Question Transcript
09:00am Kristin Madsen – Cross-calibration of Swift/NuSTAR
09:16:08 From Craig Markwardt : 100 pixels in NuSTAR is what in arcseconds?
09:16:32 From Brian W. Grefenstette : About four arc minutes. One pixel == 2.54 arc seconds.
09:17:19 From Brian W. Grefenstette : Will have to ask Kristin if she’s talking radii or diameters.
09:21:03 From Felix Fuerst : Why do the low energies drop for larger radii in the “halo metric” slide… I thought it was supposed the other way around (larger radii measure *more* soft photons)?
09:21:23 From Mike Nowak : I think its important to point out that you need to define a relationship between equivalent neutral column and halo optical depth when applying these models. Are there a uniform set of assumptions for this in the models shown in these fits?
09:23:15 From Andy Pollock : At what column density do you have to start worrying about dust halos ?
09:24:19 From Mike Nowak : Andy – depends upon energy range. Softer == lower column. Certainly at a few times 10^21 it shows up at low energy.
09:28:36 From Andy B : @AndyP I’ve seen indications of halo in SSS data at a few 10^21
09:31:17 From Mike Nowak : AndyP — if you want to see the effect of halo for a 5×10^21 source, see my Suzaku/Chandra-HETG comparison for Cyg X-1 in Nowak et al. (2011): https://ui.adsabs.harvard.edu/abs/2011ApJ…728…13N/abstract
09:31:52 From Kristin : @Andy Pollock: At around 1e22, but we want to check a lot more to get a better idea and find out if there is something else going on as well. I should mention that we have a similar problem with XMM timing mode/burst
09:29:40 From Richard Mushotzky : A simple check to see if this is the right idea is to compare fits to absorbed AGN where the dust halos are much smaller. If you get the same sort of offsets there is something else going on.
09:33:23 From Kristin : @Richard: I didn’t point out, as I should have that we get good fits with galactic sources with no haloes
09:33:46 From Kristin : So we have good fits with galactic sources, like Herx1, and a handful others
09:34:02 From Kristin : who are both bright and have low NH
09:24:09 From Javier Garcia : why are the fits so bad in the bottom panels of the second to last slide? if parameters are untie, shouldn’t you find a good fit for both detectors?
09:30:43 From Kristin : @javier : I said that wrong. I did not untie the model parameters, I just untied the cross norm and let the model fit
09:25:48 From Hugh Hudson : Fitting model parameters seems doomed to model uncertainties. Wouldn’t it be better to aim at extracting, from each instrument, some absolute property such as erg/cm^2 s erg sr at some reference energy?
09:28:54 From Jeremy Drake : @Hugh Hudson Yes, in principle, and that is what some of the cross-calibration work involves. It is less straightforward for silicon detector responses because of low energy resolution and significant wings and correlations over sometimes large energy ranges.
09:25 Craig Markwardt – Nicer cross-cal
09:35:07 From Michael Loewenstein : XMM fits to 3C273 used all three detectors, fitting them simultaneously with all parameters tied together. These were extracted by T.Toneyama.
09:35:38 From Brian W. Grefenstette : (NICER) Were all of the assumptions about the nH to the Crab the same between all analyses?
09:35:58 From Michael Wolff : Is the dust contribution to the spectrum patchy on the sky — are there certain directions in the galaxy that typically show significant dust and directions that show less dust?
09:36:12 From Kristin : Yes it is patchy
09:37:09 From Mike Nowak : MichaelW- As Kristin mentioned, dust can often come in multiple layers, e.g., correlated with galactic arms. So, halo shape & size can vary.
09:40:18 From Mike Nowak : MichaelW- also for dust, if your source is steady, you’ve got the whole halo, and it’s optical depth <~1, everything scattered out scatters back in, and your back to just N_H. For Chandra (and almost XMM), dust is a pure, additional loss term, correlated with N_H. Everything else inbetween is trying to figure out how much dust halos “put back in” to the spectrum.
09:39:07 From Hugh Hudson : Presumably the dust on the actual line of sight is also time-variable?
09:39:53 From Kristin : @hugh If the dust is local to the source it will be, but typically the clouds in the MW are not varying
09:42:11 From Mike Nowak : HughH – Timescales are 10’s ksec. So, for variable sources, it’s an issue. E.g., dips in Cyg X-1 and other “dipper” sources show time delays that might be relevant on the time scale of the observation. So, in principle, modeling dust halos might depend upon what your source was doing half a day earlier.
09:43:17 From Mike Nowak : And I should say that’s the time scales on the very innermost annuli.
09:37:34 From Herman Marshall : (NICER) was there a comparison of 3C 273 spectrum to the simultaneous Chandra data?
09:39:14 From Paul Plucinsky : With regards to the hard tail in RX J1856, see new paper on XMM and Chandra data by Dessert et al. 2020, ApJ, 904, 42.
