Secondly, I have tripped over something in MARX that doesn't seem right. I've been employing MARX in an attempt to understand the role of the cluster on the extracted AGN spectrum. But, even when I just simulate an isolated point source with a pure power-law spectrum, the spectrum I then recover has wiggles at the ±10% level (see marx.ps attached). So MARX doesn't seem to be internally consistent with CALDB. Is this unavoidable given that (I believe) MARX uses a physical model that must presumably differ in detail from CALDB? Or is there someway of creating and then extracting a completely self-consistent simulated HETG spectrum from MARX?
Debugging indicates that this is due to MARX not using QEU maps. In particular the BI chips have up to 20% drops near the chip edge. MARX does not use QEU maps, and all observations should be reduced with the ardlib qualifier UNIFORM. It turns out that fullgarf does not offer the option to do that, introducing these spurious wiggles.
Two options to solve this:
update fullgarf
have a way for MARX to include non-ideal detector effects: bad pixel maps, dead time correction, QEU
Option 1 is much faster and should solve this problem, but option 2 needs to be investigated long-term.
This bug report came to me third hand:
Secondly, I have tripped over something in MARX that doesn't seem right. I've been employing MARX in an attempt to understand the role of the cluster on the extracted AGN spectrum. But, even when I just simulate an isolated point source with a pure power-law spectrum, the spectrum I then recover has wiggles at the ±10% level (see marx.ps attached). So MARX doesn't seem to be internally consistent with CALDB. Is this unavoidable given that (I believe) MARX uses a physical model that must presumably differ in detail from CALDB? Or is there someway of creating and then extracting a completely self-consistent simulated HETG spectrum from MARX?
Debugging indicates that this is due to MARX not using QEU maps. In particular the BI chips have up to 20% drops near the chip edge. MARX does not use QEU maps, and all observations should be reduced with the ardlib qualifier
UNIFORM
. It turns out that fullgarf does not offer the option to do that, introducing these spurious wiggles.Two options to solve this:
Option 1 is much faster and should solve this problem, but option 2 needs to be investigated long-term.