iraf.net

On Thu, 26 Jun 2003 20:35:56 +0800
cd <cding@center.shao.ac.cn> wrote:> Dear Phil,
>
> I'm using the task DAOPHOT to reduce the m22 data now. There're four
> wfpc2 image of m22 in 1994: u28n0201t u28n0202t u28n0203t
> u28n0204t
>
> I got the star list successfully from 1-3 image but when I use the
> last programe: ALLSTAR on u28n0204t, the programe display a error
> information : ERROR: floating point invalid operation
>
> I don't know why it occured and how to deal with it . Could you help
> me? Thank you!
>
> Best Regards,
> Chen Ding
Hi Ding,I apologize for the delayed response. I suspect the error you are
having is related to the photometry file for that image. You may wish
to verify that your photfile does not contain unacceptable values, i.e.,
NULL or INDEF values.If you still have problems, you could send me the files that are used by
ALLSTAR. If you do send me the files, please include a header of the
image, the photfile, as well as the output form "lpar allstar".I hope that helps. Let us know if you have further questions or
problems.Cheers,Phil

The following is a forwarded message from my conversation with Chen
Ding. He was receiving a floating point exception when running ALLSTAR.
After a few email exchanges, I retrieved the data from his FTP site, and
attempted to reproduce the error. However, it was without success. In
my reply, I mentioned that I used his parameters, and the defaults (for
which he did not provide values). His new attempt was successful, in
that he did not receive the error. Hence, it appears that a setting in
the parameters was causing problems. Unfortunately I do not know which
parameter and setting was causing the problem.Begin forwarded message:Date: Thu, 3 Jul 2003 21:30:25 +0800
From: cd <cding@center.shao.ac.cn>
To: Phil Warner <pwarner@noao.edu>
Subject: Re: for help
Hello Phillip, I do the reduction again just unlearn some parameter according to
you(datapar daopar daofind,etc), and finally do the allstar
successfully! (without change other value) Thank you very much!(though I
sill don't know why the float exception occurred)... Thank you again , since you help me so much. All the best of you!

Chen & Ding

On Mon, 7 Jul 2003 11:36:13 +0800
"Chen" <cding@center.shao.ac.cn> wrote:> Dear Philip,
> Thanks for your so much help!
> I wanna to draw and overlap the outline of two epoch images(wfpc2) on
> a Ra-Dec table to see how much two epoch overlap, but I wonder how
> to fix the direction of the image, could you help me? Thank you very
> much!
> Then you wrote:> Sorry the "direction" may cause ambiguous, that means the how much
> rotation the image or angel, how can I get the value of it and how to
> use it to draw the outline of the picture(like the attachment of this
> mail), thank you!
> All the Best,
> Ding
>
Hi Ding,You can use the wcsmap (images.immatch) to get the required transform to
map one image onto another, provided your images have the appropriate
WCS in the headers. To get the angles, change the fitgeom parameter to
rscale or rxyscale. Please see the help documentation for wcsmap.I hope that helps.Cheers,Phillip

On Tue, 8 Jul 2003 09:16:49 +0800
cd <cding@center.shao.ac.cn> wrote:>
> Dear Philip,
>
> Thanks for your instruction, but maybe my ambiguous description
> mislead you. I don't want to transform the image but just wanna draw
> the outline of the two image on a table or coordinate to see how much
> these two image overlap, could u tell me how to draw or use which
> parameter? Thank you!
>
> Best Regards,
> Ding
Hi Ding,You can still do this with WCSMAP. WCSMAP prints out the transformation
of one image onto another. You can then use these values (e.g.,
transformation angles) to draw the first image with respect to the
other.As stated in the other email, you can obtain the angles with WCSMAP by
changing the "fitgeom" parameter to rscale to rxyscale. Doing so will
print the transformation (rotation) to the screen. WCSMAP is specifically used to determine the transformation equation
when registering images. However, I believe the equation (angles, in
your case) that you can obtain from this task will assist you in
determining the angle of one of your images with respect to another, and
therefore help you draw your images on a RA/Dec table.I hope that helps. Let us know if you have further questions or
problems.Cheers,Phillip

