Subsections

• 2.1 August, 2006
• 2.2 February, 2003
• 2.3 July, 2000
• 2.4 April, 2000
• 2.5 February, 2000
• 2.6 January, 2000
• 2.7 November, 1998
• 2.8 October, 1998
• 2.9 July, 1998
• 2.10 March, 1998

2 What's New?

2.1 August, 2006

• Added Section 10.1 on catalog matching tools.

2.2 February, 2003

• Added Section 10.2 on Adaptive Kernel Image Smoothing.

2.3 July, 2000

• The dataset_3d tool can now compute a ``statistic map'' using any three event properties, for example a median energy map. Both fixed-size and adaptively sized circular kernels are available. See Section 6.2.

  Once the statistic map is computed, pressing the middle mouse button will display the data used to compute the map's value at that location. If one uses a kernel size roughly equal to the typical size of sources, this may provide a rough way to browse the spectra or light curves of sources by clicking on them in an image.

  One of the statistics available will make a map of so-called ``flaring pixels''. The tool will not, unfortunately, allow you to remove such flaring pixels.

• Gaussian smoothing of densities (histograms) has been added.

• The distribution of the angles (from a reference point) to bivariate data is now displayed when Radial Profile from Reference Point is selected (Section 5.4).

• The facilities for combining images, e.g. computing the ratio of two images, has been generalized so that images from any currently running TARA tool can be used (Sections 5.4 & 7.)

• Parameters from fitting routines are now displayed in a ``clipboard'' widget to facilitate saving the values to a file.

2.4 April, 2000

• Analysis tools that use a plotting window can now be resized by dragging the window corners with the mouse.

• The default plotting region on paper now retains the same aspect ratio as the corresponding screen display.

• The controls for choosing log or linear scaling of images have been moved from the ``Z-Axis'' button at the lower left corner of widgets to the Edit dialog box.

• The true_color_image tool now supports several variations on the RGB and Hue/Brightness color models. See Section 7.

• Users may now increase the number of predefined colors (used for plot lines and symbols). See Section 9.

• Improvements in the FITS I/O routine tbget.pro in the Astro Lib have reduced the time required to read event lists by 40% and the memory required by 25%.

2.5 February, 2000

• A rare but serious bug¹in Event Browser was fixed. Prior to version 9.43 a ``race condition'' existed whenever the user loaded a new event file (Domain Dataset) and had filter widgets already defined. In that situation Event Browser automatically applies the filters to compute a new Working Dataset, i.e. it ``pushes the Apply Button'' for you. In some cases the resulting Working Dataset was incorrect. Subsequent manual activation of the Apply Filters button would produce correct results.

• Event Browser now attempts to pass on the FITS keywords it finds in the EVENTS binary table. Some keywords describing the table columns are not however passed on, so some CIAO tools may still have trouble with event lists saved by Event Browser.

• The user may now customize Event Browser by supplying a routine that computes an arbitrary domain dataset filter, and by supplying routines that define new derived properties (section 3.5).

• EB now offers more flexibility in working with multiple datasets (section 3.3).

• The working dataset is no longer automatically recomputed when Derived Property definitions or Analysis Parameters are changed (section 3.2.2). The user must press the Apply Filters button. If you're changing only one property definition or analysis parameter this new design is a minor annoyance, but if you're changing several then you will now be spared waiting for everything to be recomputed and displayed multiple times.

• A bug that prevented rotated datasets in dataset_2d (section 3.3.4) from being filtered correctly has been eliminated.

2.6 January, 2000

• The dataset_2d tool can now plot a contour map made from one dataset on top of an image made from a second dataset. See Sections 5.2 and 5.1. FYI other options outside of TARA for drawing contours on images include the FTOOL fv, Harold Ebeling's IDL routines /bulk/pkg/asc/ebeling/imcont.pro & imcont2.pro, the Astro Library routine IMCONTOUR, and the IDL intrinsic routine IMAGE_CONT.

• Two animation facilities have been added. The dataset_3d tool (Section 6) will show the distribution of trivariate datasets (e.g. produce an animation of the sky with each frame representing a slice of the photon energy axis). The dataset_2d (Section 5.4) tool can ``blink'' images made from two or more datasets.

• Support for world coordinates has been added. The mouse position is displayed continuously and the coordinates of the markers are available under the Titles dialog box (Section 4.3).

• A Trend_1d tool has been added to study CTI and other trends in one property as another property varies (Section 3.3.3).

• Crude support for de-dithering AXAF event data has been added (Section 3.2.2).

• As always then centroid of the region-of-interest (ROI) is displayed in the status line just above the plot. You can now mark with centroid with the Big Marker by selecting Analysis->Mark Centroid (Section 5.4). Since the Big Marker is also used to define the center of a annular-style ROI, you may find it useful to iteratively ``peak up'' on a source by alternately defining an annular ROI and moving the marker to the centroid.

