From dshepher@aoc.nrao.edu Mon Apr 11 07:58:43 2005 Date: Sun, 10 Apr 2005 13:10:48 -0600 (MDT) From: Debra Shepherd To: Andrew Baker , Crystal Brogan , Carlos De Breuck , James Di Francesco , Debra Shepherd , Rein H. Warmels , Martin Zwaan , Munetake Momose Cc: Joe McMullin , Steven Myers , alma-sw-admin@aoc.nrao.edu Subject: ALMA Offline test starts NOW Thank you very much for helping with the ALMA Offline tests! Everything is ready for the test to start. You should have all received the DVDs with the datasets. If you haven't please let Joe McMullin know. As mentioned in my e-mail of 21 March 2005, this test will use Glish scripting to combine the interferometric and single dish data in two ways and compare the results. The DVDs you received have *reduced* interferometric source uv-data and a final mosaic image and *reduced* single dish images. The combination methods we would like you to try are: - image "feathering" where two images are "feathered" together in the Fourier plane in an optimal way. Low spatial frequencies of the mosaic interferometer are down-weighted to ensure that the final image retains the resolution of the interferometric image. - Deconvolution in which the single dish image is input as a 'default' model, subtracted from the uv data, and the resulting dirty image is then deconvolved with, e.g., CLEAN, Multi-scale CLEAN, and/or MEM. Single-dish + synthesis image combination is a state-of-the art procedure. In the past, it has taken weeks to months to combine data and the final result is never perfect. In particular, image feathering, although easy to use and available in several packages, is not optimum when the resolution difference between the single dish and synthesis images is greater than a factor of 2 or 3. This test represents a comparison between a tried-and-true method (feathering) and an improved deconvolution method. Despite improvements in the algorithms, the images you create will not be perfect and each dataset will benefit from a different combination and imaging technique depending on the image structure and uv-coverage. The algorithms you will be testing have been improved to simplify the combination and imaging process and allow you to increase the final image quality. Your experienced evaluation of these methods will provide both input for the software development as well as scientific insight into the best way to combine single-dish and synthesis data. Joe McMullin has set up a web site with test information: http://projectoffice.aips2.nrao.edu/almatst2.0/ALMA_TST2.0.html There are several key things at this site: - email contacts if you have questions (listed at the bottom of the page). - Information about how to download and install AIPS++. See the AIPS++ Installation README file for detailed information. - A data reduction cookbook. This is the same cookbook that we used last time. Additional text has been added on mosaic techniques, single-dish and synthesis combination, and image analysis. Please also read section 1.1 ``Testing Notes - TST2.0''. Note: if you use xpdf to view the cookbook, you can jump around via links in the pdf file and drive your web browser to the on-line help documentation that is referenced in the cookbook. - See the pdf file (TST2.0.data.description.pdf) at the link: ``Detailed data information and test help. READ THIS FIRST'' This should be the FIRST thing you read when you are ready to start imaging. It contains some background on mosaicing, descriptions of each dataset (sources, field locations, frequencies), and give hints on imaging strategies for each dataset so you don't have to try to second guess the scientific intent from the data itself. - Data set summaries. Please review the summaries for your dataset: Data: Testers: ------------------------- --------------------------------- NGC 4826, CO galaxy, BIMA Andrew Baker Carlos De Breuck James Di Francesco Munetake Momose Orion, 3.6 cm continuum, VLA Crystal Brogan Rein H. Warmels Martin Zwaan Data from TST1, 1.1, and 2.0 Debra Shepherd (regression testing) Note: Orion can be particularly challenging to get a good image because the central field has hundreds of Janskys of flux while surrounding fields have relatively little. The generation of a good mask is critical for imaging success. If you have problems getting a good mask Joe has provided one on the DVD (orion.mask6). Compare your mask with this one to help you define an optimal deconvolution region. - Testing comparison page. This is a page with images of both datasets. These images were generated by internal NRAO testers within the past month (both datasets have been processed many times by many people within AIPS++. Orion was also processed in AIPS while NGC 4826 was processed in MIRIAD by the PI). Although your images will probably not match these exactly due to imaging and weighting differences, you can use these images as a guide. - A Test 2.0 Report Questionnaire - when you finish testing, please answer the questions on this form - I will merge your answers into a final report and give this to the Offline group to help guide their development. - A list of requirements that I would like you to grade based on your testing. This is an excel spread sheet that you can fill in or a pdf file that you can write on and fax to me. - As test results come in, the scripts and final images will be put on this web site. The scripts will also be used for regression testing in the future. We expect that this test should take less than 40 hours of your time. The test begins on Monday, 11 April 2005 and will end on Monday, 9 May 2005. During this month, the AIPS++ software developers and I will be available to help you with questions or problems that you may come across. Please do not start the test at the last minute! Although extensive testing of the software has been done in-house, software bugs may still exist. The developers will do their best to help you through any difficulties, including fixing any bugs and distributing new software if necessary. But they can only provide this support if they are given adequate time. If you find you cannot complete the test by 9 May, please contact me as soon as possible (dshepher@nrao.edu). If you have any questions, issues, problems, please don't hesitate to contact aips2-requests@nrao.edu (this will be routed to me and Joe McMullin). Sincerely, Debra Shepherd National Radio Astronomy Observatory Subsystem Scientist for the ALMA Offline software e-mail: dshepher@nrao.edu work phone: 1-505-835-7398 FAX: 1-505-835-7027 home phone: 1-505-838-4014 ALMA test website: http://projectoffice.aips2.nrao.edu/almatst2.0/ALMA_TST2.0.html