10:13 a.m.
Happy New Year!
I have rather large numbers (seed number, in float representation), saved with fixed
precision (although that is, I guess, not the problem). Here a screenshot from the file
opened with a text-editor (PSpad)
[cid:image001.png@01D3854C.3263B400]
When I import the data into gretl (from a tsv) and "show" the data, the
scientific format is chosen and distorts the true values
[cid:image002.png@01D3854C.696287B0]
the same "view" is presented if I print the data to the log:
[cid:image003.png@01D3854D.01E3D200]
Because same issues do not apply for numbers with large fractions, I guess it is a problem
with large integer (part) of the numbers? Is this only the visible behaviour or can it
really distort the data and computations? In this case it wouldn't matter, but in
other cases it might...
Best
Frederik
10:58 a.m.
Am 04.01.2018 um 11:13 schrieb Schaff, Frederik:
Happy New Year!
And to you! (and everybody)
There are two separate issues here, the internal number representation
and the printed format.
AFAIK in gretl basically everything is stored as a double precision
float (should be 64 bit == 8 byte wide). It is a standard computer
science thing that floating point numbers cannot exactly represent all
integers, and in principle you also have the possibility of overflow. In
this (very old and universal) sense yes in principle it could affect the
results. This is not usually a problem in practice, however.
The limited number of digits/precision which is displayed is different
from that. There are some 'set' options for that for example, quoted
from the command ref:
<doc>
- display_digits: an integer from 3 to 6, specifying the number of
significant digits to use when displaying regression coefficients and
standard errors (the default being 6). This setting can also be used to
limit the number of digits shown by the summary command; in this case
the default (and also the maximum) is 5, or 4 when the --simple option
is given.
- mwrite_g: on or off (the default). When writing a matrix to file as
text, gretl by default uses scientific notation with 18-digit precision,
hence ensuring that the stored values are a faithful representation of
the numbers in memory. When writing primary data with no more than 6
digits of precision it may be preferable to use %g format for a more
compact and human-readable file; you can make this switch via set
mwrite_g on.
</doc>
In the "show values" (zeige Werte) window there is also a button number
format (Zahlenformat) where a different display can be chosen.
hth,
sven
11:07 a.m.
Am 04.01.2018 um 11:13 schrieb Schaff, Frederik:
> Happy New Year!
And to you! (and everybody)
> guess it is a problem with large integer (part) of the numbers?
Is
> this only the visible behaviour or can it really distort the
data and
> computations? In this case it wouldn’t matter, but in other
cases it
> might…
There are two separate issues here, the internal number
representation and
the printed format.
AFAIK in gretl basically everything is stored as a double precision
float (should
be 64 bit == 8 byte wide). It is a standard computer science thing
that floating
point numbers cannot exactly represent all integers, and in principle
you also
have the possibility of overflow. In this (very old and universal)
sense yes in
principle it could affect the results. This is not usually a problem
in practice,
however.
The limited number of digits/precision which is displayed is
different from
that. There are some 'set' options for that for example,
quoted from the
command ref:
- display_digits: an integer from 3 to 6, specifying the number of
significant
digits to use when displaying regression coefficients and standard
errors (the
default being 6). This setting can also be used to limit the number
of digits
shown by the summary command; in this case the default (and also the
maximum) is 5, or 4 when the --simple option is given.
- mwrite_g: on or off (the default). When writing a matrix to file as
text, gretl
by default uses scientific notation with 18-digit precision, hence
ensuring that
the stored values are a faithful representation of the numbers in
memory.
When writing primary data with no more than 6 digits of precision it
may be
preferable to use %g format for a more compact and human-readable
file;
you can make this switch via set mwrite_g on.
> </doc In the "show values" (zeige Werte) window there is also a
button number
format (Zahlenformat) where a different display can be chosen.
hth,
sven
That trick did it, thanks!. For the record, selecting “reformat” (
[cid:image001.png@01D38554.89894C60] ) and then show (e.g.) the 15 significant figures
worked fine
Frederik
2622
days inactive
2622
days old
2 comments
2 participants
participants (2)
-
Schaff, Frederik -
Sven Schreiber