Wikipedia:Reference desk/Mathematics

Welcome to the mathematics section
of the Wikipedia reference desk.

skip to bottom

Select a section:

• Computing
• Entertainment
• Humanities
• Language
• Mathematics
• Science
• Miscellaneous
• Archives

Want a faster answer?

Main page: Help searching Wikipedia

   

How can I get my question answered?

• Select the section of the desk that best fits the general topic of your question (see the navigation column to the right).
• Post your question to only one section, providing a short header that gives the topic of your question.
• Type '~~~~' (that is, four tilde characters) at the end – this signs and dates your contribution so we know who wrote what and when.
• Don't post personal contact information – it will be removed. Any answers will be provided here.
• Please be as specific as possible, and include all relevant context – the usefulness of answers may depend on the context.
• Note:
  • We don't answer (and may remove) questions that require medical diagnosis or legal advice.
  • We don't answer requests for opinions, predictions or debate.
  • We don't do your homework for you, though we'll help you past the stuck point.
  • We don't conduct original research or provide a free source of ideas, but we'll help you find information you need.

Ready? Ask a new question!

How do I answer a question?

Main page: Wikipedia:Reference desk/Guidelines

• The best answers address the question directly, and back up facts with wikilinks and links to sources. Do not edit others' comments and do not give any medical or legal advice.

See also:

• Teahouse
• Help desk
• Village pump
• Help manual

October 13

Vectors and matrix algebra dot products et al

I have the X, Y, Z co-ordinates of 3 points in space for which I need to determine the following

1. The co-ordinate of a circles centre which passes through all 3 points

2. The co-ordinates of the point from 1. above, translated normal to the surface described by the original 3 points by a distance X.

It has been a very long time since I've used any vector or matrix so I'm a little muddled.

Any help would be great.

Solve first the special case where Z=0 for all three points. The general case is translated into the special case by changing coordinates. Bo Jacoby (talk) 07:03, 13 October 2009 (UTC).

... I don't think the OP was wanting the projection onto the XY plane? You have three vectors: OX, OY and OZ. The centre of the circle lies at the intersection of the perpendicular bisectors of the sides of the triangle XYZ, so you need to find vectors perpendicular to two sides (say XY and XZ) and passing through the middle of these sides, then find where these lines intersect. This is rather messy, so someone else will probably come up with a simpler method. For part 2, the normal to the plane of the circle is given by the vector product of any two vectors in the plane of the circle, such as XY and XZ (subtract co-ords to get these), find the vector product, divide by it's modulus, multiply by distance x, then add the result to answer to part 1. Dbfirs 07:35, 13 October 2009 (UTC)

... (later) ... If you are happy with a "magic formula" for the centre of the circle in part 1, it is given by the vector

$\alpha \,P_{1}+\beta \,P_{2}+\gamma \,P_{3}$

where

$\mathrm {P_{1}} ={\begin{bmatrix}x_{1}\\y_{1}\\z_{1}\end{bmatrix}},\mathrm {P_{2}} ={\begin{bmatrix}x_{2}\\y_{2}\\z_{2}\end{bmatrix}},\mathrm {P_{3}} ={\begin{bmatrix}x_{3}\\y_{3}\\z_{3}\end{bmatrix}}$

$\alpha ={\frac {\left|P_{2}-P_{3}\right|^{2}\left(P_{1}-P_{2}\right)\cdot \left(P_{1}-P_{3}\right)}{2\left|\left(P_{1}-P_{2}\right)\times \left(P_{2}-P_{3}\right)\right|^{2}}}$

$\beta ={\frac {\left|P_{1}-P_{3}\right|^{2}\left(P_{2}-P_{1}\right)\cdot \left(P_{2}-P_{3}\right)}{2\left|\left(P_{1}-P_{2}\right)\times \left(P_{2}-P_{3}\right)\right|^{2}}}$

$\gamma ={\frac {\left|P_{1}-P_{2}\right|^{2}\left(P_{3}-P_{1}\right)\cdot \left(P_{3}-P_{2}\right)}{2\left|\left(P_{1}-P_{2}\right)\times \left(P_{2}-P_{3}\right)\right|^{2}}}$

... but I recommend that you work through the long method to help your understanding of vector equations of lines. The vector equation of each perpendicular bisector is given by

b = OM + kp where O is the origin, M is the mid-point of a side, p is any vector perpendicular to the side, and k is an arbitrary constant (to be determined when you find where the lines intersect). Dbfirs 07:52, 13 October 2009 (UTC)

(edit conflict). I was suggesting to solve a special case, not a projection. The center (X,Y) of the circle has the same distance to the three points (X_A,Y_A), (X_B,Y_B), and (X_C,Y_C), so (X−X_A)² + (Y−Y_A)² = (X−X_B)² + (Y−Y_B)² = (X−X_C)² + (Y−Y_C)². Solving these two equations gives expressions for the two unknowns X and Y. First do the squares: X² − 2XX_A + X_A² + Y² − 2YY_A + Y_A² = X² − 2XX_B + X_B² + Y² − 2YY_B + Y_B² = X² − 2XX_C + X_C² + Y² − 2YY_C + Y_C². Then subtract X² + Y² giving − 2XX_A + X_A² − 2YY_A + Y_A² = − 2XX_B + X_B² − 2YY_B + Y_B² = − 2XX_C + X_C² − 2YY_C + Y_C². Now the equations are linear and the solution is straightforward. Bo Jacoby (talk) 08:39, 13 October 2009 (UTC).

Yes, I see. I suppose it is possible to extend this 2-D solution to 3-D, but the OP's heading suggested that they wanted a vector solution. Dbfirs 10:32, 13 October 2009 (UTC)

The center of the circle is a symmetric function of the 3 points. Your formula does not look symmetric. Can that be helped? Bo Jacoby (talk) 16:10, 13 October 2009 (UTC).

${(P_{1}-P_{2})\times (P_{2}-P_{3})=(P_{3}-P_{1})\times (P_{1}-P_{2})=(P_{2}-P_{3})\times (P_{3}-P_{1})=P_{1}\times P_{2}+P_{2}\times P_{3}+P_{3}\times P_{1}}$. — Emil J. 16:20, 13 October 2009 (UTC)

Thanks. I did make one error above, I should have said p is any vector perpendicular to the side in the plane of the triangle. I'm sure there should be a neater vector method for part 1, but I can't see it. My method is good for understanding vectors, but seems too long. There is a co-ordinate geometry formula for 3-D like Bo Jacoby's formula. I suppose I could derive it using my vector method if I had the patience, but it is a good exercise for the OP if he or she wants to understand vectors. Are there any experts who can see a short-cut? Dbfirs 16:44, 13 October 2009 (UTC)

An Almost-Primality Test

Is there any general way known to determine whether a number that fails a test for being prime is a product of only two primes without actually checking up to the number's third root for a factor? Is it just as efficient--so far as is known--to find a factor as to determine that it has at least three knowing that it has at least two? There are fast algorithms for determining whether a number is prime or not without actually finding a proper divisor. Once this is done with an affirmative (composite) result, I am left to find a factor in what may be a very long time when all I really want is the knowledge that there are or are not only two factors.Julzes (talk) 10:00, 13 October 2009 (UTC) For example, right now my computer is checking the factorization of 111110101011110010111110000010001₁₅₆₃ and it looks like it will take a week.Julzes (talk) 10:06, 13 October 2009 (UTC) I have a further question on this. I'm curious to know what the first two consecutive bases in which the above number is prime, so if anybody wants to run a search on that I would appreciate it. If it's likely that no such pair exists, I would like a good explanation also. My inclination is toward believing that such a pair probably does exist, but I haven't actually reasoned it out quantitatively.Julzes (talk) 10:10, 13 October 2009 (UTC) It is prime in both base 591 and 593, but aside from its being prime in base 2 and prime if we also allow it to be considered as in base 1, that's the closest I've come to what I'm looking for.Julzes (talk) 10:17, 13 October 2009 (UTC)

Would I be right to say that your number 111110101011110010111110000010001₁₅₆₃ is given by

1,610,516,746,593,217,788,093,321,222,359,707,265,648,849,509,955,413,660,168,610,383,205,686,470,714,500,112,576,660,497,611,201,902,703₁₀?

