New Model: RELXILL v0.2h

RELXILL is a brand new model, joining forces of the XILLVER reflection code (link; Garcia et al., 2010, 2011, 2013) and the RELLINE code (Dauser et al., 2010, 2013). The striking new feature is that for each point on the disk the proper xillver-reflection spectrum is chosen for each relativistically calculated emission angle. The implications of this improvement and more details regarding the model are provided in Garcia & Dauser et al., 2014, ApJ, 782, 76 (2014) and Dauser & Garcia et al. (subm. to MNRAS). See also this pdf for a short information on all models and their parameters.

v0.2h: Add a twice broken power law version of relxill (relxill_tbp).

v0.2g: Made this version of relxill compatible with the newest update of the xillver table.

v0.2f: fixed bug with negative heights in relxilllp_ion and set all hard limits of the height parameters to postive values by default in order to avoid problems during fitting with XSpec.

v0.2e: several bug fixes were applied (e.g., problems occured with the reflection fraction when combining relxill and xillver, the reshift parameter in xillver, and for certain combinations of the fixReflFrac switch)

v0.2d: updated to code to use the new 1 MeV Ecut table

v0.2c: added non-relativistic xillver model

v0.2c: added relxilllp_alpha, using the alpha disk density (see Shakura & Sunyaev, 1973)

v0.2a: Major Update: added models relxill_ion and relxilllp_ion, which allow to take an ionization gradient in the disk into account (see below on information on how these models work).


Left: Figure of an accretion disk, showing the predicted emission angle from GR ray tracing (upper panel: inclination 40°; lower panel: 80°; Garcia & Dauser, 2014, accepted by ApJ). It is evident that the emission angle at the inner parts, where most of the radiation is expected, does differ largely from the inclination angle. Therefore a proper relativstic treatment is necessary to take the angular effects into account.

General Information

By installing the new RELXILL model, all previous implementations of the relline-like and relconv-like models are automatically installed, too. These models are completely identical to the ones from the RELLINE installation (hence, no additional installation of relline is necessary to get the previous functionalities).

Usage and Parameters

The RELXILL model now includes by default the irradiating power law source. It's strength can be set by the parameter refl_frac. This reflection fraction is defined as the ratio of photons that hit the disk to those that reach infinity. Due to strong light bending effects, this value can easily be larger than 1. More details can be found in Dauser+2014. Note, that for a negative value of refl_frac, only the reflection compontent will be returned (similar to the pexrav model definition). All information can also be found in this pdf: download

norm: defined as the normalization of xillver divided by the ionization parameter
refl_frac: reflection fraction (relxill): calculated between 20 and 40 keV
refl_frac: reflection fraction (relxilllp): calculated directly from the lamp post geometry
Ecut: cutoff energy in keV (the same for the power law and the reflection spectrum)
angleon: 0: angle averaging (see below for more information)
1: take angles properly into account
fixReflFrac: 0: free reflection fraction (determined by the refl_frac parameter)
1: fix reflection fraction to the lamp post value (i.e., the parameter refl_frac has NO meaning in this case)
2: free, (same as 0:), but now the reflection fraction is displayed on the screen

The meaning of the parameters is the same as for the separate models (see relline description). The only exception is the additional parameter angleon. If set to 0, the relxill model acts like a simple RELCONV x XILLVER, hence using only the angle averaged reflection spectra and then applying the relativistic smear. However, if set to 1, it acts as described above: For each point on the disk the proper xillver-reflection spectrum is chosen for each relativistically calculated emission angle. Note that due to the relativistic effects the emission angle is generally not equal to the inclination angle and therefore this more complicated approach is necessary. Moreover, as the angular effects are properly treated, no "guess" regarding a limb brightening or darkening law is needed anymore. Hence, unless you really know what you are doing: Leave angleon=1. It is definitely the exacter model (no matter what your chi2 value is telling you)!


To install, please download the following files:

last changes in version
relxill_code.tgz 2015-04-22 v0.2h
tables.fits.tgz (warning: 784MB) 2015-03-19 v0.2g

The installation is similar to the relline model (look here). Note that you can use the environment variable RELLINE_TABLES to point to the complete set of provided tables, including the xillver table.

A Comparison between RELXILL and the angle averaged approach


Left: Relativistically smeared spectra using RELXILL (standard parameters: logxi=2, gamma=2, index=3, spin=0.99). Middle: Ratio between relxill and the conventional relconv*xillver. Right: Relativ distribution of flux for different emission angles. Note the angle averaged approach, which is commonly used, assumes a flat distribution here.

The code here is still in BETA-test! Please contact me if you find bugs or encounter problems with the model.