Stringification of FP numbers.

gam3 has asked for the wisdom of the Perl Monks concerning the following question:

Can anyone explain to me why 0.0001 is stringified as "0.0001" but 0.00001 is stringified as "1e-05"?

This would make sense if "100000.0" was printed as 1e05, but is seems negative exponents are treated in a different and rather arbitrary way.

print 0.0001, " ", 0.00001, "\n"
[download]

-- gam3
A picture is worth a thousand words, but takes 200K.

Comment on Stringification of FP numbers. Download Code

Replies are listed 'Best First'.
Re: Stringification of FP numbers. by ikegami (Patriarch) on Jun 12, 2009 at 15:57 UTC
If you're not happy with the default, you can easily control the format using [s]printf	[reply]
Re: Stringification of FP numbers. by JavaFan (Canon) on Jun 12, 2009 at 16:24 UTC
Well, people like to see 0.1 instead of 1e-1. But they prefer 1e-50 over 0.00000000000000000000000000000000000000000000000001. Saves space, and counting digits. Obviously, the more leading zero's the rather people see "scientific" notation. The happy medium (flip point) was set at 4 leading zeros.	[reply]
Re: Stringification of FP numbers. by ig (Vicar) on Jun 12, 2009 at 17:26 UTC
The conversion is usually done by gconvert(), gcvt() or printf with %g format, depending on what Configure determined and, for example, on CentOS5 the %g format is described as: The double argument is converted in style f or e (or F or E for G conversions). The precision specifies the number of significant digits. If the precision is missing, 6 digits are given; if the precision is zero, it is treated as 1. Style e is used if the exponent from its conversion is less than -4 or greater than or equal to the precision. Trailing zeros are removed from the fractional part of the result; a decimal point appears only if it is followed by at least one digit. The emphasis is mine. This is consistent with the result you are seeing.	[reply]
Re^2: Stringification of FP numbers. by gam3 (Curate) on Jun 14, 2009 at 18:51 UTC
I think that this is the answer I was looking for. It is dependent on the clib then if %g is being used. However %g in a C program gives this output: 0.0001 1e-05 100000 1e+06 I guess I'll have have to get the source code and look. Thanks. -- gam3 A picture is worth a thousand words, but takes 200K.	[reply]
Re: Stringification of FP numbers. by syphilis (Archbishop) on Jun 12, 2009 at 23:58 UTC
This would make sense if "100000.0" was printed as 1e05 Yes - I, too, would prefer to see the same rule applied to both leading and trailing zeroes. Unfortunately, that's not the way it's being done. Another way of avoiding the scientific notation (no matter how many leading zeroes) is to convert to a Math::BigFloat object: `perl -MMath::BigFloat -wle "print Math::BigFloat->new(1e-10)"` [download] (Replace double quotes with single quotes on nix.) Cheers, Rob	[reply] [d/l]

Back to Seekers of Perl Wisdom