## Entire Function That Agrees with Log on Natural Numbers?

For the discussion of math. Duh.

Moderators: gmalivuk, Moderators General, Prelates

Suffusion of Yellow
Posts: 42
Joined: Thu Jan 01, 2009 6:15 pm UTC

### Entire Function That Agrees with Log on Natural Numbers?

Do you think this can be done? I don't know much complex analysis, I suspect probably not, but I kind of need it.

Work so far:
I've been trying to work up something so far by cutting out a region and filling it in with something else, but no success.

The function Log(x)*[Cos((2^x)*Pi)+1]/2 agrees with Log on natural numbers but has a limit of zero as x goes to zero instead of negative infinity, so that might help with the singularity at zero, but I'm not sure what to do with the branch cut.

Danke

mike-l
Posts: 2758
Joined: Tue Sep 04, 2007 2:16 am UTC

### Re: Entire Function That Agrees with Log on Natural Numbers?

I'm pretty sure you can find an entire function with prescribed values on any set that lacks limit points, but I can't find the theorem that says that.
addams wrote:This forum has some very well educated people typing away in loops with Sourmilk. He is a lucky Sourmilk.

Kurushimi
Posts: 841
Joined: Thu Oct 02, 2008 12:06 am UTC

### Re: Entire Function That Agrees with Log on Natural Numbers?

I mean, isn't anything that you can draw on a graph that passes the vertical line test considered a function? I mean, you probably won't be able to define it in terms of functions we're used to, like polynomials, sine and cosine, exponentials etc., but its still follows all the rules of a function. And I can imagine very easily a function that lines up with natural log on natural numbers
.

Blatm
Posts: 638
Joined: Mon Jun 04, 2007 1:43 am UTC

### Re: Entire Function That Agrees with Log on Natural Numbers?

Kurushimi
Posts: 841
Joined: Thu Oct 02, 2008 12:06 am UTC

### Re: Entire Function That Agrees with Log on Natural Numbers?

Oh. I just thought he wanted the whole function. >.>
Damn common English words having separate mathematical meanings.

Eastwinn
Posts: 303
Joined: Thu Jun 19, 2008 12:36 am UTC
Location: Maryland

### Re: Entire Function That Agrees with Log on Natural Numbers?

I'd suggest log(|x|+1) but it's not everywhere differentiable. But it's close!
http://aselliedraws.tumblr.com/ - surreal sketches and characters.

skeptical scientist
closed-minded spiritualist
Posts: 6142
Joined: Tue Nov 28, 2006 6:09 am UTC
Location: San Francisco

### Re: Entire Function That Agrees with Log on Natural Numbers?

Eastwinn wrote:I'd suggest log(|x|+1) but it's not everywhere differentiable. But it's close!

It has to be C-differentiable and not just R-differentiable to be entire, and this function isn't C-differentiable anywhere. (The absolute value function is not terribly nice from the point of view of complex analysis.)
I'm looking forward to the day when the SNES emulator on my computer works by emulating the elementary particles in an actual, physical box with Nintendo stamped on the side.

"With math, all things are possible." —Rebecca Watson

jestingrabbit
Factoids are just Datas that haven't grown up yet
Posts: 5967
Joined: Tue Nov 28, 2006 9:50 pm UTC
Location: Sydney

### Re: Entire Function That Agrees with Log on Natural Numbers?

Suffusion of Yellow wrote:Work so far:
I've been trying to work up something so far by cutting out a region and filling it in with something else, but no success.

This is impossible. The values on any open set force the values everywhere else.

I think that starting with log is also a mistake. The essential singularity will be a real killer to get past.
ameretrifle wrote:Magic space feudalism is therefore a viable idea.

gmalivuk
GNU Terry Pratchett
Posts: 26820
Joined: Wed Feb 28, 2007 6:02 pm UTC
Location: Here and There
Contact:

### Re: Entire Function That Agrees with Log on Natural Numbers?

Kurushimi wrote: you probably won't be able to define it in terms of functions we're used to, like polynomials, sine and cosine, exponentials etc.
In spite of the misunderstanding of "entire", this is probably true.

Is there a limit for the sequence of polynomials of degree n-1 that match log for the first n positive integers?
Unless stated otherwise, I do not care whether a statement, by itself, constitutes a persuasive political argument. I care whether it's true.
---
If this post has math that doesn't work for you, use TeX the World for Firefox or Chrome

(he/him/his)

Yakk
Poster with most posts but no title.
Posts: 11129
Joined: Sat Jan 27, 2007 7:27 pm UTC
Location: E pur si muove

### Re: Entire Function That Agrees with Log on Natural Numbers?

Sure -- the empty function.

