Re: st: doubt on the output format %w.dg

[Nick Cox <njcoxstata@gmail.com>]

The results are certainly not random. People would have noticed and
complained long since if they were.

At this moment I am away from manuals and can not therefore check for
further discussion in the manuals. A key point is that the precise
algorithm is embedded in proprietary code. I don't recollect that a
fully explicit verbal equivalent has ever been included in the
manuals.

If you have ever tried to program a format that was adaptive to a
unpredictable variety of inputs, you will appreciate the difficulty of
doing that.

In my own programs I usually know what number of decimal places seems
sensible and use an %f format.

Nick

2011/3/12 Grace Jessie <gracejessie@hotmail.com>:
> NIck,
> I had read the -help format- file several times before I asked the question and the examples are all from that.
> I can understand %f very well, but %g is another thing.
> Does "with %g you give up most of the  control and let Stata decide" in your reply mean the result for %g is random?
> Or else,  what I want to know is how Stata dicides it.
> Additionally, in the Users' Guide, it is said that "The %w.0g format is a set of formatting rules that present
> the values in as readable a fashion as possible without sacrificing precision. The g format changes
> the number of decimal places displayed whenever it improves the readability of the current value."
> And it takes the followings for example which puzzle me also.
> . di %9.0g sqrt(2)
>  1.414214
> // The width of 1.414214 is 8. I think it should be 1.4142136 (with the width of 9) which gives less precision sacrifice.
>
> . di %11.0gc 23667902
>   23667902
> There are no commas. With regard to it, the explanation in the [D]Data Management is that "This number was too large for
> Stata to insert commas and still respect the current width of 11." Why the current width is 11? It is 8 certainly, isn't it?
> . di %12.0gc 23667902
>  23,667,902
> Together with the question in the first posting, I hope for any help.
> Thank you.
> Grace
>
>
> ----------------------------------------
>> Date: Sat, 12 Mar 2011 12:23:08 +0000
>> Subject: Re: st: doubt on the output format %w.dg
>> From: njcoxstata@gmail.com
>> To: statalist@hsphsun2.harvard.edu
>>
>> -help format- is the place to start. I think the main thing is not to
>> expect the rules for %g formats to be much like those for %f formats.
>> With %f, you keep most of the control; with %g you give up most of the
>> control and let Stata decide.
>>
>> Nick
>>
>> >> From: gracejessie@hotmail.com
>>
>> >> I can not understand the variables' output format %w.dg.
>> >> How does Stata leave up to the format the number of digits to be displayed to the right of the decimal point if d equals 0 and not more than d digits will be displayed if d!=0.
>> >> For example,
>> >> What does "12.0g" mean?
>> >> Why "di %12.0g 5231371222.139" or "di %12.1g 5231371222.139" equals "5231371222", not "5231371222.1"? (the width of "5231371222" is 10, not 12.)
>> >> Why "di %12.0g 0.0000029394" equals "2.93940e-06", not "2.9394e-06"?

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/