Hi Ding,I have another suggestion which may be useful for what you are trying to
do. If your images have reasonable WCS, what you can do is convert the
pixel coordinates of the image corners to RA/DEC and then use a task like
GRAPH to plot them. The task to use is WCSCTRAN to convert coordinates.
Here is a sketch of what you might do.Create a text file with the pixel coordinates of the corners. If both images
are the same size then you can use the same file.cl> type corners
1 1
1 1024
1024 1024
1024 1
1 1Note they should go in order around the image and the last point should
repeat.Now convert to RA/DEC:cl> wcsctran corners radec1 <image1> logical world verbose-
cl> wcsctran corners radec2 <image2> logical world verbose-If you want RA in hours instead of degrees you would need to set the
format parameter to something like "%.2H %.1h". The H is for the RA
and the h is for DEC and you need to specify them in the order in which
they are set in the header.Finally you can graph two boxes with:cl> graph radec1,radec2Note that GRAPH overplots when there are multiple text files. You might
want to fix the graph limits. Also the graph will have decimal labels
and it is not possible to use xxxx for the labels in this task.Yours,
Frank Valdes

On Mon, 14 Jul 2003 11:39:10 +0800
cd <cding@center.shao.ac.cn> wrote:> Hi, Phillip
>
> I've used the WCSMAP following your instruction and can get the
> rotation angle, however, if two image have large position shift, the
> task tell me the shift in x or y may 2000pixel or 2500pixel large
> than the 800pixel which is the size of the wfpc2 ccd. I don't know if
> the 2000pixel the task display is true in real distance or is the
> error of the data exccess. Shoud I use the WMOSAIC firstly to merger
> the four ccd into a big image and then use the wcsmap to transform the
> two big image? Another question: If I have got the star position of
> individual ccd of wfpc2 in pixel values how should I merger these four
> ccd star list to a single frame(maybe just add 800pixel to another
> ccd?). Because some star in wf1of image1 maybe in wf2 of image2, if
> merge into a single big image the compare may be easier)
> I'm eager to get your help, thank you!
> All the Best,
> Ding
>
Hi Ding,The parameters xshift and yshift are the shift between the origin of
each image, i.e., they indicate the amount of shift between the two
images, with respect to the origin. You can use WCSXYMATCH to obtain a
list containing the x and y shift, and includes the RA/Dec at several
points.You can merge your images into one big mosaic, if you like. Finding the
rotation of one large image may be less cumbersome than finding the
rotation of four. However, merging may take more time than finding the
rotation of the four images. Hence, the method is up to you.Concerning your other question, if you use WCSXYMATCH, you can determine
the shift at a particular point by giving the task a coordinate file
that contains the RA/Dec values for the star you want to match. Please
see the wcsxymatch help for details.I hope that helps. Let us know if you have further questions or
problems.Cheers,Phillip

On Fri, 18 Jul 2003 10:50:24 +0800
cd <cding@center.shao.ac.cn> wrote:> Hi,Philip,
> I draw many image or contour under iraf in irafterm, especially use
> graph task. However I wanna get the *.ps image of these plot. Would
> you please to tell me how to convert there irafterm plot into .ps
> file? Thank you very much!
>
> All the Best,
> Ding
>
Hi Ding,First, I will give you the standard reply regarding creation of eps
files from an interactive plot window.
========================================================================IRAF graphics can be saved as either Encapsulated PostScript by
using the 'eps' device name (e.g. ":.snap eps") to produce a sgiXXXX.eps
file (where XXXX is a process id number) in the current directory, or as
a GIF file using the 'g-gif' device name to produce an sgiXXXX.gif file.
In both cases, the plot will not contain any color information such as
you see on the screen due to the metacode translation. If you need this
color, you need to install the STSDAS package and can use the 'psidump'
device which uses a different kernel (let me know if you need more
information on this), or you could use a screen capture program and just
crop the image.========================================================================
Second, if you want to use graph to create a ps figure, type the
following into the cl:cl> graph <data_file> device=epsf
cl> .gflushThis will create an sgiXXXX.eps file, again where XXXX is a process id
number.I hope that helps. Let us know if you have further questions or
problems.Cheers,--Phillip