And if so, where did you get your desire to factorise such an exotic number? ~~ Dr Dec (Talk) ~~ 17:30, 13 October 2009 (UTC)

Without checking all 103 digits, it looks the same. I'm actually not so interested in the factorization as in knowing it has two prime factors and not three. I'm just following through on some research in great detail whose impetus will seem rather strange. It comes from reading the h2g2 Entry number of the article on the constellation Reticulum as a base-thirteen number. Got some rather nice arithmetical stuff from doubling the base conversion process. In this base-two case, the most interesting thing was the string of almost primes running from bases 29 through 33, along with the fact that before base 29 there are only the primes for the cases of 'base 1', base 2, base 7, and an almost-prime at base 16 (Every other smaller base has at least three prime factors). I've made an extensive set of listings for the other bases up to base 96, and I'm right now just in the process of extending the list for base 2.Julzes (talk) 23:06, 13 October 2009 (UTC)

The Dickman-de Bruijn function gives an indication of how likely a random number x is to have no prime factors above x^u x^1/u for given u. For u=3 it's about 5%. For a 103-digit number with 3 factors you could probably crank out a complete factorization with the elliptic curve method in a practical amount of time. 75.62.0.233 (talk) 23:41, 13 October 2009 (UTC)

What you just said doesn't quite make sense. I have no problem with the last sentence ('practical' is an irrelevant concept here anyway), but x obviously cannot have a factor greater than x³.Julzes (talk) 00:22, 14 October 2009 (UTC)

Sorry, misstated, see correction. Also this article. 01:16, 14 October 2009 (UTC)

Well, the number almost certainly has only two factors, but I'm still waiting.Julzes (talk) 09:16, 14 October 2009 (UTC) Still waiting. And now I've got my comp also simultaneously running for bases 1611 and 1813. It looks like I'm going to have at least two to add to the alpertron.com records list. There is something wrong with my Adobe Acrobat, so I haven't been able to open that article. I appreciate it, though. I guess it's unlikely anyone will come up with an almost-primality test any time soon, but perhaps there is a way to quicken things. I could probably declare the base-1563 case done, as it has no factor smaller than 35 digits (the borderline I need to consider), but I may as well wait it out and see just how large its smallest factor is.Julzes (talk) 04:31, 16 October 2009 (UTC)

Vectors in meterology

How are vectors used in meterology?Accdude92 (talk) (sign) 13:36, 13 October 2009 (UTC)

Wind velocity would be a vector (well, a vector field, strictly speaking - a vector for each point in the atmosphere). There will be other uses too, but that is the most obvious one. --Tango (talk) 16:40, 13 October 2009 (UTC)

• Any scalar field (e.g. temperature, air pressure, moisture, …) gives rise to a vector field; namely the gradient vector field. The gradient vector field has many useful physical interpetations. ~~ Dr Dec (Talk) ~~ 17:40, 13 October 2009 (UTC)

percent comparisons

Suppose I start with 100. Then 120 is 20% greater, 150 is 50%, 200 is 100%, and 300 is 200% greater. But 300 is also 3 times 100. So 200% greater is also 3 times greater. Is this seeming discrepancy merely a language artifact, or is there something deeper here? --Halcatalyst (talk) 19:13, 13 October 2009 (UTC)

"200% greater" is "3 times as great" and logically it would also be "2 times greater", but language isn't always logical and often uses "times greater" and "times as great" as synonyms (but only when no percent signs are involved); see this Q&A or this one, for instance. Certainly just a language quirk. —JAO • T • C 19:24, 13 October 2009 (UTC)

"200% of" is "2 times", "200% greater" is "3 times". I think "X times greater" is ambiguous and should be avoided. --Tango (talk) 19:32, 13 October 2009 (UTC)

Agreed, though it seems to be pretty much a lost cause. Even worse (to my ear) are such usages as "three times smaller" (meaning one-third as big, probably). AndrewWTaylor (talk) 09:54, 14 October 2009 (UTC)

I recall hearing "100% smaller" once, I have no idea what that was meant to mean (from context it didn't mean "reduced to zero", I think it was probably intended to mean "half", but why the speaker ever thought it did, I have no idea...). --Tango (talk) 10:23, 14 October 2009 (UTC)

Some probability questions...

1. I have a bag with 6 balls numbered one to six inside. What is the probability that I will pull the balls out in numerical order?

2. What is the probability if I do that again but this time I label the first ball as 1st (so ball marked "1" is correct) but placed sixth (so ball marked "6" is also correct); the second ball as 2nd (so ball marked "2" is correct) and placed fifth (so ball "5" is also correct)? And so on (3rd ball out must be either no.3 or number 4; 4th ball out must be either ball no.4 or no.3... etc to the end)

3. What is the probability for exercises 1 and 2 if I increase the number of balls to 7?

In case I am accused of asking a homework question, I'll tell you why I ask. I watch poker on TV and there is usually either 6 or 7 players at the table. Just for fun I try to predict the order in which people will be knocked out of the game. I used to do this by just writing down their placing. But lately I have been trying to "double" my chances (though I'm sure that isn't the truth in practice) by allowing myself to be correct using the two labels rather than just the one. I sense I may not have asked this question entirely clearly so please ask if you need me to clarify. --79.72.61.252 (talk) 23:38, 13 October 2009 (UTC)

For question 1, there are 720 permutations of the 6 balls and only one of them will satisfy the requirement, so the probability of getting that permutation (based on usual assumptions about drawing randomly) is 1/720. I don't understand question 2. In a given hand of poker, seat position has a real effect in one's winning chances--you can't assume independent draws. 75.62.0.233 (talk) 23:47, 13 October 2009 (UTC)

Thanks for your answer to question 1. That is much lower probability of success than I would have expected.

To clarify question 2, I am bringing out the numbers again but this time they can be in this order:

1st ball can be either ball 1 or 6

2nd ball can be either ball 2 or 5

3rd ball can be either ball 3 or 4

4th ball can be either ball 4 or 3

5th ball can be either ball 5 or 2

6th ball can be either ball 6 or 1

So given that there are now two correct balls at each stage what are my chances of succeeding at bringing them out in an order that is correct? —Preceding unsigned comment added by 79.72.61.252 (talk) 11:56, 14 October 2009 (UTC)

The chance of the 1st ball being right is 2/6=1/3. If that ball is right, the chance of the 2nd being right is 2/5. If that is right the chance for the 3rd is 2/4=1/2. After that only one of the options will be possible since the other will have already been used, so the chances for the 4th, 5th and 6th balls are 1/3, 1/2 and 1. To get the chance of them all being right we multiply those together. That gives us 2/180=1/90. That is quite a lot more likely than the 1/720 we had before (in fact, it is 8 times likely, that is 2³ since the first 3 balls are twice as likely to be right). --Tango (talk) 12:11, 14 October 2009 (UTC)

I don't think this is right. Once you know the value of the first ball, there is only one ball left out of the five remaining that can follow it. So we'd get the same probabilities as before, except for the first ball, which would be twice as likely to be correct. So the probablility of getting one of those sequences is 1/360 (or 2/720 - note there are 720 permutations and only two are correct). Readro (talk) 12:28, 14 October 2009 (UTC)

No, it would only be two permutations if the series had to be either 1,2,3,4,5,6 OR 6,5,4,3,2,1, however this allows for more permutations than that, such as 6,2,4,3,5,1 etc. --79.72.61.252 (talk) 14:08, 14 October 2009 (UTC)

So, therefore, if there were seven balls, the chance of pulling them out in the correct order is 1 in 7! = 5040. The chance of the numbers being correct according to the second system is 1 in 5040/8 = 630. More generally, where N is the number of balls, the odds are 1 in N! and 1 in N!/[2^INT(N/2)]. Warofdreams talk 16:03, 14 October 2009 (UTC)

Thanks everyone. --79.72.61.252 (talk) 11:40, 16 October 2009 (UTC)

October 14

Constructor

A constuctor has a 0.80 probability of making $70000, a probability of loosing $20000 and a probability of breaking even. what is the expected value? —Preceding unsigned comment added by 24.0.239.50 (talk) 02:00, 14 October 2009 (UTC)

You have to know all three probabilities. See expected value#Examples for some examples of how to compute expected value. 66.127.53.204 (talk) 02:38, 14 October 2009 (UTC)

Although you can find a range for the expected value. At the low end, breaking even has a probability of 0, and at the high end, it has a probability of 0.20. Rckrone (talk) 04:52, 14 October 2009 (UTC)

If you assume that the three cases you give are all possibilities you can use the fact that the probabilities would then sum to 1 in order to give the expected value as a function of one of the two probabilities that are not given. Taemyr (talk) 05:05, 14 October 2009 (UTC)

Probably a bit much for this but for a constructor doing one project at a time it can make more sense to use the geometric mean to give the expected nett worth rather than the normal arithmetic one. This is explained in Rate of return Dmcq (talk) 09:19, 14 October 2009 (UTC)

I'm not sure how the geometric mean applies there. I could imagine wanting to find the expected value of a joint probability distribution of the profit/loss across a bunch of different projects, but that's more complicated than the geometric mean. Alternatively, see Kelly criterion for an approach to treating each project as a straight gamble and deciding how much to bet on it given a specified bankroll. 66.127.53.204 (talk) 10:13, 14 October 2009 (UTC)

Yes that was what I was saying, if the constructor does one project at a time that gives a more useful estimate of the value. I don't why you say that and then say you're not sure how the geometric mean applies there. Dmcq (talk) 21:50, 14 October 2009 (UTC)

There is some missing data here. What is the probability of tightening $20000? -- Meni Rosenfeld (talk) 21:05, 14 October 2009 (UTC)

If x, (0≤x≤0.20), is the untold probability of loosing $20000, then the expected value is μ = 0.80·$70000−x·$20000. So $52000≤μ≤$56000. The geometric mean exp(0.80·log(70000)+x·log(−20000)+(0.20−x)·log(0)) makes no sense to me. Bo Jacoby (talk) 08:09, 16 October 2009 (UTC).

