## Is this in any way interesting?

For the discussion of math. Duh.

Moderators: gmalivuk, Moderators General, Prelates

### Is this in any way interesting?

I know it's most likely isn't, but I would still like to understand why/why not

(pi's value to n digits - pie's value to n-1 digits)/pi's value to n-2 digits = 1/constant
the same principle applies to e

i'm just curious if there is some way to represent each next digit of pi in a simple way, so that i can represent PI as
f(PI to n digits) = PI to n+1 digits

there's no reason why I want to do this, i'm just curious

Dmitry

Posts: 19
Joined: Mon May 07, 2012 12:44 am UTC

### Re: Is this in any way interesting?

Dmitry wrote:I know it's most likely isn't, but I would still like to understand why/why not

(pi's value to n digits - pi's value to n-1 digits)/pi's value to n-2 digits = 1/constant
the same principle applies to e

Huh? I guess you're treating the strings of digits as big integers... Ok, that's not particularly interesting, since you get a similar result for a random string of digits, not just the digits of pi or e. The result is not exactly a constant, though. I haven't played with the algebra yet, I just tested it with a quick program, so I can't give you an explanation for this effect yet. But your 1/constant is approximately 90 plus an "error" term of the order of 10^(2-n). Note that 90 = 10² - 10, so this effect is probably just some simple artifact of decimal representation.

Dmitry wrote:i'm just curious if there is some way to represent each next digit of pi in a simple way, so that i can represent PI as
f(PI to n digits) = PI to n+1 digits

there's no reason why I want to do this, i'm just curious

No, there's no simple formula for the decimal digits of pi, but there are compact spigot algorithms that can be used to determine digits of pi in bases which are binary powers.

PM 2Ring

Posts: 2622
Joined: Mon Jan 26, 2009 3:19 pm UTC
Location: Mid north coast, NSW, Australia

### Re: Is this in any way interesting?

Dmitry wrote:(pi's value to n digits - pie's value to n-1 digits)/pi's value to n-2 digits = 1/constant
the same principle applies to e

How did you arrive at this conclusion? I only ask because it doesn't seem to be true:

(3.1415 - 3.141) / 3.14 = 0.000159235669
(3.14159 - 3.1415) / 3.141 = 0.0000286532951

scarecrovv

Posts: 588
Joined: Wed Jul 30, 2008 4:09 pm UTC
Location: Massachusetts

### Re: Is this in any way interesting?

Also, isn't 1/constant just a constant itself? How do you figure it's 1 over something?

Dopefish

Posts: 702
Joined: Sun Sep 20, 2009 5:46 am UTC
Location: The Well of Wishes

### Re: Is this in any way interesting?

We're looking at expressions of the form
(314159 - 31415) / 3141 ~= 90.017191977

Let A be the integer formed from the first n-2 decimal places of pi (or whatever). If the next two digits are b and c, then the integer formed from the first n-1 decimal places is 10A + b, and the integer formed from the first n decimal places is 100A + 10b + c. In the above example A = 3141, b = 5, c = 9.

So the expression becomes
((100A + 10b + c) - (10A + b)) / A
= (90A + 9b + c) / A
= 90 + (9b + c) / A
For large A, (9b + c) / A is much smaller than 0 1.

Edit: Oops! Thanks, mike-l.
Last edited by PM 2Ring on Wed May 09, 2012 7:47 am UTC, edited 2 times in total.

PM 2Ring

Posts: 2622
Joined: Mon Jan 26, 2009 3:19 pm UTC
Location: Mid north coast, NSW, Australia

### Re: Is this in any way interesting?

((3.14-3.1)/3)^-1 = 75
((3.141-3.14)/3.1)^-1 = 3100
((3.14159265 - 3.1415926)/3.141592)^-1=62831840

that's what i meant

and e...
((2.71-2.7)/2)^-1 = 200
((2.718-2.71)/2.7)^-1 = 337.5, guess it doesn't work here O_o
((2.71828183-2.7182818)/2.718281)^-1 = also doesn't quite work.
so i guess it doesn't quite follow for e...

Dmitry

Posts: 19
Joined: Mon May 07, 2012 12:44 am UTC

### Re: Is this in any way interesting?

Ah, what you meant isn't that you get a constant, but that you get an integer.

Let's rearrange this:
Dmitry wrote:((3.14159265 - 3.1415926)/3.141592)^-1=62831840

First actually do that inversion from the start, giving this:
3.141592/(3.14159265 - 3.1415926)=62831840

Next multiply top/bottom of the fraction by the same power of ten to get rid of the decimal points:
314159200/(314159265 - 314159260)=62831840

Now rewrite it as the general case:
(Number formed by the first n-2 digits followed by 00) / (the nth digit) = an integer(?)

If the nth digit is 1, 2, 4, or 5, then it will automatically work there because any number ending in 00 is a multiple of 100, and so will also be a multiple of 1, 2, 4, or 5.

If the nth digit is 3, 6, 7, 8, or 9 then it will not always work, and it will generally just be coincidence if it does. With pi, it is just coincidence that 3141 is divisible by 9. However at that 6 you get
31415900/6 = 5235983.3333333333333333333333333
and at the next 3 you get
3141592600/3 = 1047197533.3333333333333333333333
so it doesn't always work for pi.

[Edited for typo]
Last edited by jaap on Tue May 08, 2012 12:14 pm UTC, edited 1 time in total.

jaap

Posts: 1731
Joined: Fri Jul 06, 2007 7:06 am UTC

### Re: Is this in any way interesting?

Alright that explains it, thanks
Dmitry

Posts: 19
Joined: Mon May 07, 2012 12:44 am UTC

### Re: Is this in any way interesting?

PM 2Ring wrote:For large A, (9b + c) / A is much smaller than 0.

What?
addams wrote:This forum has some very well educated people typing away in loops with Sourmilk. He is a lucky Sourmilk.
mike-l

Posts: 2534
Joined: Tue Sep 04, 2007 2:16 am UTC