On Wed, 23 Jul 2003 11:23:57 +0800
"Chen" <cding@center.shao.ac.cn> wrote:>
>
>
>
> Hi,Phillip,
> Sorry for my ambiguous expression.
> I've plot the two or more curve on a plot with many parameter and
> label successfully except that I can't add the name of each curve on
> the plot to distinguish each curve (I don't wanna take the color or
> line type to make difference). Can the graph or sgraph do this? Thank
> you! All the Best,
> Ding
> Hi Ding, I think I understand now. As far as I know, there is no way to do this
using IRAF. You can modify the plot by editing the postscript, which is
not easy. However, I did find that sgraph will use different line styles for your
curves. Maybe this is an option?I will ask around tomorrow concerning the curve labeling. If I find an
answer, I will let you know.Sorry I could not be more helpful.Regards,Phillip
On Wed, 23 Jul 2003 10:13:36 +0800
"Chen" < cding@center.shao.ac.cn> wrote:>> Dear Phillip,
>> I've get the .ps image successfully under your esteemed instruction.
>> Thank you very much. Through the help files of graph and sgraph, I
>can> plot and label the image completelly except that I can't add the
>name> of each file of multiply images on the last image(though the
>dotted> line or the color can distinguish, the name on the image of
>different> line maybe more direct). Could you tell me how to del with
>it? Thank> you! Best Wishes,
>> Ding
>>
Hi Ding,I apologize for my delayed response. I did receive the email you sent
two days ago as well, and again, I apoligize for not responding then.I am not quite sure what you would like to do. Do you want to add the
name of the image to the plot, and you were not successful? Could you
sned me a postscript image of your plot (or a screenshot), indicating
what you would like to add, and where?I look forward to hearing from you soon.--Phil

On Wed, 23 Jul 2003 15:40:37 +0800
"Chen" <cding@center.shao.ac.cn> wrote:> Hi, Phil
> So quick to recieve your mail! If u r in America, I guess you must be
> very late this night.^_^
> I really can't thank you enough for your so much help! I hope u r not
> getting bored to see my so many silly questions!^_^ As to the plot
> tools, if it do can't label for the each curve, maybe some other task
> (IGI?) can do? I've looked you as my instructor and friend, and your
> every mail always give me excitation and confidence. Thanks for your
> kindness! All the best,
> Ding
> Hi Ding,I found an answer to your question. After executing graph or sgraph,
you can type "=gcur" in the cl to get cursor mode. Move the mouse to
where you would like to add the label, the type "T" which will then ask
you for a text label. Type in your text label, then <enter>.After entering the labels, you can print to an EPS file by using the
colon command ":.snap epsf". Be sure you type ".gflush" to print the
EPS to a file.I hope that helps. Le us know if you have further questions or
problems.Cheers,--Phil

On Fri, 25 Jul 2003 08:53:25 +0800
cd <cding@center.shao.ac.cn> wrote:> Hi,Phil
> I can add the text on the plot now. Thank you very much!
> I still wanna ask you another question(if you don't mind). The wfpc2
> has four ccd, you know, I've analysesd each and get the position of
> star of four fields(CCD) in pixel values. But if I wanna merger all
> four field position into one field, how could I do? Will the presicion
> lose after merger(I don't like use the wmosaic)?
> I hope you won't get bored of these silly questions. You can reply
> according to your time. Thank you again for your so much help!
> All the best.
> Ding Hi Ding,You can use a couple techniques to merge your images, however I think
you will find that using wmosaic will be best, as it makes corrections
to the geometric distortion, rotation, offsets, and scale differences on
the chips.You can assemble your images into one by using imcopy. You need to
first create an artificial image using the tasks in the artdata package
(e.g., mkobjects, mkpattern). Then you can imcopy the images into
sections of the new image, e.g., imcopy im1 newim[1,800:1,800]. The
problem with this is that the new image will contain all the header
information (e.g., WCS info) of the last image copied to the new image.
You will also have to manually make the corrections that wmosaic makes.I still suggest you use wmosaic, as it was designed to properly handle
the images.I hope that helps. Let us know if you have further questions or
problems.Cheers,--Phil