The constructor would start off with some amount of money like $60000 and end up with say $130000 or $40000. If the constructor started with $20000 or less then bankruptcy is a possible outcome. Since bankruptcy stops any gain for years and years that should mean the value of a deal where that is a possible outcome is zero, in fact one would count it as a low amount as there s always a way back from it. Dmcq (talk) 01:53, 17 October 2009 (UTC)

Polynomial ideals

Hi. I've got this problem, from an Algebraic Geometry book:

Consider the polynomial $f=x^{4}+x^{2}y^{2}+y^{3}-x^{3}\in \mathbb {C} [x,y]$ and the ideal $I=\left(f,{\frac {\partial f}{\partial x}},{\frac {\partial f}{\partial y}}\right)$. Compute the dimension of $\mathbb {C} [x,y]/I$ as a complex vector space.

So, I've done the problem, and I'm pretty sure the answer is 3, but I think there's something I'm missing. I just used the 3 polynomials to compute a Gröbner basis, and although the process is slightly arduous, the basis turns out to be quite simple: $I=\left(x^{2},y^{2}\right)$. From that, I can see that the only 3 elements of the standard basis of $\mathbb {C} [x,y]$ that aren't in $I$ are 1, x and y.

Here's my question: Is there any simpler way to do this? In particular, is it possible to get any mileage from the fact that the polynomials generating the ideal are one polynomial and its two first order partial derivatives? I don't know why the book author would pose the question that way unless there's some way to use that information. I hope this question makes sense. -GTBacchus^(talk) 03:02, 14 October 2009 (UTC)

Isn't xy also not in the ideal $(x^{2},y^{2})$, making the vector-space dimension 4, or am I getting confused here?

I can't imagine any easier way to tackle this problem. It feels like much the kind of computational problem Groebner bases is meant for. When $f\in \mathbb {C} [x]$ and I see the ideal $(f,{\tfrac {df}{dx}})$, my instinct is to look for multiple roots of f, but I don't see that helping you any here. What strikes me as odd is that this problem seems to have hardly any connection with algebraic geometry. Since the ideal I isn't radical, the space $\mathbb {C} [x,y]/I$ is not isomorphic to the affine coordinate ring (or whatever it's called) of Z(I), as Z(I) is the single point (0, 0), which has affine coordinate ring $\mathbb {C} [x,y]/(x,y)\cong \mathbb {C}$.

Out of curiosity, what book are you using?

(Disclaimer: I am very much a novice at algebraic geometry and could easily be overlooking a deeper connection.) Eric. 131.215.159.109 (talk) 05:04, 14 October 2009 (UTC)

Thanks for catching my silly error. There are 3 quadratic monomials in C[x,y], as you say. (There are also 3 types of people: those who can count, and those who can't. D'oh!)

The book is Introduction to Algebraic Geometry, by Brendan Hassett (ISBN 978-0-521-69141-3). I think I see what you mean about the lack of interesting geometry in the problem. We're in chapter 2, which is just introducing Gröbner bases; varieties and coordinate rings are defined in chapter 3. -GTBacchus^(talk) 05:45, 14 October 2009 (UTC)

Ah, that explains it. Skip ahead to the interesting stuff. My own book (Robin Hartshorne's Algebraic Geometry) has the opposite problem -- it's ridiculously dense, I can scarcely make any headway at all. Otherwise a good book though. Eric. 131.215.159.109 (talk) 05:51, 14 October 2009 (UTC)

moreover he can play shakuhachi... ;-) --pma (talk) 20:35, 16 October 2009 (UTC)

Newton's Method

If I'm trying to use Newton's method to find the zero of a function over an interval, is there any way I can get an upper bound on the number of iterations I'll need to use to get to within a fixed accuracy? Like, if I were to use the bisection method on an interval [a,b], and I wanted to get the zero to within 10^-5, I could just solve |b - a|/2^n <= 10^-5, and that would tell me the maximum number of iterations n I'd need to use; I want something like that for Newton's method. I've looked at the Wikipedia page, but don't see anything like that.Deadlyhair (talk) 04:21, 14 October 2009 (UTC)

It depends on the behavior of the function and its derivatives. You can even concoct examples where it oscillates forever between two values or stays stuck on one value: see Newton's_method#Counterexamples. But for reasonable functions where the first derivative doesn't change sign between guesses, it converges much faster than bisection. There's tons of analysis of its rate of convergence in numerical analysis texts. 66.127.53.204 (talk) 04:48, 14 October 2009 (UTC)

If you really need an upper bound on the number of steps then it's best to just stick with the bisection method. Even in cases where you know Newton's method converges, it can converge arbitrarily slowly.--RDBury (talk) 13:13, 14 October 2009 (UTC)

The usual assumption for the Newton method on an interval is that the function be smooth and convex (or concave) in the interval. In that case one gets a quadratic rate of convergence (so n iterations give an error less than exp(-C2ⁿ)exp(-Cn²). Usually one starts with the bisection method (that has linear convergence, i.e. error less than exp(-Cn) ), till one gets close enough to the zero, so that the conditions for the quadratic convergence of the Newton iteration hold, and then one switch to it. --pma (talk) 19:40, 14 October 2009 (UTC)

I'm fairly certain this is actually $\exp(-C\cdot 2^{n})$. See also my response below. -- Meni Rosenfeld (talk) 20:59, 14 October 2009 (UTC)

Yes sorry you (and 195) are right! --pma (talk) 21:24, 14 October 2009 (UTC)

Just to give you an idea on what the bound may look like - if f is sufficiently smooth and $x_{n}$ is sufficiently close to the root $x^{*}\;\!$, then $\epsilon _{n+1}\approx {\frac {f''(x^{*})}{2f'(x^{*})}}\epsilon _{n}^{2}$, so $\epsilon _{n}\approx \left({\frac {f''(x^{*})}{2f'(x^{*})}}\right)^{2^{n}-1}\epsilon _{0}^{2^{n}}$. Solve for n to get your bound - the important part of the result will be $\log \log \epsilon \;\!$ where $\epsilon$ is your error goal. -- Meni Rosenfeld (talk) 20:59, 14 October 2009 (UTC)

Newton's method has local quadratic convergence. Quadratic means that (roughly) the number of correct decimal places doubles with every iteration and local means that that only holds once you are close enough to the solution. What close enough means is unfortunately not quite as easy to determine. You do *not* need concavity or convexity for Newtons method, however a zero slope at the intersection with the y-axis will harm you (giving only linear convergence). There are many ways to make Newton's method globally convergent, but they all will result in a linear convergence rate away from the solution and a quadratic rate close to it. So an estimate based on the hybrid bisection/Newton suggested above is likely about the best you can expect.195.128.250.90 (talk) 21:08, 14 October 2009 (UTC)

The short answer to the OP's question is 'no'. If he is in the lucky situation where bisection works, (continuity and change of sign over an interval), and he wants an upper bound on the number of iterations needed, then why use Newton's method? Bisection has a better worst-case behavior than Newton's method. Bo Jacoby (talk) 16:13, 15 October 2009 (UTC).

Thank you all for your help.Deadlyhair (talk) 17:23, 16 October 2009 (UTC)

Russell's paradox

I am browsing some articles about paradoxes. Can Arthur Prior's resolution for Liar paradox "This statement is false", that every statement includes an implicit assertion of its own truth, be applied to Russell's and other related paradoxes, such as Barber paradox and Grelling-Nelson paradox?

I think it would be "Every set silently includes itself as a member." or something for Russell's paradox. And if there are, what are weak points of the way of resolution? Like sushi (talk) 12:42, 14 October 2009 (UTC)

Your proposal can't help: Just think about whether Russel's set contains itself as a member. If it does contain - as you propose for every set (hence for Russel's set as well) - then that contradicts Russel's definition for the set, as a set which doesn't contain any set containing itself as a member. HOOTmag (talk) 13:08, 14 October 2009 (UTC)

I think what the article "Russell's paradox" has is "a set containing exactly the sets that are not members of themselves" (Is it the same?). What I think about it is (as in the case of Liar paradox) if every set silently contains it self, "the sets that are not members of themselves" should be rewritten as "the sets that are not members of themselves and are members of themselves", which does not exist. So the set does not contain anything (But as every set silently contains itself, it does contain it self, or if "exactly" means not even itself, it does not exist).

Like sushi (talk) 02:02, 15 October 2009 (UTC)

While such a inclusion would resolve the problem of Russell's paradox, ie. it would directly tell us that the set in question does not exist. It will not resolve the implications of Russells paradox. Since we already knew that Russel's set doesn't exist. The problem is that there isn't any direct way from the definition of a set to determining wether a description of a set corresponds to a set that can exist. While your resolution would resolve Russel's set, it would not resolve all such paradoxial sets, or if it does it would dramatically reduce the expressive power of our languag. This because of Gödel's incompleteness theorems. Taemyr (talk) 04:56, 15 October 2009 (UTC)

Also I think that either your proposed resolution would be unsound, or it would break stuff. Depending on what you mean by the inclusion beeing silent. Because either you do stuff like saying that the empty set contain a member, ie. itself. Or you say that every set is a member of itself is an assertion $\forall$ S.S∈S without actually including the relevant element. Taemyr (talk) 05:04, 15 October 2009 (UTC) could someone fix my tex-code?

I think the first ofGödel's incompleteness theorems may be dealt in simular way.

