RE: Digest for EUforum at topica.com, issue 5330
> Subject: Re: math question
>
> posted by: Larry Miller <larrymiller at sasktel.net>
>
> An easy way to extract the low DWORD from an atom is to use the
remainder() > > function as in a = remainder(a, #100000000). Bitwise
functions such as and_bits()
are restricted to numbers #FFFFFFFF or less for both parameters so can not
be used for this purpose. A potential problem is that remainder() will
return a fractional part if the original number is not an integer. The use
of floor() or a rounding function may be necessary if this is a possibility.
>
> Larry Miller
True. The original question, hoxever was about getting the exact value of
the low dworsd of the product of two dwords.
An atom is not a 64-bit integer, but a float with 53 digit precision. To
check that, look for a discussion in this list about the greatest integer
arithmetically known by Euphoria - it's power(2,53)-1.
This means that, when taking the product, only the 53 most significant bits
will be retained - this is fine as long as the product lies below the 2^53
limit -, while exact accuracy aof the 32 less significant bits is desired.
The proposed simple solution just can't work as a result, since the 11 LSB
can be lost when converting an atom to a 64-bit integer - more precisely
max(0,MSB(product)-53). The MSB of the product is the sum of the %MSBs plus
perhaps 1.
CChris
|
Not Categorized, Please Help
|
|
