[seqfan] Re: Implementing first 500 sequences in Java

[Charles Greathouse]

As best I can tell
http://developer.classpath.org/doc/java/math/BigInteger-source.html#line.629
http://www.koders.com/java/fidFBDCC33194ED039FB11131755CE7735B1FC0608C.aspx?s=261#L781
BigInteger only implements naive algorithms: grade-schools
multiplication, basic RTL binary exponentiation, etc.  No Karatsuba or
FFT-based multiplication, no Jebelean exact division, no sliding
window exponentiation, no subquadratic GCD.

So it's not going to be fast, but the results should be trustworthy.

Charles Greathouse
Analyst/Programmer
Case Western Reserve University

On Fri, Oct 29, 2010 at 6:02 AM, Richard Mathar
<mathar at strw.leidenuniv.nl> wrote:
>
>> From: "N. J. A. Sloane"
>> To: seqfan at list.seqfan.eu
>> Subject: [seqfan] Re: Implementing first 500 sequences in Java
>>
>> Does Java do exact arithmetic on large numbers?
>> Can the results of these programs be trusted?
>
> Java has "ideal" support to arbitrarily large numbers with
> the BigInteger class. Unlike C/C++ or F90 etc, this is a very
> natural reference library for these tasks. I've used this a lot
> in the FI interpreter http://www.strw.leidenuniv.nl/~mathar/progs/FI.html which uses
> mY API of http://www.strw.leidenuniv.nl/~mathar/progs/jdocs .
> There seems to be no standard NumberTheory library on top of that,
> so one has basically start from scratch to "do" prime numbers,
> rationals etc.
>
> The functionality that is available in the base library is
> documented in http://download.oracle.com/javase/7/docs/api/ (search
> for BigInteger in the lower left panel). Basically this is +, -, *,
> the integer-with-remainder,  power to an integer and some bit manipulations.
>
> RJM