... Oh, you meant the limit as n goes to infinity. Probably not.
One of the painful things about our time is that those who feel certainty are stupid, and those with any imagination and understanding are filled with doubt and indecision - BR

Last edited by JHVH on Fri Oct 23, 4004 BCE 6:17 pm, edited 6 times in total.

WarDaft
Posts: 1583
Joined: Thu Jul 30, 2009 3:16 pm UTC

### Re: Entire Function That Agrees with Log on Natural Numbers?

gmalivuk wrote:
Kurushimi wrote: you probably won't be able to define it in terms of functions we're used to, like polynomials, sine and cosine, exponentials etc.
In spite of the misunderstanding of "entire", this is probably true.

Is there a limit for the sequence of polynomials of degree n-1 that match log for the first n positive integers?

I think that'd result in some infinitely large coefficients, which just breaks things.

Though if we divide one polynomial by another, we can do it, it won't look anything like the log function off the naturals, but hey.
All Shadow priest spells that deal Fire damage now appear green.
Big freaky cereal boxes of death.

mike-l
Posts: 2758
Joined: Tue Sep 04, 2007 2:16 am UTC

### Re: Entire Function That Agrees with Log on Natural Numbers?

My idea is, take an entire function that's 0 on the naturals (for example the Weirstrass sigma function on the lattice generated by {1,i}), then multiply it by k/(z-n) to get an entire function that is 0 on the naturals except for n, and log(n) at n. Then add them up. No idea if this converges Probably not since you'll get something looking like the harmonic series... so take the Weirstrass sigma function squared and multiply by k/(z-n)^2 for appropriate k... then it might converge (what do the k's look like?).. but does it converge to an entire function?
addams wrote:This forum has some very well educated people typing away in loops with Sourmilk. He is a lucky Sourmilk.

Eastwinn
Posts: 303
Joined: Thu Jun 19, 2008 12:36 am UTC
Location: Maryland

### Re: Entire Function That Agrees with Log on Natural Numbers?

skeptical scientist wrote:
Eastwinn wrote:I'd suggest log(|x|+1) but it's not everywhere differentiable. But it's close!

It has to be C-differentiable and not just R-differentiable to be entire, and this function isn't C-differentiable anywhere. (The absolute value function is not terribly nice from the point of view of complex analysis.)

Oops, this makes the problem far more interesting.
http://aselliedraws.tumblr.com/ - surreal sketches and characters.

undecim
Posts: 289
Joined: Tue Jan 19, 2010 7:09 pm UTC
Contact:

### Re: Entire Function That Agrees with Log on Natural Numbers?

I'm a calculus newbie, but what about
$\log\left(x\right)\left[\cos^2\left(\pi x\right)-\frac{1}{\Gamma\left(1-x\right)}\right]^2$
EDIT: Apparantly, I'm a latex newbie as well
Blue, blue, blue

NathanielJ
Posts: 882
Joined: Sun Jan 13, 2008 9:04 pm UTC

### Re: Entire Function That Agrees with Log on Natural Numbers?

undecim wrote:I'm a calculus newbie, but what about
$\log\left(x\right)\left[\cos^2\left(\pi x\right)-\frac{1}{\Gamma\left(1-x\right)}\right]^2$

Even if you plug the "holes" of that function at the natural numbers by log(x), the function isn't entire since you're multiplying by the complex logarithm.
Homepage: http://www.njohnston.ca
Conway's Game of Life: http://www.conwaylife.com

undecim
Posts: 289
Joined: Tue Jan 19, 2010 7:09 pm UTC
Contact:

### Re: Entire Function That Agrees with Log on Natural Numbers?

NathanielJ wrote:
undecim wrote:I'm a calculus newbie, but what about
$\log\left(x\right)\left[\cos^2\left(\pi x\right)-\frac{1}{\Gamma\left(1-x\right)}\right]^2$

Even if you plug the "holes" of that function at the natural numbers by log(x), the function isn't entire since you're multiplying by the complex logarithm.

I thought that the only reason the complex logarithm was not an entire function is because of the pole at 0. With this equation, the limit as x approaches 0 is 0 (unless I screwed up somewhere, but the plot of that function appears to agree with me. Maybe I should stop doing math before my morning coffee)

Or did I completely misunderstand something?

EDIT: Yup, I did misunderstand something. I just looked at the surface plot of Im(ln(z)) in Genius.