>From cding@center.shao.ac.cn Sat Jul 12 18:43:37 2003
Date: Sun, 13 Jul 2003 9:43:25 +0800
From: "Chen" <cding@center.shao.ac.cn>
To: "fvaldes@noao.edu" <fvaldes@noao.edu>
Subject: - for helpDear Frank,Great thanks for your suggestion on the image plot!I can draw the single ccd, however, you know, the wfpc2 has four ccd
and each 800*800pixel, firstly,I dont't know how to draw these four ccd
together, that means how to merger the four ccd pixel coordinates on a
singel list.Second, if I don't have the fits data image but only have the basic
information from the MAST search results (Ra/Dec of the image center,
Exper time , angle sep .etc), how can I draw the outline of the image?Any advice or instruction from you is expectative to me. Thank you!All the Best,
Ding
cding@center.shao.ac.cn
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡>From valdes@noao.edu Fri Aug 1 09:31:25 2003
Date: Fri, 1 Aug 2003 09:31:23 -0700 (MST)
From: Frank Valdes <valdes@noao.edu>
To: cding@center.shao.ac.cn
Subject: Re: for helpHello Ding,I have been very busy and I saw that you were being helped by Phil. I
have not followed this closely so have you resolved this questions?Yours,
Frank

On Fri, 8 Aug 2003 11:31:52 +0800
"Chen" <cding@center.shao.ac.cn> wrote:> Dear Phil,
> I've restored my compute, thanks for your help!
> I've known the wmosaic is the best task for the image-merge, but I
> wonder can the result big image from wmosaic can be used to be
> scientific analysed? Could I use the task DAOPHOT to find the star
> position on the four-merge field image? If I've got the star position
> of each CCD under the daophot in pixel-value style(x,y f), Could I use
> wmosaic to merge the star positions into a big frame? All the Best,
> Ding
> Hi Ding,>From what I know, the resulting image from wmosaic can indeed be used
for scientific analysis. If in doubt, you may wish to contact the
maintainers, at STSCI (help@stsci.edu). The task wmosaic will
interpolate the pixel values for any overlapping pixels, however, the
image can be used for science.I hope that helps. Sorry for the delayed response.Let us know if we can help any further.Cheers,Phillip

On Mon, 18 Aug 2003 10:32:5 +0800
"Chen" <cding@center.shao.ac.cn> wrote:> Hi, Phil
> I've recieved the mail from the "help@" after sevaral days. He told me
> that the image from "wmosaic" can be scientific analysed but the
> precision of PC1 maybe lowed to the level of WF. Another I wanna ask
> you, the WCSMAP can calculated the coordinates shift and rotation of
> two image, when I calculate the proper motion of two epoch star
> positions can I used the result of "wcsmap" to substitute the plate
> constant solution? What is "wcstran"? Why I can't run of see the help
> of it? Thank u very much!
> All the Best,
> Ding
> Hi Ding,Regarding wcstran, do you mean wcsctran? This task uses the wcs in the
image to transform input coordinates to another (built-in IRAF)
coordinate system.Certainly you can use wcsctran to transform x,y coordinates to world
coordinates, thereby giving you the celestial coordinates of a star in
each of your images. You would do this for each image, i.e., you would
get the x,y center of the star, then transform the x,y coordinates to
world coordinates using the WCS from the same image, using wcsctran.You also wrote in an email forwarded to us by STScI:> The task wcsmap and geotran can transform one image according to the
> reference frame, so if I wanna to calculate the proper motion of two
> epoch stars, can I first transformed the image and then derived the
> starpositions and calculate the proper motion?I don't believe you should have to transform the image, as long as the
WCS in each image is correct. If not, then you would first have to
build the WCS in each image.
I hope that helps. I apologize for the delay. Let us know if you have
further questions or problems.Cheers,Phillip