Dear Colleagues, After too long an absence and at least one too many a tragic loss, we are resurrecting this project with the hopes of completing the comparisions begun in September 2006 and writing a second manuscript. Our best recollections of the discussions in 2006 were that the comparisons for the "TPD10" case, the off-disk-center case with varying degrees of added noise were satisfactory. The overall results were that (as far as ambiguity resolution was concerned), the addition of even a large amount of noise didn't change the results much: areas where an algorithm was successful continued to be successful and vice versa, albeit with patches of some noisy "hash". The "flowers" case was a different story. Wide-spread dissent with this model was expressed. sigh... the complaint we recall being voiced loudest was that "it wasn't solar enough", and that the fine-scale structure of the "flower petals", our crass simulation of penumbral fine-structure, had too much fine structure. Well, luckily we now have Hinode data to show that indeed there is almost no limit to fine-scale structure. Nevertheless, we have re-worked the "flowers" case, removing the most complained-about flower, and adding a patch of simulated plage. The plage compares fairly well with the structure observed by Hinode (and we do not claim that it is a true physical model by any means). The flowers data are now available on our website, http://www.cora.nwra.com/AMBIGUITY_WORKSHOP/2006_workshop/ at two different spatial resolutions generally simulating something like 0.3" and 0.9". Comments are welcome. The TPD10 data are also available. We still are lacking somewhat on metrics by which to compare the FLOWERS cases. We are investigating a metric which uses the binned field from the original model as a solution; initial tests are that it doesn't get as scrambled as originally thought. As solutions come in, I'll post the metrics that we used before (total current, inferred alpha_best), and work to come up with and present new ones. As always, we are very interested in your thoughts. Details on submitting/uploading are below and also at http://www.cora.nwra.com/AMBIGUITY_WORKSHOP/upload.html Sincerely, -KD Leka, Graham Barnes p.s., if you'd like to be removed from this email list, please let me know. If I've missed someone, please somehow let me know, too! ============================================================== To provide solutions, please use the anonymous ftp site ftp://ftp.cora.nwra.com/incoming/AMBIGUITY_WORKSHOP Solutions should be IDL ".sav" files with the data-name and a unique indicator such as "first_initial+last_name+filename.sav" For example, my solution for TPD10, potential-field solution could be called "KLeka_TPD10_pot.sav". If you have more than one solution, please indicate that in the filename. At the very least, an image-plane ambiguity-resolved azimuth array needs to be provided as a solution. Please contact me if you have any questions at leka@cora.nwra.com ============================================================== DATA DETAILS: The data sets are presented as an IDL-structure following the Hawaii data format (outlined below) in an IDL save file (".sav"). You may use this format for presenting your solutions. What is required is that results must be presented in the observer's frame, just as the original data. B.B_AZIM (or a 2-d array of image-plane unambiguous azimuth directions) is the ONLY parameter that will be used directly in the comparisons. That is, if you return simply the same structure but an updated b_azim tag, that will be sufficient. If you return simply a 2-d array of angles from the the ambiguity has been removed without any additional rotations/etc., that will suffice: we will recompute Bx,By,Bz from your provided b_azim for comparisons in heliographic coordinates. There are two cases for the flowers construct, at two different resolutions. As before, the model field is originally on a much finer grid, Stokes profiles were computed, binned, and then inverted again. The binned spectra are also available if you would like to use them. All spectra and the inversion were computed using an Unno-based Milne-Eddington approach, nothing fancy. Everyone who is presenting a method or works with vector field data is encouraged to "put their money where their algorithms are", and submit solutions to the trial data sets. THANKS! =================================================================== DATA FORMAT: The following is the data format: an IDL-save file containing a structure (named or anonymous) that presents the vector field and coordinates with tags: {i_cont,latitude,cmd,b_long,b_trans,b_azim,b_fill,point,chi2} where "point" is itself an IDL named structure containing the pointing information: point = {point,lat,cmd,b0,p,radius,pix_size} The following are two-dimensional arrays: i_cont: continuum intensity latitude: solar latitude array for each point in the field, measured northward from the solar equator, in degrees cmd: solar longitude array for each point in the field, measured westward from the central meridian, in degrees b_long: longitudinal field strength, Gauss (averaged over the pixel if the analysis calculates a fill fraction) b_trans: transverse field strength, Gauss (averaged over the pixel if the analysis calculates a fill fraction) b_azim: azimuth angle of the transverse component of the field vector in the plane of the sky, measured counterclockwise from solar north, in degrees, +/- 180 (in ambiguous data it will be +/-90). In your provided solutions this will be YOUR ANSWER. b_fill: fitted fill-fraction from the inversion chi2: a rough 'goodness of fit' from the inversion The structure "point" contains the following quantities: lat: float, solar latitude of center of the observed field, radians cmd: float, solar longitude of the center of the observed field, radians. be careful, the array for longitude in the main structure has the same name and different units!) b0: the ephemeris value for inclination of the solar rotation axis to the plane of the sky, radians p: the ephemeris value for the angle the solar rotation axis with respect to the terrestrial north direction, in radians. This quantity can be set to zero if the solar north direction is along the ordinate (slow axis) of the data arrays radius: solar radius as seen from Earth at the time of the observation, in arcseconds pix_size: two-dimensional array containing pixel dimensions [horizontal,vertical], in arcseconds Additional tags are accepted.