EDIT: Wouldn't this fix it, or does absolute value break it in another way?
$\left|\log\left(x\right)\right|\left[\cos^2\left(\pi x\right)-\frac{1}{\Gamma\left(1-x\right)}\right]^2$

P.S: I learn about math on Wikipedia and Mathworld, so if you can be explicit about anything wrong with it, it would help me learn
Blue, blue, blue

skeptical scientist
closed-minded spiritualist
Posts: 6142
Joined: Tue Nov 28, 2006 6:09 am UTC
Location: San Francisco

### Re: Entire Function That Agrees with Log on Natural Numbers?

Generally, any function defined using absolute value in a non-trivial way is going to fail to be complex analytic.
I'm looking forward to the day when the SNES emulator on my computer works by emulating the elementary particles in an actual, physical box with Nintendo stamped on the side.

"With math, all things are possible." —Rebecca Watson

squareroot
Posts: 548
Joined: Tue Jan 12, 2010 1:04 am UTC
Contact:

### Re: Entire Function That Agrees with Log on Natural Numbers?

gmalivuk wrote:
Kurushimi wrote: you probably won't be able to define it in terms of functions we're used to, like polynomials, sine and cosine, exponentials etc.
In spite of the misunderstanding of "entire", this is probably true.

Is there a limit for the sequence of polynomials of degree n-1 that match log for the first n positive integers?

This EXACT question was in a high-school-level math competition I did a week ago. (Well, log base 2.)
I didn't solve it. :-/
And the website is terribly organized, they don't have the answer up yet.
Ah well.

(Btw - it didn't seem to lead to any infinite coefficients. In fact, the terms seemed to generally shrink.)
<signature content="" style="tag:html;" overused meta />
Good fucking job Will Yu, you found me - __ -

forgetful functor
Posts: 10
Joined: Mon Nov 22, 2010 10:40 am UTC

### Re: Entire Function That Agrees with Log on Natural Numbers?

Given a set [imath]\{z_n\}[/imath] of complex numbers with [imath]\lvert z_n \rvert \rightarrow \infty[/imath], it is always possible to find an entire function with a prescribed value at each [imath]z_n[/imath]. This is a consequence of the Weierstrass Factorization Theorem and the Mittag-Leffler Theorem (see Ahlfors or any other complex variables text). Weierstrass can give you an entire function [imath]f[/imath] with a simple zero at each [imath]z_n[/imath], and Mittag-Leffler can give you a holomorphic function [imath]g[/imath] with singular part [imath]\frac{a_n}{z-z_n}[/imath] at each [imath]z_n[/imath]. Then [imath]fg[/imath] is an entire function with [imath](fg)(z_n)=a_n[/imath] for each [imath]n[/imath].

So your function exists. If you need an explicit expression, try something like:
$\sum_{n=1}^\infty \frac{\sin(\pi z)}{\pi \cos(\pi n)} \frac{e^{c_n (z-n)}}{z-n} \log(n)$

where the [imath]c_n[/imath] are some constants chosen so that the sum converges.

mike-l
Posts: 2758
Joined: Tue Sep 04, 2007 2:16 am UTC

### Re: Entire Function That Agrees with Log on Natural Numbers?

Right, I was trying to remember Mittag-Leffler, but only remembered Runge (which is used in proving it). Thanks!

A small correction, [imath](fg)(z_n) =k_na_n[/imath] where k depends only on [imath]f[/imath], so you can pick the [imath]a_n[/imath] appropriately when building [imath]g[/imath].

As an aside, I wonder why the conditions on the two theorems are stated as [imath]|z_n| \to \infty[/imath] for one and discrete for the other... these seem equivalent to me, as any discrete set can only have finitely many values in any compact subset, in particular for the set [imath]|z| \le M[/imath], and any sequence whose magnitude grows without bounds cannot have a limit point (what would it's magnitude be?) Am I missing some subtle (or not so subtle) difference? (I guess one has a prescribed order, but a discrete set must be countable, so it may as well be a sequence)
Last edited by mike-l on Mon Nov 22, 2010 2:26 pm UTC, edited 4 times in total.
addams wrote:This forum has some very well educated people typing away in loops with Sourmilk. He is a lucky Sourmilk.

skeptical scientist
closed-minded spiritualist
Posts: 6142
Joined: Tue Nov 28, 2006 6:09 am UTC
Location: San Francisco

### Re: Entire Function That Agrees with Log on Natural Numbers?

forgetful functor wrote:Given a set [imath]\{z_n\}[/imath] of complex numbers with [imath]\lvert z_n \rvert \rightarrow \infty[/imath], it is always possible to find an entire function with a prescribed value at each [imath]z_n[/imath]. This is a consequence of the Weierstrass Factorization Theorem and the Mittag-Leffler Theorem.

