Erratum to HELP for PLREFL (by A. Zdziarski, 4 Feb. 96).

1. The reflected spectrum of Lightman and White (1988) is averaged over all viewing (inclination) angles. Multiplying it by the cosine of the viewing angle does NOT approximate at all the dependence on the viewing angle. Rather, the spectrum of Lightman and White is a good approximation to the spectrum seen at the angle of 63 degrees (Magdziarz and Zdziarski (MNRAS, 273, 837; 1995). The paper of Magdziarz and Zdziarski discusses in detail the angular dependence of Compton reflection, and the xspec procedures pexrav, pexriv, and tita_a implement the Green's functions for Compton reflection of Magdziarz and Zdziarski.

2. PIEscape (par2) is simply a factor multiplying the normalization of the power law (and so it is NOT an independent parameter in the fit). It should be normally fixed to 1. Fixing it to 0 gives the reflected component.

3. Incl (par3) does not specify the inclination (viewing angle). It simply multiplies the normalization of the angle-averaged reflected spectrum by the cosine of the viewing angle, which is incorrect. Thus, it is not an independent parameter of the fit (it should be fixed at 0).

4. Emax/mc2 (par5) is the energy of the sharp cutoff of the power law. It should be fixed in the fit because the code does not reset the flux array (which leads to errrors when this parameter changes during an xspec session).

5. The reflected spectrum is computed at the upper energy of a given (xspec-specified) photon-energy bin and the number of photons in the bin is obtained by multiplying that value by the bin length. This leads to an unphysical systematic shift of the reflected spectrum on the energy scale. Instead, the mid-energy of each bin should be used to compute the reflected spectrum, and that value should be multiplied by the size of the bin.

Erratum to HELP for PLIREF (by A. Zdziarski, 4 Feb. 96).

1. The reflected spectrum of Lightman and White (1988) is averaged over all viewing (inclination) angles. Multiplying it by the cosine of the viewing angle does NOT approximate at all the dependence on the viewing angle. Rather, the spectrum of Lightman and White is a good approximation to the spectrum seen at the angle of 63 degrees (Magdziarz and Zdziarski (MNRAS, 273, 837; 1995). The paper of Magdziarz and Zdziarski discusses in detail the angular dependence of Compton reflection, and the xspec procedures pexrav, pexriv, and tita_a implement the Green's functions for Compton reflection of Magdziarz and Zdziarski.

2. PIEscape (par2) is simply a factor multiplying the normalization of the power law (and so it is NOT an independent parameter in the fit). It should be normally fixed to 1. Fixing it to 0 gives the reflected component.

3. Incl (par3) does not specify the inclination (viewing angle). It simply multiplies the normalization of the angle-averaged reflected spectrum by the cosine of the viewing angle, which is incorrect. Thus, it is not an independent parameter of the fit (it should be fixed at 0).

4. Emax/mc2 (par7) is the energy of the sharp cutoff of the power law. It should be fixed in the fit because the code does not reset the flux array (which leads to errrors when this parameter changes during an xspec session).

5. Note that the opacities are from Lang (1974) and thus it is not consistent to use this model together with, e.g., WABS, which assumes the opacities of Morrison and McCammon (1983).

6. The reflected spectrum is computed at the upper energy of a given (xspec-specified) photon-energy bin and the number of photons in the bin is obtained by multiplying that value by the bin length. This leads to an unphysical systematic shift of the reflected spectrum on the energy scale. Instead, the mid-energy of each bin should be used to compute the reflected spectrum, and that value should be multiplied by the size of the bin.