Page History

...

The bug lay in the code that computes the exposure averaged PSF. This is given by

{latex}
\newcommand{\e}{\varepsilon}
\newcommand{\Int}{{\displaystyle \int}}
\begin{eqnarray}
P(\psi, \e) &=& \frac{\Int dt P(\psi, \e, \theta) A(\e, \theta)}
        {\Int dt A(\e, \theta)}\\
        &=& \frac{\Int d\theta P(\psi, \e, \theta(t)) 
                       A(\e, \theta(t))|\partial\theta/\partial t|^{-1}}
                 {\Int d\theta A(\e, \theta(t))
                       |\partial\theta/\partial t|^{-1}}
\end{eqnarray}
{latex}

Here

$P${latex}$P${latex}

$A${latex}$A${latex}

{latex}$\theta=\theta(t)${latex}

$P${latex}$P${latex}

$A${latex}$A${latex}

{latex}$\psi${latex}

{latex}$\varepsilon${latex}

...

One can estimate the impact of the theta=70 deg cutoff by running gtexpcube2 with the option thmax=180 in the nominal case (this is the default) and thmax=70 in the truncation case. Assuming a photon index of -2, which Josh used in his gtobssim runs, one can estimate the expected fractional offset in measured flux (Fmeas) relative to the MC value (Fmc), (Fmeas - Fmc)/Fmc, using the spectrally weighted exposures, exposure

{latex}$= \left.\int \varepsilon^{-2} A(\varepsilon, \theta(t)) dt d\varepsilon\right/\int \varepsilon^{-2} d\varepsilon${latex}

...