) on the fundamental groups of the twospaces. The original questionis asking for anon-zero element a in \Ker(f1*, f2*). However, since R2-{x1,x2} is homotopically equivalent to a one point union of two circles it has fundamental group isomorphic to the free group on two generators, say where A corresponds to a loop around x1 and B to a loop around x2. So, for example, Olof's picture above is AB-1A-1B and is therefore one of the four simplest possible examples (another is mine which is A-1BAB-1). Now, writing p1(R2-{x1})=áAñ = Z and p1(R2-{x2}=áBñ = Z we get f1*(An1Bm1¼ AnrBmr=n1+¼+nr and similarly f2*(An1Bm1¼ AnrBmr)=m1+¼+mr. So the kernel of (f1*, f2*) is just the elements where the sum of the powers of the A components is 0 and the sum of the powers of the B components is 0. This allows us to construct examples of arbitrary complexity. For example, A3B2A-2B-1A-1B-1 (see the picture below, I hope I drew it correctly). Well, I hope that was interesting.

Enormously complicated example

Ian Short
Posted on Thursday, 26 February, 2004 - 11:37 am:

For people who don't know about winding numbers, here's a simple explanation.

# Imagine an ant crawls along the string, all the way round until it reaches its original starting position (the string is a closed loop that connects up behind the picture).

# Imagine a locust sitting on one of the pegs facing the ant. The locust stays on the peg and rotates as the ant moves along the string because it always stays facing the ant.

# For example, if the string was a circle and the locust sat in the middle then it would rotate 360 degrees, all the way round, as the ant encircled it once. In that case we say that the winding number of the string about the locust is 1. One rotation. If the ant did two laps, winding number would be 2. If it did one lap in opposite direction, winding number would be -1.

# Now look at Dan's diagram above this message. Imagine a locust on the right hand peg and an ant crawling round the string. As the ant crawls, locust rotates this way and that, but its total number of rotations, it winding number, is 0.

# I presume this is how Dan conceived the above complicated solution (?). Well not necessarily with ant and locust, it works with any insects.

Ian
Olof Sisask
Posted on Thursday, 26 February, 2004 - 05:55 pm:

Great stuff, Dan. Really nice.

Olof
Sarah Sarah
Posted on Friday, 27 February, 2004 - 01:00 pm:

I understand that for every time the string goes clockwise round a peg, it has to go back round anticlockwise, but not what (if any) the constraints are on which loops have to go in front of each other. I think I can see how to construct an example (by starting with a simple one, then adding loops at either end which cancel each other out) but not how to see if, for example, this dugong would work.

Sarah

PS it is a DUGONG NOT A MANATEE
Sarah Sarah
Posted on Friday, 27 February, 2004 - 01:02 pm:

(the black dots are the pegs, the red dots are on anticlockwise loops, and the blue dots are on clockwise loops)
Dan Goodman
Posted on Friday, 27 February, 2004 - 01:57 pm:

There are no constraints, as long as the number of red and blue dots on the left are the same and the number of red and blue on the right are the same, it will work. Here's a nice way of seeing why. First of all, pull your string tight so that it's tightly wound around the two pegs. Now, you should have a vertical line from the picture to the left hand peg and then from the left peg to the right peg and then down to the picture again. Now, put your finger halfway between the two pegs and pull down to the top of the picture frame. You should now have a sort of M shape in string. Now put both fingers at the middle point of this M and move them in opposite directions so that you get the string doing the following: it goes up from the frame to the left peg, down to where it started from, across the frame to directly underneath the right peg, up to the right peg and straight down again. Now your string should be divided into (a) segments which go up, round the left peg either clockwise or anticlockwise, and then down again, (b) the same except for the right peg, (c) segments going from the point directly beneath the left peg to the point beneath the right peg, or the other way round. When you remove the left peg, for example, it is as if you just cut away all the segments in (a), so you're just left with the bits in (b) and (c), in other words basically just the bits going round the right hand peg. Since you've made sure that the number of times you've gone round the right peg anticlockwise is the same as the number of times clockwise, this remaining bit will unravel. And the same thing works if you remove the right hand peg.