I think you will be able to find more information from Hewlett Packard.
See, for example
http://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1979-12.pdf
Also, I think you will find other material on wkahan's home page.
As far as the algorithms for bigfloats in Maxima, they are generally
inappropriate
for calculators because they are "unrestricted" in the sense used by
R.P. Brent.
They provide algorithms for any precision. Calculators have some fixed
precision,
e.g. 10 decimal digits. (or 13, with guard digits etc.) So quite
different routines
are used.
RJF
On 2/15/2012 9:03 AM, Michel Talon wrote:
> Jan Mueller wrote:
>> Hi,
>> I am just writing an article about how pocket calculators compute values of
> trigonometric functions (especially Volders CORDIC algorithm). I also
> contacted Casio and TI about it but they gave no information because of
> company secrets. Could anyone explain to me how Maxima solves it? sin would be
> sufficient for me
>> Jan
> Apparently the software for the HP 48 calculator has been written by the
> french mathematician B. Parisse from Grenoble, and can be downloaded here:
> ftp://ftp-fourier.ujf-grenoble.fr/linux/xcas/hpcas/erablsrc.zip
> so you can take a look at it. For the TI calculator it is the software Derive,
> i don't think one can get the source code.
>
> Another source for access to trig functions computations is to take a look at
> the source code for the libc and libm in free operating systems. For example
> for FreeBSD you can take a look at
> http://www.freebsd.org/cgi/cvsweb.cgi/src/lib/msun/
> but the code for i387 obviously uses hardware nstructions in the processor so
> it will ne be very useful to you. But there is a generic computation, for
> example for the sin here:
> http://www.freebsd.org/cgi/cvsweb.cgi/src/lib/msun/src/k_sin.c?rev=1.11
> which shows how one can proceed with approximate formula.
>
>