09:44:00 From Jeremy Drake : Yes, on RXJ1856 the Dessert et al paper confirmed the hard X-rays appear to be from the source, and also present for RXJ0420.
09:46:13 From Paul Plucinsky : I would encourage all interested to read the Dessert et al. paper on the hard X-ray excess in RX J1856. It is a weak signal and how the systematic uncertainties are handled is an interesting question. Interesting for the IACHEC.
09:58:37 From Jeremy Drake : @Paul Plucinsky et al on RXJ1856 yes, I would second Paul’s recommendation – it is a detailed and careful analysis, though warrants scrutiny from instrument experts because the conclusions are important: there is no obvious source of the emission that fits, except axions…
09:59:43 From Brian W. Grefenstette : @Jeremy It’s always axions.
10:01:40 From Jeremy Drake : And not just axions, but axions + magnetic fields!
10:07:05 From Herman Marshall : @Jeremy, @Paul P: If I remember correctly, Dessert et al dispatched the pileup issue by running marx models. It would be useful to know if such modeling adequately reflects the actual instrument performance. Are there even any flight data that could be used to test this use of the simulations?
10:09:36 From Paul Plucinsky : @Jeremy, on Dessert et al, I quote from their abstract: “We analyze possible systematic effects that could generate such spurious signals, such as nearby X-ray point sources and pileup of soft X-rays, but we find that the hard X-ray excesses are robust to these systematics to the extent that is possible to test.” The key phrase is “to the extent that it is possible to test.” They rely on an accurate pileup correction for ACIS-S/LETG data. I believe there is more work that can be done for this case. They are limited by the accuracy and applicability of the current pileup model.
10:19:49 From Jeremy Drake : Yes, @Herman, @Paul, agree. This could merit an IACHEC special investigation.
09:42:40 From Andy B : NICER – is the 3c273 double powerlaw model actually a better fit than zbody+powerlaw ? I found the latter gives a more consistent NH (with galactic) when tried on XMM and Swift data
09:44:03 From Herman Marshall : Andy B: I’ve regularly used a double PL model for fitting 3C 273 as well, for Chandra LETG/ACIS observations from 0.25 to 10 keV.
09:46:06 From Craig Markwardt : @Andy, I think Mike’L’s model was phenomenological for cross comparison, could probably be improved.
09:50:04 From Michael Loewenstein : Craig, Andy, Herman: the choice was strictly phenomenological/empirical (the simplest model that I thought of – which just happened to work). We could try more physical models as well…
09:40 Michael Smith – Cross-calibration of XMM/Chandra
09:56:30 From Andy B : @MikeNowak – on the subject of halos – can a Wing-to-core PSF ratio tell me something about the time delays, especially for a variable source ?
10:05:01 From Mike Nowak : AndyB — Yes, but it’s always a matter of S/N. See Fig. 2 of Xiang et al (2007) for a lightcurve model of a Chandra-HETG observation of a dipping source. Time delays add to be built in between the gratings lightcurve and the CCD lightcurve. But that’s a very simple two zone model (point source HETG, distributed source CCD), due to S/N issues. At the opposite extreme, Kristin showed lovely Swift movies of outbursts in V404 Cyg, where you see the dust echo propagating outward. Lynx, should it happen, probably could do some really cool intra-observation stuff. Xiang et al. (2007): https://ui.adsabs.harvard.edu/abs/2007ApJ…660.1309X/abstract
10:00 Richard Saxton :What is pileup and how should it be correctly dealt with in data?
10:16:00 From Michael Wolff : What pattern types are the best calibrated in the EPIC instruments: Singles only, or, singles + doubles or is it more complicated?
10:19:19 From Michael Wolff : And I should have also said that my question was for low flux (F < 10^-12 ) sources.
10:28:11 From Brian W. Grefenstette : (PILE-UP) If you were going to estimate the systematic errors introduced by doing the RMF-correct to pile-up vs excising the core, which do you thing is larger?
10:29:49 From James Steiner : (Pileup) For the very interesting rmf-modifying approach, because you are relying on interaction of one energy with another, it seems this effectively modify the Aeff as well. Do the rmf columns sum values other than 1, or do you introduce something new to account for this?
10:36:46 From Richard Saxton : I didn’t have time to say it but in reality because you add an extra event you increase the count rate. To remedy this you have to randomly subtract one event from each frame before reading out the piled events.
10:38:00 From Richard Saxton : The RMF rows typically sum to <1
10:38:26 From Andy B : The XRT rmfs are generated from a monte carlo simulation code which natural simulates pile-up. I’ve often wondered whether I should generate RMFs which are deliberately created for bright/piled-up souces so we don’t have to throw data away
10:42:00 From James Steiner : @Andy – Trying that out gets my vote!
10:45 Eric Miller – Background modeling
11:09:37 From Craig Markwardt : Eric, What response matrix is used in NXB spectral fit?
11:14:37 From Herman Marshall : @Eric on Suzaku background: is there a way to predict the slope of the power law component of the NXB spectrum?