The part that seems to be the explanation of it is:

"For each consistent formal theory T having the required small amount of number theory, the corresponding Gödel sentence G asserts: “G cannot be proved to be true within the theory T”. If G were provable under the axioms and rules of inference of T, then T would have a theorem, G, which effectively contradicts itself, and thus the theory T would be inconsistent. This means that if the theory T is consistent then G cannot be proved within it. This means that G's claim about its own unprovability is correct; in this sense G is not only unprovable but true. Thus provability-within-the-theory-T is not the same as truth; the theory T is incomplete.

If G is true: G cannot be proved within the theory, and the theory is incomplete. If G is false: then G can be proved within the theory and then the theory is inconsistent, since G is both provable and refutable from T."

If we assume that every sentence silently asserts itself to be provable to be true,

G should be rewritten as “G can be proved to be true and cannot be proved to be true within the theory T”.

Then it says "A and not A", and is false. That means it is provable to be false. And G is false for the italic part in “G can be proved to be true and cannot be proved to be true within the theory T”. (though it is false in "silent part").

Like sushi (talk) 10:06, 15 October 2009 (UTC)

When you say "That means it is provable to be false", you are assuming that T is complete - that T can prove all true statements such as the statement "G is false". But this is the whole point of Gödel's first incompleteness theorem - if T were complete then we can prove that G is true and at the same time we can prove that G is false, therefore T is inconsistent. Conversely, if T is consistent, then T cannot be complete. Gandalf61 (talk) 10:22, 15 October 2009 (UTC)

Not so, we do not need a complete system in order to prove that "Not(A and not A)" is a theorem, even without any information on what A is. The problem is that the statement "G and G is provable in T" is a stronger statement than just "G". So by assuming that ever statment includes the addendum "and this is provable in T" we get in a situation where we don't just have to prove the statement, but we also have to prove the provability of the statement. Which is easy in the meta-language, you just produce the proof. But that is the meta language, which is a different system. Taemyr (talk) 10:36, 15 October 2009 (UTC)

Also note that Gödel the statement that produces is a statement that is equivalent with G. It is not G itself, the common thought at the time was to avoid paradox by having systems that where incapable of referring to itself. One of the things Gödel notes is that this is impossible if the system is capable of stating any thruths about arithmetic. The actual statement G is purely an arithmetic statement. It just happens to be constructed in a way so that G is true if and only if G can not be proven. Taemyr (talk) 10:49, 15 October 2009 (UTC)

@Like Sushi, look: Russel's paradox assumes the following intuitive premise: For every definition, there exists the set containing just all elements satisfying that definition. Having assumed that, Russel found a definition which doesn't let the correspondent set exist! this is Russel's paradox: Assuming the existence of such a set - is a contradiction! HOOTmag (talk) 11:51, 15 October 2009 (UTC)

The empty set does not include itself as a member, silently or otherwise. — Carl (CBM · talk) 11:44, 15 October 2009 (UTC)

I included that as an example of how the statement "Every set includes itself as a memeber" would break stuff. Taemyr (talk) 11:50, 15 October 2009 (UTC)

You're right, I missed it. Actually, this thread is making me think I should look into proposed "resolutions" of Russell's paradox. — Carl (CBM · talk) 12:15, 15 October 2009 (UTC)

This is a different way from what I proposed before. Can the first of Gödel's incompleteness theorems be considered in this way?

Once again, the part of explanation seems to be

"For each consistent formal theory T having the required small amount of number theory, the corresponding Gödel sentence G asserts: “G cannot be proved to be true within the theory T”. If G were provable under the axioms and rules of inference of T, then T would have a theorem, G, which effectively contradicts itself, and thus the theory T would be inconsistent. This means that if the theory T is consistent then G cannot be proved within it. This means that G's claim about its own unprovability is correct; in this sense G is not only unprovable but true. Thus provability-within-the-theory-T is not the same as truth; the theory T is incomplete.

If G is true: G cannot be proved within the theory, and the theory is incomplete. If G is false: then G can be proved within the theory and then the theory is inconsistent, since G is both provable and refutable from T."

Gödel sentence G asserts: “G cannot be proved to be true within the theory T”.

If G were provable (to be true), G would effectively contradict itself. So G cannot be proved (to be true). Then it can be true but unprovable, or is false. G can be "false but is provable to be true".

Like sushi (talk) 08:08, 16 October 2009 (UTC)

First of all be very carefull about what something means when you say that something is false within a formal system. Often the system is defined by what it means for something to be true or false. Gödel's incompleteness theorems can be resolved by allowing something to be true but unprovable. This means that the system is incomplete, which is a bad thing but many feels it is a price we have to pay. It can also be false but provable, which usually means that the system is inconsistent. This can be very bad, for example clasical logic breaks down when faced with an inconsistency. Having one you can prove any statement. However some feels that logical systems that can reason in a good manner even in the presence of inconsistensy is usefull for when we are reasoning from assumptions that might be faulty. See paraconsistent logic for further treatment. however even when we are operating in a paraconsistent logic we do not actually want inconsistency. Taemyr (talk) 08:35, 16 October 2009 (UTC)

The variations of Russell's paradox seems to be in the form

The <V>er that <V>s all (and only those) who don't <V> themselves

The result is if the <V>er <V>s itself, it does not <V> itself, and if the <V>er does not <V> itself, it does <V> itself. So in either case of it <V>s or does not <V>, it may be taken to <V> and not <V>. It is in "A and not A", so false.(it is already assumed that A, and resulted in not A, this seems to be the way)

If an assumption has some false statement as a result, it seems to be right to judge it false.

The <V>er that <V>s all (and only those) who don't <V> themselves

can be a false assumption in that the <V>er does not <V> all (and only those) who don't <V> themselves.

Like sushi (talk) 08:50, 16 October 2009 (UTC)

Yes, the barber described in Barber paradox does not exist. Russell's paradox is a paradox only if you assume that every set that can be defined must also exist. If you accept that it's possible to describe sets such that the description correspond to no existing set then there is no paradox. However it creates the obvious question, when you describe a set, how do you know that a set satisfying your description exists? Taemyr (talk) 09:30, 16 October 2009 (UTC)

Sorry. About the first of Gödel's incompleteness theorems, in saying "false but is provable to be true", I had an illusion that "provable to be true" includes "true". It must be otherwise, that "true" includes "provable to be true".

And I have no clue about how to answer "the obvious question", "how can we know that a set of descripion exists?". Could you give me any hint about it or examples of inexistable sets? (Please don't just link to ZFC or something, it is too complicated for me to understand the reason)

Like sushi (talk) 10:36, 16 October 2009 (UTC)

Provable usually implies true. Otherwise it's usually seen as a problem with the proof system. However if you want something to be false but provable then either you must get rid of that implication, so you must accept that provable does not necesarry imply true, or you must accept that something can be both true and false.

Well Russells set, the set of all sets that are not members of themselves, is an example of a set that can not possibly exist. Unless I am mistaken it's not possible to creat a general method to find wether or not a proposed set exists or not. Taemyr (talk) 11:16, 16 October 2009 (UTC)

I have a suspicion about one of what are written in paraconsistent logic.

Having a single inconsistency, can we really prove any statement?

"if P and its negation are both assumed to be true, then P is assumed to be true, from which it follows that at least one of the claims P and some other (arbitrary) claim A is true. However, if we know that either P or A is true, and also that P is not true (that is true) we can conclude that A, which could be anything, is true"

My suspicion is about "if we know that either P or A is true, and also that P is not true (that is true) we can conclude that A,... is true".

P is not true, but is also true. so "P or A" is already true just with P, so A can be either true or false.

Like sushi (talk) 12:18, 16 October 2009 (UTC)

Disjunctive syllogism, that is going from (A or B) And Not B, to A, is valid in clasical logic. So the inference you quote is valid in clasical logic.

For making a system paraconsisten it's fairly common to choose to get rid of Disjunctive syllogismm for precisely the reasons you qoute. See Paraconsistent_logic#Tradeoff. The problem is that when you do so you obtain a weaker system, so you can prove less in it. Taemyr (talk) 12:31, 16 October 2009 (UTC)

By the way, about Gödel's incompleteness theorems, I think saying "Being true does not guarantee provabilty" is the easiest way to present it to non-specialists.

Like sushi (talk) 12:45, 16 October 2009 (UTC)

Given the theory T is consistent then G is unprovable and true. Does that not mean G has been proven? If it means that G has been proven, If T is consistent there is an inconsistency.

Or if so, G would be provable to be true and unprovable to be true, "A and not A", so is false. Then it must be false but provable to be true. —Preceding unsigned comment added by Like sushi (talk • contribs) 01:19, 17 October 2009 (UTC)

Like sushi (talk) 01:07, 17 October 2009 (UTC)

What you can prove, without assuming very much at all, is that if T is consistent, then G_T is true. This is not the same as proving that G_T is true, without the assumption that T is consistent.

Nevertheless, we know that for all consistent T, G_T is indeed true, and we know this precisely because we can prove the implication that if T is consistent then G_T is true.

This is more confusing to state than it is to understand — once you see it, it's completely straightforward, but it's still hard to say in a way that doesn't sound like you're pulling a fast one. --Trovatore (talk) 01:28, 17 October 2009 (UTC)

Is it that we don't know if T is consistent, so can not prove G to be true?

Like sushi (talk) 02:16, 17 October 2009 (UTC)

That's a fair way of putting it for the general case, sure.

