Paolo Mazzali mazzali@ts.astro.it I hope this is the good email address Deng From deng@astron.s.u-tokyo.ac.jp %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% David J. Jeffery Department of Physics New Mexico Tech 2002 February 25 Hi Paolo: Ken tells me you are in land of the rising Sun working on the 02ap spectra. My first look at the spectropolarimetry has not led to any revelation. Except there's intrinsic polarization order-of .5 % which you-all already knew. How are things going with spectra and life? Best Regards David P.S. How come emails to your Italian address mazzali@circe.oat.ts.astro.it never get through. I'll try mazzali@ts.astro.it this time. From mazzali@circe.oat.ts.astro.it Tue Feb 26 23:47:28 2002 From: Paolo Mazzali Hi David, very odd indeed. maybe you don't have to use the workstation name... usually oat.ts.etc or just ts.etc work fine. anyway, yes, i'm in tokyo, enjoying every second of it. 8000 miles is about the minimum distance from that place. i won't bother you with more lamenting, i'll tell you whenever we meet next. but all in all things are ok, i'd just like to solve the job problem, meaning i have to find a way to leave trieste, or better still italy. 2002ap. i'm very happy that you're becoming involved with this. i think it's an interesting one, but so far my spectra don't seem to indicate excessively large masses or energies (i mean it's at most 10foe, maybe less). i'm getting good synthetic spectra, but i'm mostly trying to id which kind of model works best, and thats not easy. The SN makes a transition from 98bw-like to 97ef-like, which is in itself interesting, plus it's got a rapid - and rather dim - LC. i'm not sure what we can get from the polarisation, i'm really not an expert in that field. any hints? so, i should be here until the end of march, i hope we can get some work done! cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 February 27 Hi Paolo: I'm sort of an old, out of date expert in polarization. My last publication on the subject was in 1991. However, last summer my interest was revived and I've a new (?) jet model of polarization which is probably wrong, but is certainly easy to calculate with. Anyway tomorrow I'll try modeling it with the jet model. The 1st real question is figuring out the interstellar polarization to subtract off. If the asymmetry is just bulk asymmetry as in the Hoeflich models, then the supernova may be ellipsoidal at the 20--40 % range. I can't compete with Peter anymore on that kind of modeling. But anyway I gather Ken is aiming at a letter in which the polarization will just be one of many facets. I'll let you know what I come up with in a day or so. Ah, Trieste. I was just reading a biography of Schroedinger. Did you know he missed studying with Boltzmann because Boltzmann committed in suicide at Duino Bay in 1906. And Rilke wrote famous Duino Elegies---don't really cotton to Rilke myself. Anyway Schroedinger spent a fair bit of the Great War with a commanding a artillery battery in the hills overlooking Trieste---but the text fails to mention whether he was shelling into the city or defending it. Wars come and go: physics is eternal. Thanks David From mazzali@circe.oat.ts.astro.it Thu Feb 28 02:50:52 2002 Hi David, i hope you get something. i'm puzzled by this jump near the CaII IR triplet. i've no idea what it may mean. the average polarisation, yes, one can interpret as asymmetry, but could it also be explained by some kind of clumping? trieste. boltzmann's story is my introduction to the city to all visitors. for me it's becoming a serious problem. i get depressedevery time i spend more than a few days there. too many bad memnories, a hostile environment. home is where you find yourself at home, i'm afraid. shelling.... during the bosnia war i was hoping that some B52 would 'lose' some heavy-duty package... now i'm thinking of passing the info that Bin Laden has been spotted around the city... cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 March 2 Hi Paolo: Ken asked for an update on my analysis efforts. So I've detailed this below. Any enlightenment you can give me would be welcome. By the way can you send me anything about what your analysis is leading to: e.g., estimate of reddening, date of explosion, photospheric velocities, line identifications. I'm afraid that the hypernovae have mostly passed me by: I'm not as au courant as I should be. I'm afraid the supernova polarization is not leading to one consistent believable picture rapidly. I'll send 4 postscript figures after this analysis message. I'll also send this material to Paolo for his comments. Koji tells me that only Feb 8 and 9 polarization data can be fully trusted and this agrees with what I see since the other two days are inconsistent with Feb 8 and 9. Since Feb 8 has the best wavelength coverage