My Math Forum  

Go Back   My Math Forum > College Math Forum > Real Analysis

Real Analysis Real Analysis Math Forum


Reply
 
LinkBack Thread Tools Display Modes
May 12th, 2010, 07:09 PM   #1
Senior Member
 
Joined: Jul 2008
From: Western Canada

Posts: 4,031
Thanks: 39

Looking for a database of constants

Reading through old scientific papers, I frequently find formulae containing numerical constants, usually given only to 4 or 5 decimal places, and no explanation of from where they came. Often they are the result of summing a series, and the value will be some simple relationship to a fundamental constant such as pi, or else the log of some integer or simple fraction. At the moment for example, I'm trying to figure out if the value -0.60835 is derived from some simple constant. So, I'm just wondering if there's an online database somewhere where one could search for these constants.
Yooklid is offline  
 
May 12th, 2010, 08:28 PM   #2
Global Moderator
 
CRGreathouse's Avatar
 
Joined: Nov 2006
From: UTC -5

Posts: 16,046
Thanks: 937

Math Focus: Number theory, computational mathematics, combinatorics, FOM, symbolic logic, TCS, algorithms
Re: Looking for a database of constants

Yes! Neil Sloane's On-Line Encyclopedia of Integer Sequences:
http://oeis.org/classic/

If you have any trouble finding a constant there, feel free to post and I'll help you. (I'm involved with that project.) Other possibilities include:
RIES http://mrob.com/pub/ries/
Plouffe's Inverter: http://pi.lacim.uqam.ca/

If the value is an obscure infinite product or special constant, essentially your only hope is that it is in the OEIS; the others won't have it. If it's a 'simple' combination of fundamental constants, but not very well-known, RIES may be best at recognizing it. I haven't found much use for Plouffe's Inverter, and his newer Inverse Symbolic Calculator seems to have been lost to the web. (It still wasn't generally as good as the other tools, though...)
CRGreathouse is offline  
May 13th, 2010, 09:23 PM   #3
Senior Member
 
Joined: Jul 2008
From: Western Canada

Posts: 4,031
Thanks: 39

Re: Looking for a database of constants

Thank you very much.

I looked at Neil Sloane's site, but wasn't able to figure out how it would apply to my current situation.

I did try both Plouffe's Inverter, and RIES, both of which came up with possible solutions, which I will now have to work through.

Meanwhile, I chopped up my current problem up into four smaller pieces, which I hope will be easier to solve.
I now need to find the the following sum:



where H(k) is the kth Harmonic number, and a and b are constants, which for the four different parts of my problem, take on the following sets of values:
a=1, b=-1
a=2, b=-1
a=1, b=+1
a=2, b=+1

Hence, I need to evaluate these:

, , ,

Any idea whether these have been solved previously?
Yooklid is offline  
May 14th, 2010, 06:18 AM   #4
Global Moderator
 
CRGreathouse's Avatar
 
Joined: Nov 2006
From: UTC -5

Posts: 16,046
Thanks: 937

Math Focus: Number theory, computational mathematics, combinatorics, FOM, symbolic logic, TCS, algorithms
Re: Looking for a database of constants

Quote:
Originally Posted by Yooklid
I looked at Neil Sloane's site, but wasn't able to figure out how it would apply to my current situation.
Enter the constant into the search bar. For pi, type 3.1415926. This will actually search for sequences with 3,1,4,1,5,9,2,6 but as you can see it pulls up the right sequence without any real trouble.

There are a *lot* of constants there, over 5000. Random obscure example: the Khinchin mean . The others wouldn't be able to pull something like that up.

Quote:
Originally Posted by Yooklid
I did try both Plouffe's Inverter, and RIES, both of which came up with possible solutions, which I will now have to work through.

Meanwhile, I chopped up my current problem up into four smaller pieces, which I hope will be easier to solve.
I now need to find the the following sum:



where H(k) is the kth Harmonic number, and a and b are constants, which for the four different parts of my problem, take on the following sets of values:
a=1, b=-1
a=2, b=-1
a=1, b=+1
a=2, b=+1

Hence, I need to evaluate these:

, , ,

Any idea whether these have been solved previously?
Stab in the dark, but is the first one zeta(3)? I certainly can't prove it, but numerically it seems plausible. I calculated the number to 11 digits (using an integral to estimate the terms I can't sum to) and it matches:
http://oeis.org/classic/?q=1.20205690315

