Notes on zeta Ori line profile draft for coauthors: - Table 2, showing the Gaussian fits - I did these fits quite a while ago, using CIAO/Sherpa. The results certainly are OK (though I used the "uncertainty" command to get the errors on the fitted parameters; doing a "steppar" in XSPEC might be more accurate). I could redo this fitting (in XSPEC) and I could fit the same lines we fit the wind profile model to (including line blends fit simultaneously). This would make the results more comparable to the wind profile results. In fact, Kevin and I have done all this already, but haven't run steppar in order to get confidence limits. Think it's worth doing some or all of this? A separate but related issue is whether we should list monte carlo C statistic-based rejection probabilities for these fits (and maybe for the Gaussian fits with a fixed centroid). Note that this information is already contained in the text of the paper for the stron lines at 18.97 and 15.01. And we report that, overall, the rejection probabilities got better for 10 out of our 11 lines when we fit the wind profile model. - I've used R_min rather than R_o, which was the notation used in Owocki and Cohen (2001), but I define all the variables, so I think this is OK. - Note the discussion of the wind opacity in sec. 5. The model shown in my 1996 paper is for an early B giant, epsilon CMa. (I've put the relevant figure from that paper on my site: http://astro.swarthmore.edu/~cohen/projects/zori/wind_opacity_Cohenetal1996.rtf if you want to have a look.) But there's nothing unusual about the abundances used in that calculation; in fact, for that old model I assumed "standard" B star abundances. Who knows what the abundances are in the wind of zeta Ori. The values used in that old calculation are probably a reasonable guess. And the details of the wavelength-dependent wind opacity really then just depend on the ionization balance. This can't be radically off, and even big changes really just have the effect of shifting the wavelength of various ionization edges. I recently did another simulation using Joe MacFarlane's PrismSpect code, the results of which are at: http://astro.swarthmore.edu/~cohen/projects/zori/sample_wind_opacity.jpg (note that the units on the y-axis are mislabelled; they should be cm^2/g). This new modeling is very preliminary. It does not account for spatial variation in the wind, nor does it account for x-ray photoionization (and its effect on the ionization distribution). Again, though, that'll just shift the locations and relative strengths of some edges around a bit. I'd like to follow up these preliminary models with some more sophisticated ones, but that won't happen for a little while (I'm waiting for an updated iron model from Joe, not that that'll make much of a difference). We could discuss some more of this in the paper, but I think the way it is now - just referencing the 1996 paper, is fine. I'm open to suggestions though. - What do you think of the clumping discussion? Do I skirt to many issues here? I'll note here that the Bouret et al. work assumes a quite extreme clumping factor (0.02 - 0.04). - Discussion of numerical simulations of the wind instability - Glenn Cooper's dissertation is referenced. Think that's OK? It's available online and I could include the URL in the bibliography. Not sure if I should keep the reference to Cohen et al. 1996, where one simulation snapshot is shown. For the low density wind of epsilon CMa, the onset of shocks (albeit weak ones) close to the photosphere might be physical or it might be numerical. ...I am without access to most online journals here at the beach...I can't recall if Feldmeier has a paper earlier than the 1997 work that shows self-excited instability structure, which I could reference. Similarly, I've referenced Stan's paper with Marc Runacres (the first refereed one), but don't have a copy in front of me. I know that statistical analysis of the inner wind shows a slight decrease in shock velocity dispersion (and thus temperature) with radius, which I've invoked in the discussion of (the lack of) trends from line to line in our data. But I'm not sure the paper shows this explicitly. I'll check into this when I return to my office, but any feedback on this would be welcome. - It looks like Alex Fullerton and Derck Massa still have not published their work on FUSE observations of P V and the problem of M-dot overestimates. I'll contact Alex next week and see what's up. For now, I'm referencing their AAS poster from the San Diego meeting (which I studied pretty carefully at the time). - Maurice: O VII at 21 A - fit/data look a little weird. Another feauture between the two components? What do you think? - Delta C = 3.53 vs. 1 or 4 (3 "parameters of interest" vs. 1; should I describe further what I actually did - esp. how the maximum extent of each parameter could be thought of, taken all together, as describing a cube, but that this cube in parameter space contains far more than 68% of all realizations. In other words, the confidence limits on any one parameter, assume correlations with the other parameters.) You can see it's hard to describe this stuff clearly! Do you experts - Maurice and Roban - think this is OK? - Take a look at the linear fits to the derived parameter values on my website. Not worth showing - or even discussing these in any more detail than we already do, right? Note there's already a brief mention to the linear fits and consistency with zero slope. The figures are at: http://astro.swarthmore.edu/~cohen/projects/zori/Rmin-stats.jpg http://astro.swarthmore.edu/~cohen/projects/zori/q-stats.jpg http://astro.swarthmore.edu/~cohen/projects/zori/taustar-stats.jpg * Is my (implied and explicit) criticism of Waldron and Cassinelli too much? How should I modify it?