In specific cases, we may well "know" that T is consistent, for whatever value of "know" is being considered, and we will therefore also know that G_T is true. But we still won't be able to prove it from T. We may be able to formalize a proof of G_T, but it won't be a proof in T; it will be a proof in some theory of higher consistency strength. --Trovatore (talk) 02:21, 17 October 2009 (UTC)

"Given the theory T is consistent then G is unprovable and true." is not a proof in T, that G is true? (I think it may be for "T is consistent" is not in T.)

Like sushi (talk) 02:46, 17 October 2009 (UTC)

Exactly, this is not a proof in T that G_T is true, because given that T is in fact consistent, T cannot prove that T is consistent. That's the content of Gödel's second incompleteness theorem, and this exchange here is essentially the proof of the second incompleteness theorem. --Trovatore (talk) 03:09, 17 October 2009 (UTC)

Gödel's incompleteness theorem#Second incompleteness theorem does not have enough explanation about why "given that T is in fact consistent, T cannot prove that T is consistent"

Why cannot it?

Like sushi (talk) 06:29, 17 October 2009 (UTC)

For exactly the reason you intuited in your message of 02:46! If T could prove T is consistent, then since T can also prove that, if T is consistent, then G_T is true, it would follow (by modus ponens) that T could prove G_T is true. But we know that, if T is in fact consistent, then T cannot prove G_T. --Trovatore (talk) 06:40, 17 October 2009 (UTC)

Well, thank you. I think I understand it now.

Like sushi (talk) 07:19, 17 October 2009 (UTC)

About "Having a single inconsistency, can we really prove any statement?"

Again, in paraconsistent logic:

"if P and its negation are both assumed to be true, then P is assumed to be true, from which it follows that at least one of the claims P and some other (arbitrary) claim A is true. However, if we know that either P or A is true, and also that P is not true (that is true) we can conclude that A, which could be anything, is true"

My suspicion is, in this time, about "at least one of the claims P and some other (arbitrary) claim A is true". Can A not be either true or false independent of P?

Like sushi (talk) 02:16, 17 October 2009 (UTC)

I would strongly advise you to forget about the concepts of "true" and "false" when thinking about paraconsistent logics and Gödel's theorem, at least at first. In a classical logic, every proposition P has an opposite ¬P. Ideally, for every proposition P, you want the formal system to prove either P or ¬P, but not both—i.e. you want it to "settle every question". Such a system is said to be consistent and complete. If neither P nor ¬P is provable for some P, the system is incomplete. If both are provable, it's inconsistent. What Gödel's theorem shows is that in any formal system that contains enough arithmetic, there's a proposition P that's provable if and only if ¬P is provable, so the system is either inconsistent or incomplete. That's all. The stuff about P being true or false is interpretation on top of that, it's not something that Gödel proved. In nonclassical logics this binary pairing of P and ¬P doesn't work any more because there's no dualizing operation that exchanges them. Instead of going P ⇒ ¬P ⇒ P ⇒ ¬P ⇒ ⋯, negation goes P ⇒ ¬P ⇒ ¬¬P ⇒ ¬¬¬P ⇒ ⋯. So you no longer have a black-and-white true-and-false world of yes-no questions. Just because you have P and ¬P doesn't mean you've necessarily covered all the bases. -- BenRG (talk) 10:55, 17 October 2009 (UTC)

For paraconsistent logics, which really have no interpretation, it might make sense to forget about truth. If for no other reason that, once you start thinking about truth, you'll realize that paraconsistent logic is bogus. (Oh, not completely bogus — it probably does model certain aspects of commonsense reasoning that deserve to be modeled — but bogus in the sense that it's not a useful view of mathematical reality.)

But I completely disagree with you that you shouldn't think about truth in order to understand the Gödel theorems. On the contrary, keeping track of what's actually true, separate from what's provable in T, is the most effective way to avoid getting confused. As to what "Gödel actually proved", you have to keep in mind the times — philosophers in the thirties tended to view truth with some suspicion, and it's quite likely that Gödel wanted to avoid that fight. --Trovatore (talk) 19:03, 17 October 2009 (UTC)

Sorry, the explanation was not suffice. In saying "in paraconsistent logic" I just wanted to show the source. I wanted to link to Definition section, which shows "principle of explosion" (the explanation of which I quoted) in classical logic. The question in the post above is about that.

Like sushi (talk) 12:14, 17 October 2009 (UTC)

(Deindenting). A logic is a set of rules for what constitutes a valid inferrence within that logic. In Classical logic we are allowed to go from (A or P) and ¬P, to A. There really isn't anything more to say, if you create a system where Disjunctive syllogism is an invalid inference then you create a system that can prove less. This might be a tradeoff you are willing to make if you can't know that your assumptions are not contradictory. But it is a different system. Taemyr (talk) 14:24, 17 October 2009 (UTC)

I see now that it is that P and not P then P, and Disjunction introduction P then P or A. So it is that whatever A is true

Thank you.

Like sushi (talk) 03:38, 18 October 2009 (UTC)

Lottery: 6 numbers out of 49, expected values in case of large jackpots

Let’s assume that one lottery ticket costs 1$. I can calculate the probabilities of matching 6, 5, 4 or 3 numbers (cases when prizes are paid). 40% of money paid into lottery is for prizes. 44% of total prizes goes for the main prize (for matching 6 numbers), 8% for the second prize (matching 5 numbers). There is a cumulation/jackpot (If no one wins the main prize the money goes for next drawing). Is there a way to calculate when (in terms of the size of the jackpot) buying a lottery ticket gives expected value greater than zero? —Preceding unsigned comment added by 213.158.197.100 (talk) 13:28, 14 October 2009 (UTC)

Sorry, I don't understand. Does 44% of the jackpot go to the first prize (i.e. all six numbers) and 8% of the jackpot go to the second prize (i.e. five numbers)? Why do you say that 40% goes to prizes and what are the percentages for four and three numbers? If you know the percentage of the jackpot awarded to three, four, five and six numbers then just calculate the size of the jackpot that makes the expected payoff more than $1. Zain Ebrahim (talk) 14:09, 14 October 2009 (UTC)

100$ - money paid as lottery tickets; 40$ - total prize fund; 17.6 $ - money for the main prize (44% * 40$); Percentages for four and three numbers are not stated —Preceding unsigned comment added by 213.158.197.100 (talk) 14:44, 14 October 2009 (UTC)

Since the expected winning is a function of the relative amounts paid to the third and fourth prizes, the most I can see is a range of jackpots. Use the hypergeometric distribution to calculate the probabilities. The upper bound of the range will be when 48% of the prize fund goes to the third place (i.e. the less likely) and the lower bound when 48% goes to fourth place. Zain Ebrahim (talk) 15:16, 14 October 2009 (UTC)

Ignoring the possibility of sharing the jackpot, it should be this easy: For each $1 you invest, you can expect to win back $.40, if there's no jackpot. So the expected jackpot winning must exceed $.60, which means that the jackpot must be at least $.60/P(6), where P(6) is the probability of matching 6 numbers. To take jackpot sharing into account, we would also need to know how many people are playing. —JAO • T • C 09:54, 15 October 2009 (UTC)

Also, the cumulative effect would seem to make your odds better when the bonus is built up to a higher level, but remember that this also causes many more people to play (and for each of them to buy more tickets), such that the prize is more likely to be split multiple ways (especially if you bet any type of common pattern that many others will bet, too). StuRat (talk) 14:55, 18 October 2009 (UTC)

October 15

simple formula for fractions in computer code

i am writing a simple computer program that has a draggable slider in it. this slider at its minimum can have a value of 1 and at its maximum it can have a value of 1000. now, i have another object, let's say it is a light. when off, it is black and has a value of 1, it rises through shades of grey and at complete brightness (pure white) it has a value of 100.

i want this light to interact according to the slider between the numbers 400-500. so if the slider is valued at 1 to 400, the light would be black, between 401-499 would be shades of grey and from 500 to 1000 would be pure white.

what mathematical formula, without the use of code, would i use to solve this? i want my light to interpret the numbers 1-100 the same way my slider interprets its value of 400-500 in its range of 1-1000.

i hope i made this clear enough D:

thanks! Bonusbox (talk) 01:42, 15 October 2009 (UTC)

I'm not sure what you mean by "without the use of code". Do you mean, no "if" statements? Are you looking for a single mathematical expression that could be translated into a single line of mathematics-only code?

In general, what you want to do is take your slider's value (1-1000) and subtract 400 from it. Then, if the result is less than 0, make it zero. If the result is greater than 100, make it 100.

If you have MIN and MAX functions that return the smallest and largest passed values (respectively), then you could do something like this (make sure that your math operations are done in signed integers):

lightValue = MIN(MAX(sliderValue - 400, 1), 100)

Note that slider value 401 translates to light value 1, which you initially said was black. That's a bit different from what you said later, "between 401-499 would be shades of grey".