I've concentrated on that day. Figure 1 is an overlay of observed relative flux, polarization, and position angle spectra: the latter two box-car smoothed over 100 A. I've corrected for foreground reddening (E(B-V)=0.075 from Leda data base) and heliocentric velocity shift (v=631 km/s from narrow lines in the SN which Koji approves). But I've not corrected for interstellar polarization (ISP) which is hard to decide on: see below. There are several remarkable points about Figure 1. First, there is a polarization continuum without obvious features, except the big peak across the Ca II triplet drop: I'm assuming that most of the wiggles are still noise even after 100 A box-car smoothing. The Ca II line feature proves there is intrinsic polarization. But I would have expected there to line features associated with some of the other broad absorptions: not as deep as the Ca II feature, but distinct---the Ca II flux absorption is a 50 % drop from the nearby continuum, the other ``absorptions'' are relatively smaller. That there is only one feature is a puzzle. It's possible that the continuum is mostly ISP and the other features are small because depolarizing line opacity in the supernova kills them. (I'm assuming that the electron scattering is the polarizing agent.) But if the ISP is small, then the continuum polarization is intrinsic and one should have more line features. The position angle (PA) on Figure 1 is very constant which is good. But there is some variation over the whole continuum and over the Ca II line polarization peak. To zeroth order intrinsic polarization PA should be constant (if electron scattering is the source which I believe) no matter what the asymmetry since electron scattering is wavelength independent: all flux gets polarized the same way if lines don't interfere. If lines do interfere with polarization, then PA can exhibit shifts unless the asymmetry is axisymmetric: 90 degree shifts can occur in axisymmetry, but perhaps are hard to arrange. Any significant ISP should cause PA to vary strongly over the continuum. This is clear on a QU Stokes diagram (see Fig. 2). Intrinsic polarization with constant PA on the diagram would just be a radial line segment pointed toward the origin. Any ISP would be a vector (Q_isp,U_isp) translating the QU line off the radial line to the origin. Since PA is given by $$ PA=.5*atan(U/Q) , $$ PA would vary with an ISP translation---unless the ISP was accidently colinear with the intrinsic polarization. That the observed PA is fairly constant, except through the Ca II line, suggests that ISP is smallish and the SN asymmetry is somewhat non-axisymmetric. Unfortunately, this contradicts the evidence of the lack line polarization features other than for Ca II. Maybe the resolution is the ISP is accidently aligned with the intrinsic polarization. Or maybe it's some messy mixture of effects of near alignment and weakish, but not negligible ISP (relative to the intrinsic polarization). Of course, maybe I'm all wrong. Figure 2 is the QU plot for Feb. 8. The data have been binned in 200 A intervals from 4200 A to 8000 A: I've left of the far ends of the QU spectra supposing them to be less certain. ---I think all the points shown are significant, except maybe near the shown ends: but observers must tell me how good the ends (shown or unshown) of the QU spectra are. The low wavelength end is where all the little loops are: the big quadrilateral is the Ca II line: the two regions are inexplicably joined by a somewhat out of the trend zone that reappears after the Ca II line. No ISP translation is going to turn this QU curve into a nice radial line segment, but it's actually confined to a radial cone of 40 degrees (20 degrees in PA) which is not so bad in arguing for near constant intrinsic PA. If the ISP correction moved the QU curve further away from the origin in the negative Q direction, then the intrinsic PA would become better more constant. That would be a coincidence on the part of the ISP---but stranger things have happened. I've tried modeling the polarization spectrum with my bipolar jet polarization model. This model assumes electron scattering polarizing jets that are well detached from the photosphere. It is a purely axisymmetric model, and the synthetic position angle is a constant aligned perpendicular to the jet direction. It's a very, very simple model. Figure 3 shows synthetic polarization spectrum (dashed line) consistent with the model's assumptions. I took the SN photosphere to be at beta=0.1 (velocity in units of c) and the jet to be detached to beta=0.2. This means that there is a huge redshift and high polarization can only form to the red of the broad continuum maximum. I've adjusted the parameters to fit the Ca II line. The polarization spectrum starts only at 5200 A because it used the observed