• The axis range controls have been improved (Section 4.3).

2.7 November, 1998

OBSOLETE JULY 2000

A bivariate dataset may be weighted by a third event property, forming greyscale images which depict either the sum or average of the weights in each pixel. For example, if you weighted event positions by the event energies and specified that the weights should be summed, then you would get an image which approximates a raw CCD frame integrated over a long time. If instead you specified that the weights should be averaged, then you would get an image which depicts the average energy of the events at each location.

 $$

The difference or ratio of the densities of any two datasets loaded into the same Bivariate Analysis Tool may now be computed and displayed. See Section 5.4

 $$

Taking inspiration from the very nice color images produced by the AXAF HETG team, we've added a tool that will produce two kinds of ``true color'' images. By ``true color'' we mean images whose colors are derived from two or three separate datasets, as opposed to false color images which are produced by simply mapping the intensity values of one dataset onto a color map instead of mapping them onto a grey scale. See Section 7

2.8 October, 1998

• The way in which Event Browser reads intrinsic event properties from FITS event lists, calculates derived event properties, and stores all event properties has changed significantly. One goal for these design changes was to make Event Browser more flexible internally so that handling new ACIS modes (e.g. graded mode) and handling new file formats (e.g. ASC Level 1.5 files) would be easier to code. Fortunately, from the user's point of view these internal changes are visible only has modest changes in the controls on the Domain Dataset widget (at upper left of screen). Please see Section 3.2.2.

• We have implemented a new strategy for dealing with the situation where a user has integer-valued positional data (e.g. TDETX, TDETY) and wishes to apply an annular region-of-interest (ROI). In previous releases the bivariate analysis tool itself would dither the integer positions just prior to applying the ROI. Audrey Garmire recently pointed out that this design is flawed because the working dataset changes each time the annular ROI is applied to the data, even though the user has changed neither the data nor the ROI. In this release a different approach has been taken. The bivariate analysis tool does not dither its data. Instead, the user has the option of defining the event properties x_position and y_position as dithered copies of integer-valued FITS columns. Please see Section 3.2.2.

• Event Browser can now work with ACIS graded mode datasets.

• Event Browser can now work with any column of data that appears in an event list. For example the Penn State CCD simulator can write a column named "DEPTH" which contains the depth into the CCD at which a simulated photon was absorbed; a user can now work with those depths in Event Browser, for example plotting depth versus energy.

• Event Browser can now work with event properties which were previously not accessible to the user. For example the event property right_pix (the left-hand pixel of the event island) is now visible to the user. For example, one could plot right_pix versus CHIPX for one amplifier to visualize serial CTI.

• The user now has full control over which event properties, both intrinsic (read for the input files) or derived (calculated by Event Browser), are to be saved in an output event list.

• A colorbar - a type of legend for the scaling used in an image display - is now an option in the bivariate analysis (dataset_2d) tool.

• All the keywords found in the primary FITS header of the input event list file can now be viewed by the user and are of course carried through to the output event list file.

• My earlier claim that Event Browser can read MARX-produced FITS event lists was incorrect. The ASC's current FITS design stores the nine pixels of an event island in the order in which they were telemetered. This is different from the ASCA standard island order which Event Browser expects and which is produced by ATICA, by the ACIS high-speed tap software, and by MIT's subassembly calibration software. The ASC is currently reviewing their format decision so I have postponed modification of Event Browser to accept their format.

• The tool function_2d has replaced the old tool ccd_viewer. You may call function_2d directly and pass a 2-D array (image) or you may use the tool fits_viewer to read an image from a FITS file.

2.9 July, 1998

• Individual model parameters can be frozen in the multi-gaussian fitting tool in the univariate dataset analysis widget.

2.10 March, 1998

• Event Browser can now work with several different spatial coordinate systems (see Section 3.2.2) commonly found in event list files (X/Y, DETX/DETY, TDETX/TDETY, CHIPX/CHIPY, RAWX/RAWY).

This release includes a major overhaul of the tools used to analyze univariate datasets (e.g. spectra and light curves) and bivariate datasets (e.g. images) in Event Browser:

• Direct application of the Maximum Likelihood Method to fitting multiple gaussian models to univariate datasets (as opposed to least squares fitting to a histogram)

• New methods for visualizing 2-D density functions (images): wire grid surface plot, lego surface plot, shaded surface plot, contour plot

• Smoothed histograms (an approximation to the kernel smoothing method of non-parametric density estimation)

• Angled regions-of-interest and angled cuts (inelegant method)

• More explicit and convenient axis range controls, in a pan-and-zoom style

• More convenient image scaling controls

• Resizable plot windows