Does that help? –RHolton≡– 04:29, 15 October 2009 (UTC)

excellent. i guess code was required. in actionscript it turned out to be: lightValue = Math.min(Math.max(sliderValue - 400, 1), 100);

all working now though. thanks! Bonusbox (talk) 21:51, 15 October 2009 (UTC)

Code is not strictly speaking required. You could use polynomial interpolation. Assuming the slider only gives integer values you would get a degree 999 polynomial. This might be prone to overflow though. Taemyr (talk) 09:40, 16 October 2009 (UTC)

I was wondering if there was a strictly math-based (no code) approach (no matter how impractical). Would polynomial interpolation result in a one to one mapping? –RHolton≡– 13:30, 16 October 2009 (UTC)

Well, let's see... You are only interested in integer values, right? Thus you have a function that is defined in 999 points (x; y) and no other points really matter. So, you can act as if you wanted to get some values between them - in such case polynomial interpolation would give you exact fit in those 999 points (that is, if we could somehow avoid the rounding errors). Of course, finding a polynomial with a degree of 998 (and then calculating its value) can safely be considered to be "impractical" in presence of such a simple alternative... --Martynas Patasius (talk) 16:50, 17 October 2009 (UTC)

You've made me curious as to why you want to do this, instead of using the full slider to adjust brightness. Do you intend to use slider positions below 400 and above 500 to also do something besides brightness adjustments ? StuRat (talk) 14:49, 18 October 2009 (UTC)

October 16

cbse mathematics question papers

i want cbse 11th maths chapterwise question papers. Please help me....... —Preceding unsigned comment added by 59.92.241.69 (talk) 12:45, 16 October 2009 (UTC)

I think "cbse" means Central Board of Secondary Education. I do not understand the rest of your request. Do you want study questions? –RHolton≡– 13:44, 16 October 2009 (UTC)

It sounds like OP's asking for past papers? Vimescarrot (talk) 22:38, 16 October 2009 (UTC)

The Limiting Function of a Fourier Series

So given a Fourier series, we can say a lot about the function from it was generated. Things like convergence (in various norms), if the function is even, odd, or neither, if it is continuous, differentiable, how many times is it continuously differentiable, even things like the period of the function. My question is, given a Fourier series, is there some sort of an "inverse" operation we can do to retrieve the original function exactly? Is there any way to tell what function was used to generate the given Fourier series? My questions in particular is about this
$g(x)={\sqrt {5}}-\sum _{n=1}^{\infty }\left({\frac {\sin(n{\sqrt {5}})}{n^{4}}}\cos(nx)+{\frac {1}{n^{4}}}\sin(nx)\right).$
What is this converging to? Thanks! -Looking for Wisdom and Insight! (talk) 23:25, 16 October 2009 (UTC)

I'm not exactly sure what you're asking. I mean, there exist various manipulations you can do on series to rewrite them as other series, but the series you've given seems to be a perfectly good closed-form expression for a function to me. Ray^Talk 02:03, 17 October 2009 (UTC)

The answer is no. For example, look at two periodic functions each with one jump discontinuity, differing from each other ONLY in that one is continuous from the right and the other from the left at that jump. They both have the SAME Fourier series. Michael Hardy (talk) 02:17, 17 October 2009 (UTC)

...but I should add that there is an inverse operation if, instead of pointwise defined functions, you look at Fourier series of certain sorts of generalized functions, or Fourier series of well-behaved measures. Michael Hardy (talk) 02:19, 17 October 2009 (UTC)

This seems like a circular question. The function that generates the series above is g(x). In other words series converges to a function that generates it. If your asking if there is some way to generate a closed form expression for the function given the series then the answer is probably no.--RDBury (talk) 14:22, 17 October 2009 (UTC)

"series converges to a function that generates it" is well known to be false. Michael Hardy (talk) 14:40, 17 October 2009 (UTC)

We have an article on this: Convergence of Fourier series. Michael Hardy (talk) 14:47, 17 October 2009 (UTC)

I should have added "with assumptions about absolute convergence etc." The terms in the series given are all O(1/n⁴) so I kind of assumed that would be implied by the context.--RDBury (talk) 15:03, 17 October 2009 (UTC)

The fact remains (as I pointed out above) that more than one function generates the same series. Isn't that what the question asked? Michael Hardy (talk) 15:10, 17 October 2009 (UTC)

My interpretation of the question is he wants an expression for the function that generates the series. What I should of said is that g(x) is one such function. But I think what the original poster was really looking for is a closed form expression and I doubt it exists. Not really sure why this is turning into an argument, you clearly know a lot more about the subject than I do and I've already agreed that I should have been more careful about the statement I made.--RDBury (talk) 15:30, 17 October 2009 (UTC)

OK, maybe I'm getting distracted by the language that was used. The poster used the word "inverse". The way I am accustomed to think of it, summing the series is the inverse operation. So if the question is whether the inverse operation can tell you the original function, the answer is that there are certainly limits on how far you can take that, and the question of what the limits are is complicated.

But if the poster meant "How do you sum the series?", that's another matter. In this case n&sqrt;5 looks as if it might not be easy to deal with, but I'm not sure. Michael Hardy (talk) 16:59, 17 October 2009 (UTC)

Rewrite the sines and cosines as complex exponentials; $g(x)$ can then be written in closed form quite easily in terms of polylogarithms. It's doubtful that it can be reduced to elementary functions, but I'm not sure. Fredrik Johansson 17:22, 17 October 2009 (UTC)

I see your point. I was thrown by the n root 5 as well but you can expand the product into functions of $x\pm {\sqrt {5}}$. If it was n³ in the denominator then it looks like you could get a piecewise polynomial function by repeated integration of the sawtooth function, but the parity is wrong for that in this case.--RDBury (talk) 18:27, 17 October 2009 (UTC)

Okay so I understand, in general, two functions may be distinct (on a set of measure zero? or just countably many points? or just a finite number of points?) and yet may still generate the same exact Fourier series. But for this specific example, the convergence is uniform so the limiting function must be continuous for all real numbers. In fact, it is continuously differentiable 3 times (and the Fourier series can also be differentiated term by term 3 times). So the generating function must be unique. The problem is that I was given this Fourier series WITHOUT the function which generated it so I am just curiously investigating if there is any way to retrieve the original generating function. -Looking for Wisdom and Insight! (talk) 04:59, 18 October 2009 (UTC)

Two L¹ functions have the same Fourier coefficients if and only if they disagree on a set of measure zero. In particular, the generating function is never unique. — Emil J. 11:58, 19 October 2009 (UTC)

October 17

Quadratic forms on real matrices

Hi all - I've been given a problem to show that the map $A\mapsto tr(A^{2})$ is a quadratic form on Mat_n($\mathbb {R}$), and find its rank & signature (where tr(A) denotes the trace of A).

The first part is no problem - I'm using the definition of a quadratic form where Q is a q.f. iff $Q(\lambda \,v)=\lambda ^{2}\,Q(v)$ and the map $(u,\,v)\,\mapsto \,{\frac {1}{2}}(Q(u+v)-Q(u)-Q(v))$ is bilinear, so that's fairly simple. However, I don't even understand how to begin finding the trace or the signature - with quadratics like $x^{2}+2xy+4xz+z^{2}$ I know to write them in the form x^TAx for vector x=(x,y,z) and symmetric matrix A, but I'm in a little over my head here, how do I get started when I'm working with n*n matrices instead?

Many thanks, Mathmos6 (talk) 01:08, 17 October 2009 (UTC)

Well you can write it out as a quadratic: If A=[a,b;c,d] then tr(A^2) = a^2 + d^2 + 2bc. The matrix of the quadratic form is just a permutation matrix: [1,0,0,0;0,0,1,0;0,1,0,0;0,0,0,1]. If A=[a,b,c;d,e,f;g,h,i], then tr(A^2) = a^2 + e^2 + i^2 + 2bd + 2cg + 2fh. JackSchmidt (talk) 04:37, 17 October 2009 (UTC)

You can write it out as a quadratic for fixed n, but I think the OP wants it for general n. --Tango (talk) 09:54, 17 October 2009 (UTC)

If A = (a_ij) is an n × n matrix then, as the trace article shows:

${\mbox{tr}}(A^{2})=\sum _{i=1}^{n}\sum _{j=1}^{n}a_{ij}a_{ji}.$

This gives an explicit expression for the quadratic form. The trace and signature should follow from calculations. Find the symmetrix matrix, say M such that vMv^T = tr(A²) where

$v=(a_{11},\ldots ,a_{1n},a_{21},\ldots ,a_{2n},\ldots ,a_{nn}).$

Finding M shouldn't be all that difficult. For example, in the case where n = 2 then tr(A²) = a₁₁² + 2a₁₂a₂₁ + a₂₂² and so

$M=\left({\begin{array}{cccc}1&0&0&0\\0&0&1&0\\0&1&0&0\\0&0&0&1\end{array}}\right).$

The eigenvalues are given by −1, 1, 1, and 1, the eigenvectors are (0,−1,1,0), (0,0,0,1), (0,1,1,0), and (0,0,0,1) respectively. The rank and signature follow. ~~ Dr Dec (Talk) ~~ 11:32, 17 October 2009 (UTC)

If A is symmetric and B is skew-symmetric then tr(AB)=0. Moreover, Q(A):=tr(A²) is positive definite on the space of symmetric matrices, and it is negative definite on the space of skew-symmetric matrices. So it's a non-degenerate quadratic form with signature ( n(n+1)/2 , n(n-1)/2 ). --pma (talk) 21:26, 17 October 2009 (UTC)

To help the OP "get started when [he's] working with n × n matrices", could you please explain how you came to this conclusion? ~~ Dr Dec (Talk) ~~ 13:03, 18 October 2009 (UTC)

Details: we know that the n²-dimensional vector space Mat_n is the direct sum of the n(n+1)/2 dimensional subspace Sym_n of all symmetric matrices plus the n(n-1)/2 dimensional subspace Skw_n of all skew-symmetric matrices; the decomposition being M=A+B, with A:=(M+M^T)/2 (symmetric) and B:= (M-M^T)/2 (skew-symmetric). The above quadratic form Q splits in this decomposition: Q(M)=Q(A)+Q(B), because tr(AB)=0. Since Q is positive definite, respectively negative definite, on Sym respectively on Skw, the conclusion follows. In terms of the Hilbert-Schmidt norm, we can write Q(M)=|A|²-|B|².

