Iris Bugs & Caveats

IRIS

GUI Menus

The Help menu is still not operational. Users are directed to the Iris web documentation for information about running Iris. If you are not able to find an answer to your question in the documentation, please inform the VAO Helpdesk.

The Preferences menu dialog is misleading, as only a certain subset of items listed are actually enabled, and other items, like the symbol style, are present to break ties in the case of coplotting. There are no resources yet to enable display of user-selected points with different colors and/or symbols.

Visualization/Interaction

On certain computer platforms the cross-hair cursor has a tendency to eventually leave behind graphical artifacts (line segments) when the mouse pointer is moved too fast over the plotting surface, or when the plotting surface gets covered and then uncovered by another window. The 'Redraw' button can be used to clean up the plot in these situations, although a more permanent and elegant solution is being sought.

If the user edits the plot title (by clicking on it and using the edit dialog that shows up) and then changes the units of the plot, the plot title reverts back to its original form. This bug will be fixed on the next release.

File Input/output, SED building

The contents of the log file created by Iris in the user's home directory are still preliminary. Currently, the log file records which files were read during the session, which files were written, as well as the input and results from every attempt to fit a spectral model. Future versions will provide a much more detailed record of each session.

Input SED files can have multiple segments, although when such files are read by Iris, the segment structure is 'flattened out', so to speak. Iris does not currently provide any way for the user to interact with the segment structure, although such capability is a high priority and will be included in a future release.

Currently there are no tools in Iris to handle bad data points - e.g., to mask bad data on input - although SED data points with no associated error are displayed with a special color and symbol. This is pending a more detailed definition of 'bad data point' within the SED context.

SED Model and Fitting

Fitting Interface

A model fitted to data is discarded when the model fitting window is dismissed. Iris issues a warning dialog to prevent the user from accidentally losing a spectral model before saving it to file. In a future release, the model, once created, will remain associated with the SED data it came from until the data itself is deleted from memory.

When a spectral component is deleted from the model fitting window, the component is not removed automatically from the Sherpa expression in the same window. It is only when an actual fit is attempted that Sherpa will issue an error dialog that warns about the inconsistency between the expression and the list of components.

Models

The blackbody model gives valid results only for wavelengths greater than 60 Angstroms (or frequencies below 5.0 * 10^16 Hz), as the model was originally implemented for UV/optical/IR data. If the blackbody model is fit to a SED, then the user needs to define range(s) that do not include data points with wavelengths less than 60 Angstroms (or frequencies greater than 5.0 * 10^16 Hz). This limitation will be resolved in the next release.

Units

While the user may plot the SED using a variety of units, for modeling and fitting Iris always converts the data to one set of standard units: photon-flux density vs. wavelength (i.e., photons/s/cm^2/Angstrom vs. Angstroms). If the SED is displayed in different units, Iris internally converts the SED to photons/s/cm^2/Angstrom vs. Angstroms; fits the model to the SED; and then converts the SED and model values back to the units the user chose for plotting, to display SED and model together in one plot.

Therefore, the model parameters that have physical meaning always use these internal units. For example, the center and FWHM of a Gaussian, or the break for a broken power-law, etc., always are expressed in Angstroms - even if the user has chosen to display the data in, for example, MHz. Model parameters cannot be converted to different units within Iris.

Chi-squared Fitting / Flux Density Errors

For chi-squared fitting, each SED data point must include an error for the flux density. (In the plot this is indicated with a symmetric y-error bar on the data point.) Data points with such errors are plotted as black squares; data points lacking error bars, or having only upper or lower limits, are plotted as magenta diamonds. Data points lacking error bars are automatically excluded from fitting when using chi-squared. (The points lacking error bars are included in fitting when using least-squares, Cash or C-statistics, since these statistics do not make use of the errors.)

Calculating Confidence Intervals

When calculating confidence limits, the user must use either chi-squared or C-statistic. Confidence limits cannot be calculated with the least-squares or Cash statistics.

SED IMPORTER

Interoperability with VO Tools

Simple Application Messaging Protocol (SAMP) support is limited to incoming messages. You can transmit a table to the SED Importer from other SAMP-enabled VO applications (e.g., Topcat) and import it as a segment; however in the current beta version, users cannot send data outward from Iris. Also, each time a table is sent to the SED Importer through a SAMP connection, a new SED is created. These issues will be resolved in the 1.0 release.

For compatibility with Iris, unit strings in SED Importer output files do not follow the IVOA standards.

For compatibility with Iris, users cannot import segments with asymmetric errors.

Input File Format

If you import FITS or VOTable files with multiple tables, only the first one will be imported. This limitation affects the SED Importer, only, and not Iris.