Wikipedia:Reference desk/Archives/Mathematics/2011 September 8

Welcome to the Wikipedia Mathematics Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

September 8

Did I use neighborhood right?

this has been bugging me. I am a monitor (kinda like a high-school version of an AP, I am an upperclassman and do this in an underclassman calculus class during my apparently mandatory study hall). I got to teach today :) since the teacher is out sick and the substitute was not a math teacher. There was a question based on a problem I presented from the regular teacher's notes that if there were a function defined say, over (-∞,0] and (3,∞), what would be the limit at 2 (a modification of the problem in the notes, which had it defined over R but with a jump at x=2). I said it would not exist because f(x) is not defined within a neighborhood of x=2. Did I use the term right? (It's been a while since I took calc and I went the discrete route in my math studies instead, so I'm not sure; I mused as much out loud but if anyone in the class thought I used it wrong they were afraid to correct me because they're all freshies and I'm the scary senior ;)) thx — Preceding unsigned comment added by 24.92.85.35 (talk) 00:06, 8 September 2011 (UTC)

Yes, that's correct.--Antendren (talk) 10:58, 8 September 2011 (UTC)

Maximising winnings on Euromillions

Back when I was in a syndicate for the National Lottery (the game they named Lotto), we used to buy tickets in a way that maximised the winnings. So if we chose the numbers 1-7 we'd buy 123456, 123457, 123467, 123567, 124567, 134567 & 234567. (We actually did this for 8 numbers, because we were a large syndicate) This would mean that if we got 3 numbers, we would win 4 times, and if we got 4 numbers we'd win on every line and so on. Basically, we didn't really expect to win the jackpot, but wanted to increase the actual winnings we made.

I was thinking of setting up another syndicate, but for the Euromillions. I can see how we'd do a similar method if we held the lucky stars still, but is there a better way to do it? At the moment, I'm seeing we could buy 6 tickets if we want 6 numbers with 2 fixed lucky stars, or 18 tickets if we want 6 numbers with 3 lucky stars. I'm sure I could sit down and work out the levels of probability, but I thought I'd see if some illustrious reference deskers had any thoughts. Worm^TT · (talk) 08:34, 8 September 2011 (UTC)

For information, Lotto takes 6 random numbers between 1-49, with prizes if you get 3 or more matching. Euromillions has takes 5 random numbers from 1-50 and 2 lucky stars from 1-11, with prizes for 2 main numbers upwards. Worm^TT · (talk) 08:38, 8 September 2011 (UTC)

(-1/3)^(-1/3)

So google gives this answer as 0.721124785 - 1.24902477 i. But -(3^(1/3)) is also a solution? Why do they give the complex solution? --163.202.48.109 (talk) 11:10, 8 September 2011 (UTC)

Since it was a negative number that you were getting the root of they treated it as a complex number and gave the principal value of the cube root. Getting a real cube root of a negative real number only works for rational exponents and is a bit of a mess, even if they did do it they would have to recognize 1/3 as a rational number rather than just some random real number. Dmcq (talk) 13:01, 8 September 2011 (UTC)

What he said. Google doesn't "know" that 1/3 is rational, and so doesn't extract the real root. Here is how to get Google to report the real root: [1]. Sławomir Biały (talk) 13:16, 8 September 2011 (UTC)

I'd also recommend that you read the root of unity article and, depending on your level. From that you'll see that

$(-1/3)^{-1/3}=-{\sqrt[{3}]{3}},{\sqrt[{3}]{3}}(1/2-i{\sqrt {3}}/2),\ {\text{or}}\ {\sqrt[{3}]{3}}(1/2+i{\sqrt {3}}/2).$

Maybe you already knew how to do that, but I thought I'd mention it anyway. — Fly by Night (talk) 15:10, 8 September 2011 (UTC)

Toric Configuration Space

Say I've got a robotic arm that operates in a two dimensional plane. The has a "shoulder" and an "elbow", both joints can rotate about 360°. I'm interested in the final position of the "hand". The configuration space is a torus. But clearly, different shoulder and elbow configurations can lead to the hand being in the same place. Taking the quotient space T/~, where p ~ q if and only if the hand position for joint configuration p is the same as that of q. Topologically, the space of final positions of the hand is an annulus, but what's the topology of T/~? There's obviously an injective mapping from T/~ to the annulus; but I'm not sure about continuity. — Fly by Night (talk) 17:55, 8 September 2011 (UTC)

I'm not really an expert on this but the closed map lemma (see Open and closed maps) implies that the map T→A is closed and our article on quotient maps says this implies that the map is a quotient map (see the last paragraph in the Properties section). That proves that T/~ is homeomorphic to the annulus unless there's something I'm missing.--RDBury (talk) 21:08, 8 September 2011 (UTC)

If I understand the configuration of your arm correctly, then you have two segments A and B, with A attached to a fixed pivot at the "shoulder" and B attached to A at the "elbow", and each joint free to rotate independently through 360 degrees in the same plane. So if A and B have lengths a and b respectively, and we assume a > b, then the "hand" at the far end of arm B can reach anywhere in the annulus a-b ≤ r ≤ a+b.

Let's call the shoulder O, the elbow joint P and the hand Q. Then the two parameters of your configuration space are the angle that the arm A makes with some fixed direction - call this α - and the angle that arm B makes with arm A - call this β. So OPQ is a triangle with |OP| = a, |PQ| = b, angle OPQ = 180 - β, and you want to find the polar co-ordinates (r, θ) of Q in terms of a, b, α and β.

The cosine rule tells us that

$r=|OQ|={\sqrt {|OP|^{2}+|PQ|^{2}-2|OP||PQ|\cos(\angle OPQ)}}={\sqrt {a^{2}+b^{2}+2ab\cos \beta }}$

and the sine rule tells us that

$\sin(\angle POQ)={\frac {|PQ|\sin(\angle OPQ)}{|OQ|}}={\frac {b\sin \beta }{r}}$

so we have

$\theta =\alpha +\angle POQ=\alpha +\sin ^{-1}({\frac {b\sin \beta }{r}})$

Now, if we are given r and θ and want to find α and β, then we have

$\cos \beta ={\frac {r^{2}-a^{2}-b^{2}}{2ab}}$

which gives two possible values for β (except when β is 0 or 180 degrees). Once we fix β then we have

$\alpha =\theta -\sin ^{-1}({\frac {b\sin \beta }{r}})$

So the projection from configuration space to the annulus is, in general, 2-to-1, except when β is 0 or 180 degrees, when it is 1-to-1. Gandalf61 (talk) 11:33, 10 September 2011 (UTC)

Topologically the map T → A is just the obvious flattening. If you have the torus defined in cylindrical coordinates (r, θ, z) = (a+bcosβ, α, bsinβ) for shoulder and elbow angles α and β, the map onto the annulus isn't a straight projection, but it's close. The radial distance is preserved. The points in the torus that map down to a spoke in the annulus form a sort of slanted circle in the torus, since height in the torus translates into radial movement in the annulus. Think of squashing a slinky that's lying on its side. Rckrone (talk) 05:09, 11 September 2011 (UTC)

Just an observation: Suppose the elbow is a socket rather than a gear (so it permits motion through a sector of a sphere). Then I think motion planning becomes a more interesting question. Sławomir Biały (talk) 22:31, 11 September 2011 (UTC)