Note that for n=2 Dr Dec's computation gives a signature (3,1) as it has to be. Moreover, if you check the eigenvectors he provides, they correspond to the skew-symmetric 2x2 matrix with 0 on the diagonal (of the eigenvalue -1), and three symmetric 2x2 symmetric matrices (of the eigenvalue 1). This suggests the generalization of the diagonalization of Q for any n. Let E_ij be the matrix with all entries 0 but the ij-entry equal to 1. Consider the base of Mat_n given by the n(n+1)/2 symmetric matrices S_ij:=E_ij+E_ji for i≤j and the n(n-1)/2 skew-symmetric matrices S_ij:= E_ij-E_ji for i>j: clearly tr(S_ij S_i'j')=0 if (i,j)≠(i',j'); Q(S_ij)>0 for i≤j and Q(S_ij)<0 for i>j. (As usual, making computations is a necessary step in order to avoid making computations ;-) --pma (talk) 22:40, 18 October 2009 (UTC)

Good way to remember meet vs join?

I can never remember which of Meet and Join is which (and the symbols $x\land y$ and $x\lor y$). Any suggestions for a good way to remember? — Matt Crypto 17:29, 17 October 2009 (UTC)

Meet looks like intersection which looks like an 'n' and join looks like union which looks like a 'u'. Does that help?--RDBury (talk) 18:10, 17 October 2009 (UTC)

The meet of two sets is where they meet, i.e. the intersection (and has a symbol resembling the intersection symbol). The join of two sets is the two sets joined together, i.e. the union (and again, the symbol is similar). Algebraist 18:15, 17 October 2009 (UTC)

(ec) One way I can think of is that if you have a power set with partial order x ≤ y for $x\subset y$, then $x\land y=x\cap y$ and $x\lor y=x\cup y$.

More generally, $\land ,\cap$ are like 'A' for "and" and $\lor ,\cup$ are like 'U' for "union". Rckrone (talk) 18:28, 17 October 2009 (UTC)

Let's add $\scriptstyle \prod$ and $\scriptstyle \coprod$. Note that cup-like symbols denote inductive limits while cap-like symbols, like the product, denote projective limits. Also, it may help the mnemonic to recall that (like $\cup$ is for "Union"), $\lor$ stands for the Latin conjunction vel, "or" (indeed, x is in $A\cup B$ iff x is in A, or in B). --pma (talk) 21:47, 17 October 2009 (UTC)

Flip the symbol upside-down and that's what the Hasse diagram would look like. Or, remember that "AND" can only make truth values go down, and "OR" can only make them go up. — Carl (CBM · talk) 01:01, 19 October 2009 (UTC)

Certain conformal map projections

At Talk:Dymaxion map, I posted a hastily scrawled question, which I then amended a few minutes later. This seems like a better place to find an answer than that page.

The Dymaxion map is a world map on the surface of an icosohedron (or maybe in some cases some other polyhedron?). It can unfold to make a flat map, perhaps with less distortion (except maybe along the edges and certainly at the vertices) than some other projections of the whole world, or nearly the whole world, onto a page.

The simplest way to do that would be 20 separate stereographic projections, I would think. Stereographic projections are conformal.

So that would make the Dymaxion map conformal except at the vertices and possibly along the edges.

My guess is it's not conformal along the edges.

And obviously it could not be conformal at the vertices.

Two questions:

• Is my guess about the edges right? (Maybe I'll fiddle with this and see if I can find the answer before anyone answers here; it doesn't seem like a hard question.)
• If so, a possibly harder question to answer: is there some other way to do it, i.e. not stereographic, that makes it conformal everywhere except at the 12 vertices?

Michael Hardy (talk) 23:04, 17 October 2009 (UTC)

If the map consists of portions of 20 stereographic projections, is the centre of projection in each case the centre of the globe? Is so, the edge of each triangular face will, I think, correspond to a great circle on the globe, in which case conformality will be maintained on each side of that boundary line and, I think, across it. But intuition can only go so far, and my 3D trigonometry isn't up to a formal proof.→217.43.210.13 (talk) 12:37, 18 October 2009 (UTC)

Oh: I was being stupid. I don't think it's a stereographic projection if the center of projection is the center of the sphere. But the center of the sphere seems like the obvious place from which to project. Is projection from the center of the sphere conformal? Michael Hardy (talk) 18:03, 18 October 2009 (UTC)

I think the radial projection (from the south hemisphere S₊ to C, seen as the tangent plane at the south pole) is not conformal. Without making the computation, the reason is: if it were conformal, we could compose the inverse of the radial projection, C → S₊, with the standard stereographic projection S → C; the result would be a radial holomorphic map C →C, that is, of the form f(re^it)=φ(r)e^it, which can be holomorphic only if f(z)=az, which is not the case here (alternatively, it would be a bounded holomorphic map on C, hence constant by Liouville theorem, again not the case here). --pma (talk) 21:37, 18 October 2009 (UTC)

Just to fix some data of the problem. An icosahedron I minus the set of its vertices V is a Riemann surface with the structure given by quotient of an open set U of C (this corresponds exactly to the concrete construction: U is the union of 20 open equilateral triangles of the plane expansion, together with the tabs to be glued to the adjecent faces). Otherwise, we can define the holomorphic structure by an atlas, the charts being the maps that take the rhombus made by 2 equilateral triangles in C, onto a pair of adjacent faces (the transition maps are then of the form az+b). This Riemann surface is of course homeomorphic to the Riemann sphere S minus a set of 12 points (that we may identify as V, the set of vertices of I, if we think I as inscribed in S). The piecewise stereographic projection is certainly a homeomorphism of S onto I, but it is certainly not holomorphic on S\V : it is a stereographic projection on each face, but exactly for this reason it can't be globally holomorphic (by unique continuation it would be a unique stereographic projection, read on the charts. In particular, without all the icosahedral symmetry of the piecewise sereographic map). However, it seems at least $C^{1}$, even at the edges. So I agree with you, as to the first question (however I couldn't find out there the details of the construction of the Dymaxion map, and we are not sure if the Dymaxion is actually the piecewise stereographic projection). The second question is very interesting: is there a homeomorphism h between I and S which is bi-holomorphic between I\V and S\V? I think the answer is yes (and if so, I guess this one is the true Dymaxion map). I suspect the point is that, even if the holomorphic structure of I\V can't be extended to I, there is a 6m-fold branched covering of I that admits a global holomorphic structure, and that is bi-holomorphic to the analogous branched covering of S. This should give the map h. In any case, if we could make this thing clear and at elemenary level, it should be a very beautiful addition to the article. --pma (talk) 21:11, 18 October 2009 (UTC)

October 18

Getting pi with a limit

I was reading Method of exhaustion, and I wanted to try out Archimedes' method for calculating pi with a limit. Finding the limit as the number of sides of a regular polygon approaches infinity of the ratio of its area to its apothem squared gave me:

$\lim _{n\to \infty }\ n\tan \left({\frac {180^{\circ }}{n}}\right)$

Obviously, this equals pi, but I have no clue how to show that. I only have a very basic understanding of limits; could someone show me how to evaluate that expression to obtain pi? Thegreenj 00:04, 18 October 2009 (UTC)

If you know about derivatives, your limit is just the derivative of tan(πx) at x=0, π indeed. That's not Archimede's method, of course. He compared the perimeter of the inscribed and of the circumscribed regular n-gons, that are very close for large n -and π is in between. If n is a power of two, meaning that you double the vertices recursively, the formula for the perimeter is somehow simpler, giving rise to Viète's formula. --78.13.138.118 (talk) 00:20, 18 October 2009 (UTC)

Remember that tan = opp/adj. Suppose the circle has unit radius, as you often do in trigonometry. Draw the triangle in which the "adjacent" side is from (0,0) to (1,0) and the "opposite" side is a vertical line of height tan(180°/n). Then the angle between "adjacent" and "hypotenuse" is 180°/n (draw the picture). That side is longer than the arc of the circle between the "adjacent" side and the hypotenuse. Stack up n of these, going counterclockwise from the positive half of the x-axis to the negative half. The sum of the lengths of those "opposite" sides is n tan(180°/n). And it approaches the length of the arc. Michael Hardy (talk) 00:49, 18 October 2009 (UTC)

• Thanks to both of you, though I admit I don't wholly understand 78.13.138.118's response, and I seem to have misunderstood Archimemdes' method (the text and the graphic in the article are different methods). Thegreenj 01:52, 19 October 2009 (UTC)

