I don't know about fast, but... for example, using the definitions at
https://code.jsoftware.com/wiki/Essays/Extended_Precision_Functions

   0j100 ": 100 exp 1
2.7182818284590452353602874713526624977572470936999595749669676277240766303535475945713821785251664274

If I wanted fast, though, I'd probably do something along the lines of:

   htm=. gethttp 'https://www.math.utah.edu/~pa/math/e.html'
   (LF,' ')-.~({.~ '...' I.@E. ]) (}.~ '2.71828' I.@E. ]) htm
2.71828182845904523536028747135266249775724709369995957496696762772407663035354759457138217852516642742746639193200305992181741359662904357290033429526059563073813232862794349076323382988075319525101901157383418793070215408914993488416750924476146066808226...

(Except I'm sure there's better urls for this sort of approach.)

Probably not what you were looking for, though...

-- 
Raul

On Thu, Jul 25, 2019 at 11:04 AM Ulrich Vollert <[email protected]> wrote:
>
> Hello,
>
> I came across ’The Google Test’ by Eugene McDonnell 
> (https://www.jsoftware.com/papers/play211.htm 
> <https://www.jsoftware.com/papers/play211.htm>) and was wondering how to 
> calculate many digits of Euler’s number e.
>
> I found an (ancient) article "The calculation of e to many significant 
> digits" by A. H. J. Sale in The Computer Journal 11(2) · August 1968 
> (https://www.researchgate.net/publication/266281843_The_Calculation_of_e_to_Many_Significant_Digits
>  
> <https://www.researchgate.net/publication/266281843_The_Calculation_of_e_to_Many_Significant_Digits>)
>  which uses a clever algorithm implemented in Algol 60 (!) to calculate e 
> digit by digit and using only integer arithmetic.
>
> With this algorithm in J (see below)  I could calculate as many digits of e 
> as I want (I used 140 digits) and solved the Google Test which asks for:
>
> 1. First 10-digit prime found in consecutive digits of e.
>
> 2. The number F(5) which follows
>
> F(1)= 7182818284
> F(2)= 8182845904
> F(3)= 8747135266
> F(4)= 7427466391
> F(5)=__________
>
> (BTW In the article of Eugene McDonell the second number F(2) contains 
> transposed digits 9 and 0.)
>
> Here is my question:
>
> How would you calculate many significant digits of e in J?
>
> Regards,
> Ulrich
>
>
> NB. *** problem 1 ***
>
> NB. value to check m
> checkm =: 3 : 0
> r =. -: ^. 6.2831852 * y
> r + y * (^. y) - 1
> )
>
> NB. given the number of digits of e which are wanted,
> NB. calculate the number of required terms
> number_of_terms =: 3 : 0
> test =. 2.30258509 * >: y
> (>: ^: (test&>: @: checkm) ^:_) 4
> )
>
> NB. from JforC
> LoopWithInitial =: 2 : 'u&.>/\.&.(,&(<v))&.|.&.(<"_1)'
>
> NB. x is coeff and y is j, carry, follwed by the coeffs so far
> NB. answer are the next j and carry, the new coeff, and old coeffs
> term =: 4 : 0
> j =. {. y
> carry =. 1 { y
> coeffs =. 2 }. y
> temp =. carry + 10 * x
> (<: j), ((0, j) #: temp), coeffs
> )
>
> NB. initialise with m
> init =: 3 : 0
> term LoopWithInitial (y, 0) (<: y) $ 1
> )
>
> NB. given is an output of term LoopWithInitial
> NB. find next output
> next =: 3 : 0
> NB. take last row of output of term LoopWithInitial
> NB. the row contains: j carry coeffs
> last =. {: y
> NB. carry is the next digit of e, forget j
> ee =: ee , ": 1 { last
> term LoopWithInitial ((<: #last), 0) |. 2 }. last
> )
>
> NB. calculate the number of digits of e
> NB. the digits are collected in a global variable ee
> calc_e =: 3 : 0
> ee =: ''
> next^:y init number_of_terms y
> ee
> )
>
> digits_e =: ". 10 ]\ calc_e 140
>
> solution1 =: {. (I. 1 p: digits_e) { digits_e
> NB. -> 7427466391
>
> NB. *** problem 2 ***
>
> NB. cross sum of a number
> csum =: 3 : '+/ "."0 ": y'
>
> NB. the given numbers have all the same cross sum of 49
> NB. csum"0 (7182818284 8182845904 8747135266 7427466391) -> 49 49 49 49
>
> NB. find the next number in the digits of e with a cross sum of 49
> solution2 =: 4 }. (] {~ [: I. 49 = csum"0) digits_e
> NB. -> 5966290435
>
>
>
>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to