flux spectrum as an input and with the big redshift the pushes the start of the polarization spectrum far to the red. The polarization spikes in the synthetic spectrum are because of the telluric lines: I'd get rid of those in any final figure. The parameters of the model can be played with a little, but for a parameter set consistent with well detached jets, I don't think I can fit the Ca II line and the continuum. So either the well-detached jet model is wrong or else ISP forms the continuum (which is consistent with the lack of other polarization lines beside Ca II), but is inconsistent with the fairly constant PA (unless the ISP is accidently highly aligned with the intrinsic polarization which is possible but somewhat unlikely). The jets could, of course, be closer to the SN photosphere (or embedded in it), and so require less of a redshift, but then my simple formulae would only be heuristic and not offer even a semi-accurate way of determining asymmetry parameters. The ellipsoidal asymmetry modelers I think have some of the same dilemmas I'm facing. They should get more line polarization features or no continuum polarization blueward of Ca II line unless the ISP provided the continuum polarization, but in which case the ISP would have to be accidently closely aligned with the intrinsic ISP to prevent large PA variation. Figure 4 shows a better synthetic fit to the polarization spectrum with the bipolar jet model. But this fit is purely heuristic since I have to embed the jets inside the photosphere. As aforementioned this is physically not impossible---the jet model then becomes not so distinct from the ellipsoidal asymmetry model. But also as aforementioned my fitting parameters become somewhat meaningless because my simple formulae assume a well detached jet as a simplifying hypothesis. As you can see I get more line features. There might be some correlation between my line features and the observed bumps in the 4000--7000 A region, but it doesn't look significant to me. Even though my model is only heuristically good for this set of input parameters, I think those line features are hard to avoid for any realistic model unless the continuum polarization is mostly ISP accidently aligned with the intrinsic SN polarization. So firm conclusions elude me, except that there is intrinsic polarization---which you already knew. I'd like to say that there is ISP accidently aligned with the SN intrinsic polarization, and thus save my well-detached bipolar jet model---but that would also save the ellipsoidal models. My analysis is pretty tentative though. Maybe someone can explain the observations without largish ISP accidently aligned with the intrinsic SN polarization. More spectropolarimetry from other epochs would help---or at least so one can say before it comes along and confuses the issue still more. Best Regards David From mazzali@supernova.astron.s.u-tokyo.ac.jp Sun Mar 03 20:46:40 2002 Subject: ap Hi David, thanks for the plots. i think it's interesting. even though your model does not exactly apply to that situation, the 'embedded jet' case is interesting. how different would that be from just a prolate ellipsoid? me, i've now got some decent results on the first spectrum. it's taken me quite some time, but i now think a good model is CO100 (which we used for 97ef) but with 1/4 of the mass, ie 2.5Mo, and a KE of 4.5foe, roughly. there is a firm outer edge to the matter distribution at about 60000km/s, else CaII IR gets too blue. i prefer an epoch of 2 days for the 1st spectrum, thus setting t=0 at about jan 29.0 UT. as for the reddening we've derived and are using Eb-v=0.08, from the IS (interstellar lines) lines and various formulae, including my unpublished and never-to-be-published heuristic SN reddening one - which gives a good result i should note. i think we have quite a lot here, and what you say is already worth publishing. so i think we should start finalising all of this. cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 March 3 Hi Paolo: I think the embedded jet has to be regarded as just heuristic: translation to English ``phoney''. I can't see how to map it into any of the ellipsoidal models. One can map it into polarization from pre-existing dust clouds as in the Wang-Wheeler model or flash-ionized clumps circumstellar material from the late evolution which may not be a bad idea though I'd have to be Peter Lundqvist to work out the numbers. In those cases there would be no need for redshifts at all and the optical depth in electron scattering might be relatively constant since the clumps or clouds would be relatively unchanging in time. The original idea by the bipolar jet model was to keep the bulk of the ejecta spherical. David Branch's Ib spectral analysis paper suggested that many Ib's should be highly alike and spherical therefore---and