Formulation and selection of objects for QORG-style galaxy-objects charts. Eric Flesch, 16 June 2006 The galaxy-objects charts presented here use the latest published catalogs as sources. Selectivity is used to exclude objects which are judged unreliable. Following is a precis of the method, consisting first of a listing of the data included, then a discussion of object selection, and considerations in selecting galaxy fields for display on these pages. Included data starts with the 2004 Quasars.Org (QORG) all-sky optical catalog of radio/X-ray sources (Flesch & Hardcastle, 2004 A&A 427,387) which presented 501761 objects. QORG uses a density-based likelihood algorithm to calculate odds that an optical object is associated to a corresponding radio/X-ray detection, and how likely that it is a QSO or galaxy or star. QORG used ROSAT catalogs exclusively for its X-ray data; ROSAT was the X-ray satellite which operated from 1990-99. There are now two next-generation X-ray satellites in operation, Chandra and XMM-Newton, which have improved capabilities over ROSAT. The QORG paper used early data from these to demonstrate the correctness of the QORG astrometric solutions of the ROSAT fields. Since QORG there have been new large QSO catalogs published, and catalogs of X-ray sources from the Chandra and XMM-Newton surveys. These are used in tandem with the QORG data, and likelihoods have been calculated for the X-ray data using the QORG algorithm which is an appropriate tool as it is a purely data-driven method. The individual added catalogs are listed hereafter, plus notes on how they were used. (1) Chandra (aka ChaMP or CXOMP) data: This is available chiefly from the xassist site (xassist.pha.jhu.edu/xassist/index.jsp), which provides 34609 astrometrically unique X-ray sources which match to 8989 APM-USNO-A optical objects using the QORG algorithm. Optical solutions of Chandra fields result in fields typically shifting by 0-2 arcsec against the APM astrometry, and 15 fields (out of 658) shift by 3 arcsec. Raw APM astrometry can be discrepant by up to 2 arcsec, so these solutions imply a Chandra astrometric pointing uncertainty of about one-half arcsec. There is also the ChaMP X-ray Point Source Catalog (2nd release, 2006), at hea-www.cfa.harvard.edu/CHAMP/IMAGES_DATA/index.html which however has only 6512 sources and does not provide the field ID, so that no optical solution can be done. Comparison to the xassist data shows that the ChaMP catalog performs no optical solution. Frankly, this catalog should have been 15x larger and included the field ID information. However, I include it for those detections which are not included in the xassist data. Optical-Xray associations are cropped using the ChaMP error circle as a classification parameter. Altogether, the yield from these two catalogs, as at June 2006, is 7187 Chandra detections associated to APM/USNO-A optical objects. (2) XMM-Newton data: This is available chiefly from the XMM-Newton Serendipitous Source Catalogue (2003), which provides 41815 astrometrically unique X-ray sources which match to 12463 APM-USNO-A optical objects using the QORG algorithm. The XMM data, as presented in the catalog, has already been astrometrically corrected using an optical solution. Analysis shows this has been inconsistently applied, with some fields corrected and others uncorrected. I have done my own optical solutions against the presented data, resulting in typical shifts of XMM fields of 0-2 arcsec, with a few fields shifting by 3 or 4 arcsec against the raw APM astrometry. The xassist site provides some additional XMM data, 14296 astrometrically unique sources which map to 2808 APM-USNO-A optical objects. There is no astrometric correction in this data, so applying my QORG-method optical solutions results, again, in typical shifts of 0-2 arcsec, and 19 fields (out of 149) shift by 3 arcsec, and 3 fields shift by 4 arcsec. As this is excess to the maximum 2 arcsec error of APM astrometry, I infer that XMM astrometric pointing has a typical uncertainty of 1 to 2 arcsec. Optical-Xray associations are cropped using the XMM error circle as a classification parameter. Altogether, the yield from these two catalogs, as at June 2006, is 8273 XMM detections associated to APM/USNO-A optical objects. (3) Veron & Veron-Cetty QSO catalogue version 12 has been released (March 2006) which includes all QSOs in the literature to the end of 2005. This has been included. (4) SDSS Quasars: The QORG catalog included only Data Release 1 from the SDSS, which contributed 17959 QSOs which are found in the APM/USNO-A data. Since then there have been SDSS DR2, DR3 and DR4, and I have changed the method from requiring APM/USNO-A matching, to using SDSS photometry where APM/USNO-A photometry is missing or one-color. The total SDSS QSOs thus now number 62804. (5) NBCKDE Quasars: This is a large catalog of photometrically redshifted quasars from the SDSS DR3. Home page is sdss.ncsa.uiuc.edu/qso/nbckde . The catalog presents 273287 objects which are adjudged 95% likely to be QSOs. Merging against catalogs of known objects leaves 189760 objects, not previously identified, which are likely QSOs. However, these are listed with various probabilities of having redshifts in various ranges. For our purpose of display on these charts, we need a fair indication of the QSO's redshift. Thus I have analyzed the NBCKDE objects against the full set of SDSS-redshifted objects, and have identified a subset which is >85% likely to have a redshift within 20% of the