The DM virtual file syntax lets you pick out selected columns
from a table.
dmcopy "evt.fits[cols time,pha]" tmp1.fits
makes a file containing only the columns time and pha,
whereas
dmstat "evt.fits[cols energy]"
passes to dmstat a virtual file containing only the column
called "energy".
dmcopy "pi.fits[cols rate=count_rate]" npi.fits
renames the column count_rate in pi.fits
to be simply rate in the file npi.fits.
In this example only the count_rate
column is copied; to copy across the other
columns you can either specify them individually
or use the symbol "*":
dmcopy "pi.fits[cols rate=count_rate,*]" npi2.fits
The cols directive can be used to reorder the columns in a file,
since they will come out in the order you specify.
In the previous example - which used
- the rate column is the first column in the output file,
whatever its position in the input file.
Note that if you have a vector column sky(x,y), then
dmcopy "evt.fits[cols x,y]" table.fits
will retain the information that (x,y) is a vector column
called sky, but any of
will not, and the vector-dependent
coordinate systems like RA and Dec will be lost.
The command
dmstat "evt.fits[cols #3]"
passes to dmstat a file which contains only the third column
in evt.fits (where `column' is used in the DM sense, in which
sky(x,y) is a single "vector" column; use
to find the column numbers).
You can also exclude columns; both the following commands
remove the status column from the file, leaving the
other columns intact.
dmcopy "evt.fits[cols -status]" tmp1.fits
dmcopy "evt.fits[cols !status]" tmp1.fits
When used from the Unix shell the "!" symbol needs to
be written as "\!", which makes the first form more
convenient.
A table can contain virtual columns; these are
columns whose values are calculated by applying a
mathematical transform to an existing column.
They are mainly used to represent coordinate
systems such as those described in "ahelp coords".
For a Chandra event file, the sky(x,y) column
contains a coordinate transform to create the
eqpos(ra,dec) column - which gives the RA and Dec
of each event. These values can be listed by
using either of the following two commands:
dmlist "evt.fits[cols ra,dec]" data
dmlist "evt.fits[cols eqpos]" data
You can find the names of the coordinate
systems defined for a file by using the 'cols' option to dmlist:
and then using these coordinate system names in a DM
qualifier lets you turn them into actual columns.
It is also possible to filter and bin on virtual columns.
For example:
dmlist "evt.fits[ra=233.5:233.7,dec=23.4:23.6]" data
dmcopy "evt.fits[bin eqpos=0.01]" img.fits
However, filtering and binning do not work correctly with
virtual columns derived from non-monotonic coordinate
transforms (e.g. MSC(THETA,PHI), or EQPOS near the poles; see
'ahelp coords' for more information on these coordinate
systems). Hence, the results of filtering and binning in these
cases are not reliable.
Also, world-coordinate region filters on eqpos, such as
[eqpos=circle(15:34:55.534,+23:29:57.21,3.9032')]
are not handled as one might expect. When the region filter is
applied, eqpos is treated as a vector on R^2 (the plane). For
a circular region, a point (ra,dec) passes the filter if it
satisfies the equation
sqrt( (ra - ra0)^2 + (dec - dec0)^2 ) ) < radius
However, what one most likely wants in this case is for eqpos
to be treated as a vector on S^2 (the sphere), with a point passing
the filter if the spherical arc distance between it and
(ra0,dec0) is less than the radius. Currently, the best
approximation to this can be obtained by applying the region
filter to sky rather than eqpos, e.g.
[sky=circle(15:34:55.534,+23:29:57.21,3.9032')]
Hence, this alternative should be preferred over region filtering
on eqpos in most cases.
![[CIAO Logo]](../imgs/ciao_logo_navbar.gif){kind=logo}