yet one presumes that they exhibit polarization like other core collapse events. There isn't a lot of spectropolarimetry data for Ib's, but Marshall McCall's unpublished 83N data showed a polarization line feature. By putting the polarizing outside the bulk supernova, I could keep the geometry simple. Essentially a point source and a point scatterer with a few correction factors of marginal validity. Anyway I thought all you hypernova people were going to give us jets and gamma-ray bursts. Your E(B-V) is virtually the foreground value of .075 so the SN is mostly unextincted in the host galaxy. Alas an empirical law says that ISP can be has high as 0.65 % with that amount of reddening: it could still be nearly zero of course. I don't think we can get a better grasp on the ISP without more polarimetry. So for an observational paper, I'd say the analysis I've done so far for the polarization may be about as far as one should go. I could write something up that's much shorter and less rambling than what I sent you-all for a section. I'd mention the jet model and cloud/circumstellar variations and ellipsoidal asymmetry relying on Peter H's results. I take it Ken is aiming at a letter. Tchues David David J. Jeffery Department of Physics New Mexico Tech 2002 March 3 Hi Paolo: I take back what I just said about clouds and ionized clumps. They would give a pretty constant polarization without line features unless the clouds were far enough away to give time delay effects which what Wang-Wheeler invoked. Tchues David From mazzali@supernova.astron.s.u-tokyo.ac.jp Sun Mar 03 23:23:42 2002 Hi David, my ignorance about polarisation is unfathomable. so whatever you say is gospel. i don't yet see evidence of asymmetry in the ejecta, so the 20% lever suits me. i just got a spectrum on the 21st, and it shows CaII emission. something i had predicted almost from the outset. the quick development (that was the prediction) testifies that the mass is small. so i'm quite pleased with myself. i think ken is aiming for a letter, yes, but we must wait for a decision on the part of the various observing groups. but then we already have material for one or possibly more main journal papers, depending on whether or not we want to address different issues in different papers. all that will have to be decided at some point soon. i do agree that polarisation may not tell us much more than what you're finding, (because i know nothing about it!) so we should certainly publish that. cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 March 3 Hi Paolo: There is just too much degeneracy in the polarization-to-asymmetry problem. What we need is a fast multi-dimensional NLTE code that by-passes all the hard work. I've just been glancing at Lucy's A&A paper on Monte-Carlo NLTE. Maybe that is the way of the future. Polarization could be incorporated at the later steps of the photon. It's really only the last 3 or scatterings before escape that determine polarization. Tchues David From mazzali@supernova.astron.s.u-tokyo.ac.jp Sun Mar 03 23:40:15 2002 From: Paolo Mazzali Hi David, well if we could have that! but it's a dream for now, or rather something we're working towards. i have a couple of students here and in munich who are building some of the necessary things, but there's a long way to go. i wouldn't make predictions as to how far in the future this lies. so i take it the detached shell must be opt. thick in electron scattering, so that every photon scatters at least once, right? cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 March 4 Hi Paolo: It's good to know someone is building a better code. The trouble with polarization is degeneracy (many imaginable asymmetries giving the same spectrum) and that it's so hard to calculate except in very parameterized ways like me. For maximum polarization with my model, the best thing is to have the jet optically thick radially and thin perpendicular to radially. This way photons tend to scatter in a non-random way relative to the supernova and so give a high net polarization. But actually I don't have a problem getting high enough polarization with the jet model for 2002ap, but rather getting it in the right wavelength place. Tchues David From mazzali@supernova.astron.s.u-tokyo.ac.jp Tue Mar 05 00:04:55 2002 Hi David, `building' is too big a word! `dreaming about' is more realistic. anyway, i suppose in that picture you also have some clue about the orientation of the jet relative to the LoS have you? cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 March 4 Hi Paolo: Not really. The orientation of the symmetry axis (if there is one) is problem for all axisymmetric models. The smaller the angle to the line of sight, the more asymmetry you need to get the same polarization. Except that probably we arn't looking very close to a symmetry