11:15:25 From kuntz : What does the long term variation of the NXB look like?
11:19:15 From Eric Miller : @Kip (assuming that’s Kip): In high-inclination low-Earth orbit, there is obviously a huge variation over the orbit due to varying cut-off rigidity. We do see Solar cycle variability as well, but it really appears to just be an overall normalization rather than spectra variation, with a caveat that I haven’t explored it in detail yet.
11:15:42 From Richard Mushotzky : How do you take the variation in spectrum and flux of the soft x-ray background (E<2 keV) into account (e.g the Rosat maps)
11:20:33 From Eric Miller : @Rich: Mike is right. I haven’t attacked that yet, but you would have to use a local offset region and allow some reasonable priors on the spatial and spectral variation.
11:17:32 From Mike Nowak : Richard M: If you fit the background directly, then that should allow for variation of the soft X-ray model. You could go more fancy by imposing a prior on the allowed parameters for said fit.
11:19:09 From Gulab Dewangan : (NXB) Eric, are you assuming no X-ray emission at all from night Earth? Cosmic rays interacting with Earth may be giving rise to some line emission?
11:22:37 From Eric Miller : @Gulab: For the 6 keV range I showed, I’m assuming no X-ray emission. I think that is safe for this case as it should be well below the NXB and unresolved AGN CXB, but I should verify that.
11:20:11 From Varun Bhalerao : @Eric can you give some reference about the cut-off rigidity?
11:25:25 From Eric Miller : @Varun: This somewhat dated paper describes the variation of the Suzaku background with COR: http://adsabs.harvard.edu/abs/2008PASJ…60S..11T That uses an older geomagnetic cut-off rigidity model which has been updated since then.
11:27:29 From Koji Mukai : @Eric: one thing I worry about the COR is that we tend to use static, 2D maps for it. In reality, the geomagnetic field is three-dimensional and time variable …
11:33:20 From Eric Miller : @Koji: That’s right. I believe there have been some COR model updates in HEASOFT recently, but I think this is still an issue. Maybe Mike L knows a bit more about this. It also seems that we can use in-situ spacecraft measurements like the HXD PIN on Suzaku, which seemed a better metric than COR.
11:34:11 From Craig Markwardt : (Eric & Koji) – the output of the “prefilter” tool (COR_SAX column) is a full computation of COR based on the magnetosphere. And I keep the magnetic field model updated (updaes every 5 years).
11:00 Herman Marshall – Concordance results
11:31:14 From Richard Mushotzky : Are there results for AXCIS-S and ACIS-I?
11:32:15 From Javier Garcia : @herman: can you quote a number for the wavelength accuracy of the HETG
11:32:15 From Andy Pollock : Why do you work in log space ?
11:33:35 From J. Michael Burgess : What family of MCMC is used to compute the posterior?
11:34:30 From Vinay Kashyap : @MichaelBurgess, we used Stan/HMC
11:36:06 From Andy B : @Herman – the XRT PC data on E0102 are slightly piled-up, hence the line norms are suppressed somewhat compared to the WT data. You should factor that into your model…
11:37:23 From James Steiner : @kristen – if you or others have any practical lessons learned about xscat to impart, that would be useful to hear. (E.g., does the default dust model work best?)
11:15 Kristin Madsen – How can we help the Community; how can the Community help us?
11:42:35 From James Steiner : Comparison to a gold standard would be good to see
11:46:48 From Catherine (she/her) Grant : We have looked at the variation on Chandra. Decided it was too small to bother with.
11:49:41 From Andy B : Calibration updates/improvements happen. Do the now “old” papers on E0102 or G21.5 need to be revisited?
11:53:37 From Herman Marshall : @Andy, I think we definitely should be reviewing cross-cal results regularly. I think the IACHEC working groups do examine particular calibration source results regularly.
11:56:13 From Paul Plucinsky : @Andy B for the Chandra data of E0102 we have not changed the calibration in a way that would affect the results before 2016 that were included in the paper. The biggest issue for Chandra is the contamination model and we currently make sure that a revision to the contamination model does not change results earlier in the mission. Now, in the future, we might decide that we do need to change the calibration before 2016, but at the moment we have not made such a decision.
11:53:53 From J. Michael Burgess : What about a GitHub repo / org with folders which can also have a DOI?
12:02:06 From Vinay Kashyap : https://iachec.org/calibration-statistics/#2020dec1
12:04:00 From Karl Forster : The IACHEC supports new missions with coordination of cross calibration observations between active observatories (and was very helpful in the initial in-flight calibrations of NuSTAR). So it in the interests of future missions (IXPE, XRISM, Athena, etc.) to have a vibrant and active community of calibration scientists. So hopefully the support for IACHEC will continue as long as new missions continue to fly!
12:10:20 From Adam Foster : IACHEC Slack invite: https://join.slack.com/t/iachec/shared_invite/zt-jdpqd0jq-svDze~jP4lbw8nHKpJqIyQ