listed value, which I feel meets the standard for display on these charts. This criterion results in a set of 111174 NBCDKE QSOs included in our compendium. They are displayed on the charts with redshift rounded to the nearest 0.1, eg, 1.700, so they can be easily recognized upon sight. In the data tables they are named as NBCKDEhhmmss.s#ddmmss even though the authors did not moot such a notation, so that it is clear to the reader that these are NBCKDE photometric quasars. (6) The Serendipitous Extragalactic X-ray Source Identification (SEXSI, astro-ph/0603556, March 2006) catalog has been included. This survey took spectra of optical counterparts to 477 hard X-ray sources. We retain those quasars and ELGs which were not previously identified and have good optical data, which amount to 391 objects. We take the ELGs as AGN galaxies by the judgement of the paper which states in its conclusion "the majority of these sources are Seyfert 2 galaxies", as Ne V emission was detected. (7) the background: I use SkyMap Pro v5 (1998) to produce the charts, and the galaxy positions, size and tilt are provided by the included data. I import the data listed above into SkyMap to produce the charts that you see. Discussion: (1) Object Selection: The compendium of all data listed above amounts to 664165 objects. However, some of these are just eg, radio galaxies or stars, or radio/X-ray emitting objects which QORG calculates as being less than 40% likely to be a QSO. I think of 40% as being a minimal criteria of interest, in that the intent here is to display objects which are *likely* to be QSOs, and QORG's actual record is that a listing of 40% translates into >50% true odds (see quasars.org/docs/Testing-QORG-via-SDSS.txt for a full discussion). So I wish to display only known QSOs and objects with QSO pcts of >=40%. With the benefit of the additional data (especially SDSS) to hand since publication of QORG, I have gone back and analyzed the performance of QORG QSO candidates against known objects, categorizing these by psf type (ie, stellar, or fuzzy, etc) by color, and listed QSO pct. I find that some types of objects are highly overperformed, ie, candidates stellar in both red and blue, yield over 40% QSOs for QSO pct values as low as 21. Other types, notably one-color objects, perform below the listed value, possibly because of APM one-color artifacts. I have devised cut-offs for all object types below which I judge that the likely QSO result is less than 40% QSOs, and so have removed those from the data pool. The remaining data consists of 353875 objects available for display on these charts. (2) Selection of galaxy fields: The standard to aim for is that of OVERDENSITY, where QSOs (known and presumed) are packed around a galaxy more thickly than would be expected randomly. But all the surveys listed above are heterogeneous, that is, they are done in some places and not in others. One does not want to display a difference in object numbers, when the entire basis for the difference is that observations were made in one place and not the other. To avoid this trap, for every galaxy field, essentially uniform observations must have been made of the field out to a distance of one degree from the galaxy. Radio surveys are uniform so pose no problem. For X-ray surveys, it is required that a ROSAT PSPC observation be done, as these are (usually) 2-degree observations allowing good comparison between inner field close to the galaxy, and the outer areas. It frequently happens that XMM / Chandra have re-observed the inner area and added some objects, but the difference is usually slight. The difference is slight because when the ROSAT fields are correctly shifted (as the QORG catalog did systematically), their accuracy in optical object selection is very good, e.g NGC 891 had 18 identifications compared with 25 when Chandra is added. So if we require a PSPC field and an overdensity of 10+, then we should be on safe ground that a truly interesting field has been selected. Another trap is that it is expected that randomly, some fields will be denser than others. There may be, say, a distant cluster of QSOs (although QSOs do not seem to come in large clusters). If all surveys were uniform, we could calculate the expectation of random clustering. But, again, the surveys are very heterogeneous. So we fall back on certain naive notions, that is, if a clustering looks *far in excess* of our notions of fair randomness, then we presume some causality. So in the absence of a formal caculation of significance, we rely on our sense of balance and/or absurdity to look at a galaxy-quasar field and decide if this arrangement can be random, or if it "just ain't so". The Big Bang model has become so well-entrenched that it seems immovable. If one makes the common- -sense observation that, say, in the Big Bang model elliptical galaxies steadily shrink in real size as redshift increases, and that this is evidence that the model is wrong, then the reply is that it "shows evolution"; in other words, even the precept of homogeneity must give way to the Great Paradigm. For those of us who feel that elliptical galaxies have not changed in size through the aeons, the only way out might be to compile cases of such visual impact that the paradigm of "This Is Not Hard" (a.k.a. "D-uh") manages to overcome the paradigm of the Big Bang. These pages look for such pictures while keeping in mind that perhaps they do not exist. But better that they do not exist, than if they did and we did not try to find them.