axis (which would take huge asymmmetry to get 0.5 % polarization), there is no simple way of determining orientation. Detailed modeling a la Hoeflich with both flux and polarization treated jointly has a chance, but even then I think there's still a lot of degeneracy. Buenas naches. Tchues David David J. Jeffery Department of Physics New Mexico Tech 2002 March 19 Hi Paolo: I havn't heard anything from your side of the world lately. Has there been any progress on 2002ap or plan to write the paper of which I'm supposed to contribute a section? Tchues David From mazzali@supernova.astron.s.u-tokyo.ac.jp Tue Mar 19 23:11:02 2002 From: Paolo Mazzali Subject: Re: Your Message Sent on Tue, 19 Mar 2002 22:56:59 -0700 To: jeffery@kestrel.nmt.edu Hi David, yes, there has been progress. we have defined a range of parameters for the possible model using spectral fits, and got a LC . i am not convinced this is really the final word, but Ken is putting an enormous amount of pressure on me to write this up and send it as a letter. about 1/2 of the letter will be taken up by the names of millions of observers coauthors and UoTokyo ghosts... polarisation is completely out of this because the Subaru lot don't want to release the data until they publish them. i asked Ken about your contribution, and he says you will place this in the paper on polarisation. i don't know what else to say. what is your understanding of this? i would prefer for the two of us to do some spectrscopy/polarimetry paper together. what do you think? cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 March 19 Hi Paolo: Ken's plan is probably best. I could write up whatever I have to say in the observer's paper which isn't a lot. Probably the interstellar polarization is of order 0.2% perhaps roughly aligned with the supernova polarization. If the supernova has ellipsoidal asymmetry then the Hoeflich models would suggest 20--30 % asymmetry. My jet polarization model is not likely to be right---not for this supernova anyway. Polarization always just messes things up: best to put it harmlessly out of the way. I don't know if there is enough in the polarimetry that I can extract to make much of it in a spectra/spectropolarimetry paper. In fact the data are being recalcitrant. I think it's clear whatever the asymmetry it's not strictly axisymmetry: globs and elongation. It's also not very rich: only one significant line polarization feature---the same one shifted---at two epochs a month apart. Off to bed. Tchues David From mazzali@supernova.astron.s.u-tokyo.ac.jp Tue Mar 19 23:38:07 2002 Hi David, as you like. i see some 'strange' things in those spectra, like Fe-peak elements' abundance decreasing inwards... may be signs of a HN near the jet axis. this will not be discussed in this letter, but when i bring it up in a bigger paper it would be nice to compare to asymmetry predictions. also, i hope soon enough nebular spectra will give us some hint re. clumping. another odd thing is the behaviour of CaII IR. Alex swears the big abs is OI and the weak one CaII. i have a different idea: CaII is both. does polarisation tell us that the strongly polarised line is CaII? if so , that is useful. cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 March 20 Hi Paolo: Actually I only seen spectra from Feb 8--11 and March 11. From those two epochs it looked like a vary fast Ca II decreased rapidly and left the O I to show up only on the latter day. The polarization peak seems to have just followed the retreating Ca II line. Could you send me a figure with spectral evolution. Tchues David From mazzali@supernova.astron.s.u-tokyo.ac.jp Thu Mar 21 00:10:16 2002 Hi David, i've asked Deng, our postdoc here, to send you that plot. you'll find the evolution of the O/Ca feature interesting. let me know your opinion on it. cheers paolo From mazzali@supernova.astron.s.u-tokyo.ac.jp Thu Mar 21 01:56:11 2002 Hi David, what i mean is look at the apparent shift from feb 4 to feb 6. there seems to be a weak CaII low velocity abs forming, which then just keeps increasing in strength. the question is can polarisazion follow that. of course there are no data. i think there must be a double component to the Ca line Deng fitted that well with synow, but then you can get anything from synow... so what do you think? paolo David J. Jeffery Department of Physics New Mexico Tech 2002 March 21 Hi Paolo: Sorry for my slow response. It's been another busy---and late day. Unfortunately, I could print the spectrum figure Deng sent. It looks like ordinary postscript but it gives an error printing or ghostviewing. Can you ask him to try again. Your last message was a bit cryptic. I'm puzzling over what ``o.c.'' is. Tchues David David J. Jeffery Department of Physics New Mexico Tech 2002 March 21 Hi Deng: I have to admit I'm unable to read an attachment. Could