Ah thanks. Mike-l guessed this was true, and I also thought it was probably true, but we didn't have the proof.
I'm looking forward to the day when the SNES emulator on my computer works by emulating the elementary particles in an actual, physical box with Nintendo stamped on the side.

"With math, all things are possible." —Rebecca Watson

forgetful functor
Posts: 10
Joined: Mon Nov 22, 2010 10:40 am UTC

### Re: Entire Function That Agrees with Log on Natural Numbers?

mike-l wrote:As an aside, I wonder why the conditions on the two theorems are stated as [imath]|z_n| \to \infty[/imath] for one and discrete for the other... these seem equivalent to me, as any discrete set can only have finitely many values in any compact subset, in particular for the set [imath]|z| \le M[/imath], and any sequence whose magnitude grows without bounds cannot have a limit point (what would it's magnitude be?) Am I missing some subtle (or not so subtle) difference? (I guess one has a prescribed order, but a discrete set must be countable, so it may as well be a sequence)

No difference, must just be the text you are looking in. If you have any set [imath]S \subset C[/imath] without a limit point in [imath]C[/imath], then you can find a holomorphic (or meromorphic) function with zeros (or poles) at the points of [imath]A[/imath].

mike-l wrote:A small correction, [imath](fg)(z_n) =k_na_n[/imath] where k depends only on [imath]f[/imath], so you can pick the [imath]a_n[/imath] appropriately when building [imath]g[/imath].

Thanks, I knew what I had seemed to easy. The example I gave should still work, though.

Suffusion of Yellow
Posts: 42
Joined: Thu Jan 01, 2009 6:15 pm UTC

### Re: Entire Function That Agrees with Log on Natural Numbers?

forgetful functor wrote:Given a set [imath]\{z_n\}[/imath] of complex numbers with [imath]\lvert z_n \rvert \rightarrow \infty[/imath], it is always possible to find an entire function with a prescribed value at each [imath]z_n[/imath]. This is a consequence of the Weierstrass Factorization Theorem and the Mittag-Leffler Theorem (see Ahlfors or any other complex variables text). Weierstrass can give you an entire function [imath]f[/imath] with a simple zero at each [imath]z_n[/imath], and Mittag-Leffler can give you a holomorphic function [imath]g[/imath] with singular part [imath]\frac{a_n}{z-z_n}[/imath] at each [imath]z_n[/imath]. Then [imath]fg[/imath] is an entire function with [imath](fg)(z_n)=a_n[/imath] for each [imath]n[/imath].

So your function exists. If you need an explicit expression, try something like:
$\sum_{n=1}^\infty \frac{\sin(\pi z)}{\pi \cos(\pi n)} \frac{e^{c_n (z-n)}}{z-n} \log(n)$

where the [imath]c_n[/imath] are some constants chosen so that the sum converges.

Cool, thank you very much. I do need an explicit form (or something close to it), i'll look into that.

I've been messing around with something that looks like c1*(e-x)+c2*(e-x)(e^2-x)+c3*(e-x)(e^2-x)(e^3-x) but I haven't shown convergence (yet).

Suffusion of Yellow
Posts: 42
Joined: Thu Jan 01, 2009 6:15 pm UTC

### Re: Entire Function That Agrees with Log on Natural Numbers?

mike-l wrote:I'm pretty sure you can find an entire function with prescribed values on any set that lacks limit points, but I can't find the theorem that says that.

In this context, is infinity a limit point of the natural numbers?

jestingrabbit
Factoids are just Datas that haven't grown up yet
Posts: 5967
Joined: Tue Nov 28, 2006 9:50 pm UTC
Location: Sydney

### Re: Entire Function That Agrees with Log on Natural Numbers?

No. Only limit points in the complex numbers count.
ameretrifle wrote:Magic space feudalism is therefore a viable idea.

skeptical scientist
closed-minded spiritualist
Posts: 6142
Joined: Tue Nov 28, 2006 6:09 am UTC
Location: San Francisco

### Re: Entire Function That Agrees with Log on Natural Numbers?

Suffusion of Yellow wrote:
mike-l wrote:I'm pretty sure you can find an entire function with prescribed values on any set that lacks limit points, but I can't find the theorem that says that.

In this context, is infinity a limit point of the natural numbers?

If you're talking about the Riemann sphere, then infinity is a limit point, and you can't necessarily find a function which is analytic on the Riemann sphere with given values on the natural numbers. But for this question, we're only interested in the function being analytic on C (an entire function), and the only possible obstruction would be a limit point in C.
I'm looking forward to the day when the SNES emulator on my computer works by emulating the elementary particles in an actual, physical box with Nintendo stamped on the side.

"With math, all things are possible." —Rebecca Watson