next up previous contents
Next: XSTAR Fixed Parameters Up: Parameters Previous: Parameters   Contents

Physical Parameters

XSTAR models are based on 21 physical parameters described in detail in Chapter 4. In quick summary, they are cfrac, temperature, pressure, density, trad, rlrad38, column, rlogxi, habund, heabund, cabund, nabund, oabund, neabund, mgabund, siabund, sabund, arabund, caabund, feabund & niabund. Each of these parameters needs at least one additional parameters and as many as four additional parameters to specify is variation during the program run.

Each parameter has three levels for classifying its variability:

Constant (variation type = 0):
This parameter is held constant in all the XSTAR runs.

Additive (variation type = 1):
The Additive class of parameters provides a simple method for varying parameters that are reasonably independent of the others. For more info on how additive parameters function, see the XSPEC manual and OGIP Memo OGIP 92-009.

Interpolated (variation type = 2):
Interpolated parameters provide the greatest accuracy in building table models. They also require the most processing time. For each interpolated parameter, you can define some number of points between a maximum and minimum range. The placement of these intermediate points is determined by the interpolation type - linear (interpolation type = 0) or logarithmic (interpolation type = 1).

In estimating the running time of XSTAR2XPEC, the key factor is the number of times XSTAR is called. Consider a run with ${\rm N}_{\rm I}$ interpolated parameters, where interpolated parameter $i$ is evaluated at $n_{i}$ points $(1 \leq i \leq {\rm N}_{\rm I})$. The total number of times XSTAR is called is then $\prod\limits_{\rm i=1}^{{\rm N}_{\rm
I}} n_i$. However, if ${\rm N}_{\rm A}$ additive parameters are also defined, then for each set of interpolated variables, there is one run with all the additive parameters are zero and the remaining ${\rm N}_{\rm A}$ runs have one of the additive parameters at it's maximum value and the rest all zero. This means that the total number of times XSTAR must be run is given by

({\rm N}_{\rm A}+1) \prod\limits_{\rm i=1}^{{\rm N}_{\rm I}} n_i
\end{displaymath} (7.1)

and this can provide you with a feel for how long a complete XSTAR2XSPEC run will require. As an example, if we defined 2 interpolated parameters (one evaluated at 5 points and the other evaluated at 4 points) and 8 additive parameters, XSTAR would be run a total of
(8+1)(5)(4) = 180 calls
\end{displaymath} (7.2)

which means that if the XSTAR runs for the appropriate model range averages five minutes, it will take approximately (180)(5 minutes) = 900 minutes = 15 hours for a complete XSTAR2XSPEC run.

next up previous contents
Next: XSTAR Fixed Parameters Up: Parameters Previous: Parameters   Contents
Tim Kallman 2014-04-04