you please send the figure just by old-fashioned mail in postscript so it can be read without interpretation. Sorry about this. Best Regards David From mazzali@supernova.astron.s.u-tokyo.ac.jp Sat Mar 23 20:18:36 2002 Hi David, sorry for the silence. i was just too busy. today is my last day, and i want to submit the paper. i hope you could finally see that figure. i'm leaving tomorrow, but won't be back in trieste for another couple of weeks, that place affects my sanity. i should be on email contact anyway. cheers paolo ps o.c. is, o.c., of course David J. Jeffery Department of Physics New Mexico Tech 2002 March 21 Hi Paolo: No, I never saw the figure. Your friend Jinsoon (?) probably gave up in disgust at person who couldn't decode an attachment. Why not send me a postscript of your paper---by ordinary UNIX email so that it comes across without undecipherables. I havn't had a chance to get a real overview of this supernova. I am working on a little paper that I hoped you could glance and tell me if it was rightish (probably) and novel (probably not). Not quite ready yet, but if you're willing I could mail it along sometime soon. Tchues David From mazzali@supernova.astron.s.u-tokyo.ac.jp Sun Mar 24 01:37:17 2002 From: Paolo Mazzali Hi David. i can only try. i looked at the file, it looks funny, it occupies the bottom lh corner of the page, but at least you should see it. paper's gone, off tomorrow cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 March 26 Hi Paolo: I did get the figure and yes the evolution looks stange to me, but you know I've never tried modeling these fast Ic's myself and so don't have a great intuition. But yes, the Ca II does seem to have a double component around Feb 8. But maybe it's just O I turning on and Ca II fading back in some combined way. But you've already been all over that ground. I don't think polarization can tell use too much. The asymmetry can't be exactly axisymmetric, but on the other hand it probably isn't wildy non-axisymmetric---some axisymmetry. The shape of the polarization profiles I'd say is wrong for my jet model. The polarization maxima are strongest on the blue edge of the line absorption trough: this is just like 87A. This blue edge maxima suggest to me that the strong polarization arises at wavelengths where the Sobolev optical depth of the line is falling below 1 in the CD plane. So it seems that the polarized flux is originating in the bulk ejecta itself and be internally depolarized by the line until the line suddenly starts becoming optically thin, but still reducing unpolarized light from the photosphere. So I think that probably bulk asymmetry must cause the polarization. One could still claim a jet emerged (and there is some evidence that that happens), but one can't account for all the polarization that way and so it's not a great analysis tool. Anyway I intend to get to work on the polarization data this weekend: teaching and bureaucracy is keeping me occupied, not to mention preoccupied. Tchues David %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% David J. Jeffery Department of Physics New Mexico Tech 2002 March 31 Hi Paolo: Do have an estimate of the optical depths of the lines near the photosphere at about March 8 for SN 2002ap? My eye suggests that they are only about optical depth 1 including combined Ca II line. Is that about right? Tchues David From mazzali@circe.oat.ts.astro.it Mon Apr 01 03:38:48 2002 Hi David, i'm reading you from home, so i'm not sure this will work but i'll try. i'll include some of my o/p for a synthetic spectrum on feb 8. as you can see, the Soboloev tau is >> 1. however, if the CaII IR triplet really only is the small lime to the red, then the estimate there may be much less. you still need > 1 i think because there is quite sa lot of abs. overall. does this make sense? cheers paolo tau Sob wvl obs wvl rest El El log gf 13.990 8040.9 8446.25 O I 9.52 76794.98 -.520 55.951 8034.1 8446.36 O I 9.52 76794.98 .170 37.138 8037.8 8446.76 O I 9.52 76794.98 -.050 51.236 7607.1 8498.02 CA II -1.69 -13650.19 -1.312 381.037 7637.1 8542.09 CA II -1.70 -13710.88 -.362 2.543 8193.0 8581.56 CO II 10.49 84586.96 .598 2.088 8241.7 8632.35 CO II 10.60 85481.80 .577 237.373 7748.8 8662.14 CA II -1.69 -13650.19 -.623 David J. Jeffery Department of Physics New Mexico Tech 2002 March 31 Hi Paolo: It was March 8, not Feb. 8, that I was interested in. I'm trying to determine the ISP from the data itself which is really the only way it can be done. It would help if the lines were optically thinnish then: i.e., of order optical depth 1 at the photosphere. They look about that actually. Was there a real photosphere left by March 8. Given the very high velocities and the presumed lowish mass, I would guess that the atmosphere would be semi-transparent except for lines and electron