Accurate Bernoulli numbers on the 41C, or "how close can you get"?

[Dieter]

(03-12-2014 04:34 PM)Thomas Klemm Wrote:  You'll find all you need on Warren Furlow's HP-41 web-site.

(03-14-2014 10:05 AM)Ángel Martin Wrote:  It's slightly more involved than just applying a cookie-cutter recipy, even if it's all published somewhere (like TOS).

Yes, I am absolutely sure it is all "somewhere". ;-) But what I am looking for is something like a structured introduction to MCODE programming: what are the hardware options, what else is required, what is covered by the instruction set and what are basic programming techniques. I am quite familiar with FOCAL programming, and I also appreciate the the benefits of doing things in machine code, so MCODE sounds like an interesting new project to me. However, I need some basic info first.

Quote:The 13-digit routines allow for intermediate calculations done with more precision, so the final result is more accurate (or less inaccurate, in a more strict speak). There are many 13-digit versions of the standard math routines in the 41 OS, a good starting point could be the mini-paper I prepared a while ago; available at TOS (can't paste the link here as you probably know).
I've attached it to this post as well (hope it works)

Thank you very much, and yes, it worked. That's exactly what I am looking for. In a way I am a kind of "accuracy junkie" - that's why I love the 34s project and the option of 34-digit precision for certain things. Defining special functions like the Normal distribution quantile, Lambert's W or just the Bernoulli numbers will full 10-digit accuracy is virtually impossible in FOCAL. With 13 digits all this becomes easy.

Now - where and how to start?

Dieter

By the way, do you know of a kind of overview with the actual accuracy of the 41's internal 13-digit routines?