I calculate the third one as approximately 3.049047873156, but I can't find any simple form for it. I haven't tried the other two yet.
CRGreathouse is offline  
May 14th, 2010, 06:31 AM   #5
Senior Member
 
Joined: Apr 2010

Posts: 215
Thanks: 0

Re: Looking for a database of constants

Wolfram Mathematica says:







brangelito is offline  
May 14th, 2010, 06:47 AM   #6
Global Moderator
 
CRGreathouse's Avatar
 
Joined: Nov 2006
From: UTC -5

Posts: 16,046
Thanks: 937

Math Focus: Number theory, computational mathematics, combinatorics, FOM, symbolic logic, TCS, algorithms
Re: Looking for a database of constants

Impressive. I think that these might be provable by hand by expanding the H(n) and rewriting in terms of what is divided by each n^2, but I haven't tried yet. (I was about to, but then you posted with the answers...)

Code:
S(lim)=s=0;sum(n=1,lim,s+=1./n;s/(n+1)^2)+(1+Euler)/lim+log(lim)/lim
CRGreathouse is offline  
May 14th, 2010, 04:21 PM   #7
Senior Member
 
Joined: Jul 2008
From: Western Canada

Posts: 4,031
Thanks: 39

Re: Looking for a database of constants

CRGreathouse,
Thank you for the explanation on looking up constants on OEIS. This will be very useful.

Brangelito,
Thank you for finding these solutions. Now, I'm embarassed to confess that I made a transcription error on the second and forth summations. They should really have been:



and



Could I possibly twist your arm, and have you look these up too?


After my post last night, I wrote a short program to calculate these, and let it run overnight. It's calculating to 100 digits, and isn't terribly fast. It is still running, and is now up to about 21,000,000 terms. The sums, so far are only accurate to about 6 figures. I was hoping that it might generate enough terms to for OEIS to recognize it. Anyway that appears to be moot now.

Out of interest, I also included the following sum which isn't part of my problem, but seems to be the simplest case:


and appears (with only 6 significant figures so far) that it may be equal to

I thought that this last one might be simple enough for me to attempt a hand solution, but when I started expanding the terms, I could find no discernible pattern.

I'd be fascinated to learn what process Mathematica used to find these results.
Yooklid is offline  
May 15th, 2010, 03:48 AM   #8
Senior Member
 
Joined: Apr 2010

Posts: 215
Thanks: 0

Re: Looking for a database of constants

Mathematica had some problems this time around, it seems. Anyway, here's a picture of the output (I'm not sure how to decipher it):



Quote:
Out of interest, I also included the following sum which isn't part of my problem, but seems to be the simplest case:


and appears (with only 6 significant figures so far) that it may be equal to
This is correct!:)
brangelito is offline  
May 16th, 2010, 01:34 AM   #9
Senior Member
 
Joined: Jul 2008
From: Western Canada

Posts: 4,031
Thanks: 39

Re: Looking for a database of constants

Thanks Brangelito.
I'm afraid that I don't know what the H(2n-2) case means either. However, it gives me an idea how to proceed. Since we now know the value of the H(2n-1) sum, and the H(2n-2) case would differ only by a (1/n) in each term, the overall difference would be the following series:




Then using partial fractions it should be simple to split this into:

<<Half asleep ramblings edited out. See following post>>

It's too late for me to think clearly right now. I'll look at this again in the morning when I'm actually awake.
Yooklid is offline  
May 16th, 2010, 02:38 PM   #10
Senior Member
 
Joined: Jul 2008
From: Western Canada

Posts: 4,031
Thanks: 39

Re: Looking for a database of constants

The partial fraction expansion wasn't quite as simple as what I thought in my last post, but still reasonable. So, I worked though the simple series and found:



So combining that with the H(2n-1) series gives:



or approx. 1.576973

This result agrees with the value my program has now calculated up to 7 figures so far (with 59,000,000 terms calculated).
Yooklid is offline  
Reply

  My Math Forum > College Math Forum > Real Analysis

Tags
constants, database



Thread Tools
Display Modes


Similar Threads
Thread Thread Starter Forum Replies Last Post
database, composite key ungeheuer Computer Science 1 May 29th, 2013 08:11 AM
Database of Number Correlations etc virtue121 New Users 11 July 23rd, 2009 05:56 PM
Access database CRGreathouse Computer Science 2 June 22nd, 2009 09:04 PM





Copyright © 2018 My Math Forum. All rights reserved.