opacity like Ia's are at a comparable epoch. Any idea what the electron optical depth to the center is on March 8? Are most of the lines on March 8 aside from Ca II and O I, weakish Fe II lines? Tchues David From mazzali@circe.oat.ts.astro.it Mon Apr 01 13:46:01 2002 Hi David, the latest spectrum i tried to model is feb 21. sorry i misread feb for mar, as i said, working from home is not so easy. anyway, on feb 21 i still have strong CaII and OI lines. i suppose by mar 8 the photosphere should be almost gone, but i think CaII would still absorb quite a bit. why do you need it to be optically thin? cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 March 31 Hi Paolo: Koji and I are trying to figure out what the interstellar polarization is. The intrinsic polarization should be low at the peak of the Ca II emission just due to dilution by the unpolarized non-scattering emission from those lines. But the polarization would be even lower, if the lines depolarized electron scattered polarized flux. If the lines had optical depths much greater than 1 everywhere, then likely the polarization right at the emission feature is pure ISP. But to my eye the line optical depths look one-ish, but it's hard to tell with the jolly big emission feature maybe filling them in. At one-ish some polarized flux may escape depolarization. In fact lines can polarize at a weaker level than electrons: but the Ca II triplet in an old-fashioned approximation anyway is very weakly polarizing. Complicating things there is another way that polarized photons can evade even a strong line. Electron scattering actually has a broad frequency redistribution: it is possible for a blueward-of-the-line photon to be scattered and polarized and end up redward of the line. But it can only do this once and still have globally significant polarization (as opposed to a merely random one). This is probably a very weak effect, but it is hard to be sure. Anyway I've almost convinced myself that Koji is right to assume that polarization at the Ca II triplet emission is nearly all ISP. It's a pity the February 8 data doesn't extend to the is wavelength: if they had both agreed on the same value, that would be pretty definitive. With that ISP value, 2002ap turns out to have a rather low polarization in the continuum: only of order 0.1 and nearly consistent with zero. The line polarization is real enough. The object probably is really very asymmetric, but line depolarization may lower it's polarization compared to a hydrogen atmosphere object. But if the bipolar jet model were right then line depolarization wouldn't correlate with any particular lines ... on and on musing. Tchues David %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% From mazzali@circe.oat.ts.astro.it Sat May 18 10:43:48 2002 Subject: Re: polar. in 2002ap Hi David, Ken sent me the draft of the paper and your further comments to see if i have anything clever to say, which i doubt. i had one probably stupid thought, though. Could the redshifted polarised be due not to a jet but to something like a 56Ni blob? the wvls are those where Fe lines form the spectrum, so this is suggestive, and the velocity happens to be very similar to the highest velovity i could accommodate in the model. 56Ni should create free electrons, as you say, and so there may be polarisation. what do you think? Also, you mention montecarlo. this sounds interesting. i can provide any codes if you are interested... cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 May 18 Hi Paolo: To me the paper turned out rather interesting. I hope it's OK. You might notice that I had a heavish hand in shaping the text. However, Koji and/or the collaborators over there had all the bright ideas actually. The jet idea is highly speculative and I hope you find that the text makes that clear. The agreement of the redshifted scaled spectrum and polarized flux is suggestive, not overwhelmingly convincing to me. If you think we say too much, suggest we tone it down. But who knows. Cas A seems to have had a jet. I will sometime in some coming months try to see if I can make any more progress with the jet idea. But I've got two papers, a course, and a mandatory vacation to do. To me a fast blob of Ni-56 and a jet arn't distinguishable. So it's possible. Maybe you should add some text and join the paper. At least Ken doesn't disagree with the idea of jet. But then Ken doesn't seem to mind going out on limb. I am interested in Monte Carlo but I'm not ready to borrow advanced codes. I have to do more playing around to understand what's happening with my toy codes. Which comes to asking you for a favor. A good way to learn something new is try to write a paper on it. So I'm writing a paper on Monte Carlo technique. It's mostly derivative from Lucy's recent papers. Would you have a look at a