[Cfp-interest 2105] Subnormals in decimal
Mike Cowlishaw
mfc at speleotrove.com
Wed Aug 18 06:45:24 PDT 2021
As promised, I checked how I originally defined this for decimal numbers;
links to the full document are at:
http://speleotrove.com/decimal/#arithmetic
The most pertinent section is at:
http://speleotrove.com/decimal/damodel.html which states:
"In any context where exponents are bounded most finite numbers are normal.
Non-zero finite numbers whose adjusted exponents are greater than or equal
to Emin are called normal numbers; those non-zero numbers whose adjusted
exponents are less than Emin are called subnormal numbers."
[8] <http://speleotrove.com/decimal/damodel.html#backref.8> That is,
numbers whose absolute value is non-zero and is closer to zero than ten to
the power of Emin.
'Adjusted exponent' is defined earlier in that section, and is, in effect, a
normalized exponent but described independently of encodings etc.
This does avoid the use of the term 'normalized', which perhaps avoids the
entanglement we discussed yesterday.
Mike
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20210818/1891b942/attachment.htm>
More information about the Cfp-interest
mailing list