The value of $\tan(x)$ at $x=0$ is 0, and the derivative of $\tan$ at $x=0$ is 1. Consider $x:={\frac {\pi }{n}}$, that goes to 0 as $n$ goes to infinity: then

$\lim _{n\to \infty }n\tan \left({\frac {\pi }{n}}\right)=\lim _{n\to \infty }\pi {\frac {\tan \left({\frac {\pi }{n}}\right)}{\frac {\pi }{n}}}=\lim _{x\to 0}\pi {\frac {\tan(x)}{x}}=\pi \lim _{x\to 0}{\frac {\tan(x)-\tan(0)}{x}}=\pi {\frac {\mathrm {d} }{\mathrm {d} x}}\tan(x){\Big |}_{x=0}=\pi .$

Archimedes compared π with the perimeter of the inscribed and circumscribed 96-sided regular polygons, getting the celebrated bounds 3+10/71 < π < 3+1/7. The choice of 96 = 3·2⁵ corresponds to starting with an equilateral triangle and bisecting 5 times). PS: Indeed you are rigth about the article being a bit unclear: Archimedes did not use exhaustion to compute π, but to prove the formula πr² for the area inside a circle, that is, to prove that the ratio of a circle to a square with edge equal to the radius is the same as the ratio of the circumference to the diameter, that is π. And, of course, the bounds he gave on π use no exhaustion. (Fixed). --pma (talk) 10:39, 19 October 2009 (UTC)

Application of the PNT to a sequence of numbers

82818079787776757473727170696867666564636261605958575655545352515049484746454443424140393837363534333231302928272625242322212019181716151413121110987654321 has just been determined to be the only prime in the implied sequence up to the thousandth term. However, it seems to me by using the PNT that there should be infinitely many primes in the sequence, with the probability of each third (two out of three have 3 dividing them) term being prime going like C/D, where D is the number of digits and C is a constant. Is this right?Julzes (talk) 04:10, 18 October 2009 (UTC)

Yes that's surprising. Scrubbed my reply after thinking more about it. I thought you'd get 2/7 removed for factor 7 etc but my reasoning doesn't follow Dmcq (talk) 10:38, 18 October 2009 (UTC)

It seems a little tricky to determine the constant, since you have clear conditioning for the primes 2, 3, and 5 (never divisible by 2 or 5, and also never by 3 for the terms worth looking at). It's possible to use modularity for other primes too, I suppose, to eliminate other sets of numbers. Right now I'm looking at whether I can get a second term in the sequence. Perhaps someone might be interested in taking a more certain route (or less human-intensive) than the one I am, which is to simply keep trying numbers input by prefixing three numbers at alpertron.com. The base-7 case also looks intriguing. I know there is no solution up to the term beginning 1000 (=343), and only a much smaller fraction of cases are even worth looking at. First terms are trivial for bases 2 through 6, but that is about all I know about the alternative base problem.Julzes (talk) 11:34, 18 October 2009 (UTC)

I posted at User talk:PrimeHunter#New Prime Discovered. I guess you mean http://www.alpertron.com.ar/ECM.HTM. That applet is designed for finding prime factors and is relatively slow at primality testing. PrimeHunter (talk) 14:53, 18 October 2009 (UTC)

This is Sloane A000422. Dmcq (talk) 16:59, 18 October 2009 (UTC)

Can kite be a trapezium?

A kite.

Kite is a quadrilateral which has two pairs of adjacent sides equal. By any means can a kite become a trapezium or trapezoid i.e. two of its opposite sides are parallel? Srinivas 10:23, 18 October 2009 (UTC)

See square. --Tango (talk) 11:08, 18 October 2009 (UTC)

No, no, I mean it should be a kite only i.e. adjacent sides only must be equal and opposite sides must be unequal. What Tango (talk · contribs) said is also correct as square is a kite but I forgot to mention that opposite sides must be unequal and only adjacent sides must be equal. Srinivas 12:50, 18 October 2009 (UTC)

Any kite which is also a trapezoid must be a rhombus. Let the kite be ABCD where AB = BC and CD = DA and suppose AB is parallel to CD. Draw the two diagonal AC and BD and let O be the point of intersection. By alternate angles, angle OCD = angle OAB. Then by angle-side-angle, traingle OAB is congruent to triangle OCD, so AB = CD and all four sides are equal. It's a cute problem and it might work as an extra credit in a high geometry class.--RDBury (talk) 13:29, 18 October 2009 (UTC)

Thanks a lot, RDBury (talk · contribs), you solved my doubt clearly. Srinivas 14:38, 18 October 2009 (UTC)

Smallest number which have the first n positive integers as divisors

Is there are name for the numbers which are the smallest to have the first n positive integers as divisors? 88.104.157.141 (talk) 11:46, 18 October 2009 (UTC)

Sloane's just calls the lcms of {1, 2, ..., n}, so if there is a specific name, it can't be very widely used. Algebraist 11:54, 18 October 2009 (UTC)

Aside from the formula (name?) lcm(2,3,...,n), it could also be written as a simple product of primorials. I can't get it into notation, but it's the product from i=1 to infinity of the product of the primes less than or equal to n^(1/i). If I were naming it I would call it the "superprimorial up to n" or something. "Least common multiple of the numbers up to n" is clearer and almost as short, though.Julzes (talk) 12:08, 18 October 2009 (UTC)

Its logarithm has a name - it is the second Chebyshev function ψ(n). Gandalf61 (talk) 13:21, 18 October 2009 (UTC)

Modular forms: Finding the fundamental domain for a finite index subgroup

Argh, I can not believe I closed this window when I was almost done with my question! :) I will just write it shorter this time.

First, I must give some background. I know any matrix in SL(2, Z) can be written uniquely in the form

$T=(-1)^{r}P^{2p_{0}}VP^{2p_{1}}V\cdots VP^{2p_{n}},$

where $P={\begin{bmatrix}0&-1\\1&1\end{bmatrix}}$ and $V={\begin{bmatrix}0&-1\\-1&0\end{bmatrix}}$. Based on this representation, define $h(T)=n+2r$ for any matrix in SL(2, Z). Now, consider

$\Gamma ^{4}=\{T\in SL(2,\mathbb {Z} ):h(T)\equiv 0\mod 4\}.$

My understanding is, to get the fundamental domain of this subgroup of SL(2, Z), which is index 4 by the way, I will need to find a set of coset representatives and apply each to the "standard" fundamental domain of SL(2, Z), which is the hyperbolic triangle with vertices $\infty ,{\frac {1}{2}}+i{\frac {\sqrt {3}}{2}},-{\frac {1}{2}}+i{\frac {\sqrt {3}}{2}}$. Then, the union of the 4 images I get will be a fundamental domain of $\Gamma ^{4}$.

So, my question: How do I figure out the 4 coset representatives. Obviously the identity matrix is one of them. I know of several standard matrices from the book I am using that are not in this group, so of course these are candidates for the other 3 representatives:

$U=P^{0}VP^{4},(n+2r=1),\qquad V=P^{0}VP^{0}(n+2r=1),\qquad W=-P^{0}VP^{2}(n+2r=2),\qquad P=-P^{4}(n+2r=1)$

I noticed that $-I\Gamma ^{4}$ is going to be the set of all matrices with $h(T)\equiv 2\mod 4$ since it just increases or decreases r by 1. And, I believe that $V\Gamma ^{4}$is all those with $h(T)\equiv 1\mod 4$, and then $-V\Gamma ^{4}$ is all those with $h(T)\equiv 3\mod 4$, unless I messed something up. So, it seems as if $\pm I,\pm V$ would be a set of coset representatives. But, the action of $-I$ is trivial on the upper half plane so I end up with only two different images of the "standard" fundamental domain of SL(2, Z), and not 4. Am I doing something wrong here? Or is it fine that I end up with only two copies of that? StatisticsMan (talk) 21:26, 18 October 2009 (UTC)

What's going on is that, since the element $-I$ acts trivially, it's really the quotient group PSL(2,Z) = SL(2,Z)/$\{I,-I\}$ that's acting on the hyperbolic plane. The image of Γ⁴ in this quotient group has index 2, which is why the fundamental domain that you're getting only has two triangles. Jim (talk) 04:27, 19 October 2009 (UTC)

Use of abacus in United States education

I was reading an article on the abacus:

http://en.wikipedia.org/wiki/Abacus#School_abacus

When was the abacus first used in the United States education system to teach counting skills and things like this?

Thanks for the help!

137.81.112.223 (talk) 23:23, 18 October 2009 (UTC)

Euclidean scaling

I believe the following is correct for geometric figures. If an n-dimensional element (like an edge) is scaled by s_n, then its m-dimensional measure will scale by s_m according to the following power law:

$s_{m}\propto s_{n}^{\frac {m}{n}}$

First of all, is this entirely correct? Second, does this law have a name? I know that Euclid and Aristotle proved it for special cases and Galileo showed it for n=1 with m=2 and m=3, but I can't find this generalized form anywhere. Does it have a name? Is it attributed to anyone?

Of course these scaling laws are frequently used in fractal geometry, but explanations of this generally jump from Euclid straight to Hausdorff and Mandelbrot. risk (talk) 12:12, 19 October 2009 (UTC)

Just to clarify, the figure is m-dimensional, and made up of n-dimensional elements (like a cube made up of 2-dimensional faces). risk (talk) 13:57, 19 October 2009 (UTC)