Post Reply 
Accurate Bernoulli numbers on the 41C, or "how close can you get"?
03-10-2014, 08:23 PM
Post: #3
RE: Accurate Bernoulli numbers on the 41C, or "how close can you get"?
(03-10-2014 12:20 AM)Thomas Klemm Wrote:  Did you ever consider to write the program in MCODE? That provides the possibility to use 13 digits for your calculations and only round the result to 10 digits.

Getting 10 valid digits out of 13 digits working precision is not quite a challenge. ;-)

The basic question here was: how close can you get without any additional guard digits. Otherwise things are trivial: I tried the basic formula on a 12-digit 35s. Even with sloppy programming 10 valid digits ±1 ULP are not very hard to achieve. So with 13 digit precision things become trivial if the goal is merely a correct 10-digit result.

But nevertheless I would love to try MCODE programming. Years ago I always enjoyed combining efficient x86 assembler code with the beauty of Pascal. But honestly, I do not know the slightest thing about MCODE. I do not even how what is required to build a working programming environment. Maybe there is some basic info somewhere to start with?!

BTW, in the meantime there is a shorter and even slightly faster Bernoulli program with comparable accuracy: ±1 ULP over the whole working range. Yes, with a standard 10-digit 41C. ;-)

Dieter
Find all posts by this user
Quote this message in a reply
Post Reply 


Messages In This Thread
RE: Accurate Bernoulli numbers on the 41C, or "how close can you get"? - Dieter - 03-10-2014 08:23 PM



User(s) browsing this thread: 1 Guest(s)