draft when ready (today, tomorrow, months) and tell me if anything is worthy or novel? Regards David From mazzali@circe.oat.ts.astro.it Mon May 20 04:33:49 2002 Hi David, ok, i'll try to write a few lines and give it ti Ken, he can decide what to do with it. There is one thing that confuses me, though, when you discuss PA (something i know less about than ancient Sanskrit): you (?) say that in Feb the PA is <~120 for the OI/CaII line and ~80 for the continuum. Now, if i look at Fig 1c or 1a and naively read off the axes, i see nothing of the sort... Rather, i see PA (cont, ie <6500A) ~120, and ~100-120 in the lines. What am i doing wrong? cheers paolo ps. i'll be glad to read/comment whatever you write. If we decided to develop a MC polarisation code togeter, i think we could be rather quick. David J. Jeffery Department of Physics New Mexico Tech 2002 May 20 Hi Paolo: You've missed the ISP correction nonsense. The 80 and 120 degrees refers to after the correction has been made: see Figure~3d. ISP is a great pain. It can pretty much distort the polarizationa and PA spectra altogether. On a QU plot ISP just translates the QU data: well nearly just: ISP has some wavelength dependence and so there is some distortion. I think we make a good argument for the ISP correction we adopt, but we could still be wrong. I hope someone will be able to spectropolarimetry some months down the road when the electron optical depth goes to zero. The measurement then should be pure ISP and thus nail down that quantity. Thanks for willing to read my attempt at a Monte Carlo paper. I'll send you a postscript by tomorrow night: I'm trying to pull the last threads of a full draft together. On Wednesday I'm off to Canada for two weeks (back June 6), and so there's no need to hurry about reading. Anything you have to say will be welcome: probably it's all been done better before. I think we certainly could write such a polarization mc code. The problem I ran into a decade ago (you remember the good old days) with my Sobolev polarization code was that I couldn't get sufficient accuracy for real asymmetry. One needs to know the two components to better than 1 in 1000 in order to have about 10 % accuracy in a polarization calculation. And my poor old code just couldn't really do that in a trustworthy way. So that's when I left off. Now in a straight count-the-packets Monte Carlo polarization is straightforward. You just need to treat scattering a bit differently and do extra book keeping to keep track of three intensities: 4 if you keep track of circular polarization too. In the analytic averaging Monte Carlo that you and Lucy now do I need to give some thought to how the Stokes parameters must be accounted for properly. The trouble is that polarization is such a finicky thing. SN beams should be fairly highly polarized. Maybe 30 % or so tyically, but integration cancels most of this and one observes a measly 1 % or so do almost any kind of asymmetry. One thing one could do (and I planned to do this but never did) was just to calculate the polarization along impact parameter every time one calculates a spherically symmetric model. In principle this should be a trivial modification of spherical Monte Carlo code. One still doesn't know what polarization one gets from asymmetry, but one will get some insight into the polarizing properties of any model that one calculated. This would be a first step to figuring out how to do polarization calculations for real asymmetric models. Of course, Peter Hoeflich is years on in doing this kind of thing. Anyway see what you think of my understanding of Monte Carlo when you read my ``paper''. Regards David From mazzali@circe.oat.ts.astro.it Tue May 21 04:26:57 2002 Hi David, now i get you. Obviously the order of the presentation is a bit confusing. Anyway, i have prepared some text and sent it to Ken, who says he'll try to get me in the paper. We'll see. Somehow the blob sounds like a possibility to me. I'm attaching the 'new' latex file for the paper. mind you, it's all up to Kawabata et al to accept it. If you think it's ok, you can help by pushing a bit. I'm looking forward to reading your draft. Also don't expect instant reaction from me: i've loads to do in the next week or so, then i'll say bugger all and go home for a world cup holiday. no point in trying to work with football on from 8am to 4pm.... but that may be the time i can do some reading... the idea of starting with a simple polarisation calculation sounds like a good one. in the new code all the old features are preserved, so the only q. is whether using a smaller no. of packets still gives you a good result. but one may do this as an on/off thing: do spectrum synthesis first, then take the 'final' model, increase the no. of pkts, switch pol. on and just run this once. Anyway, we'll see. cheers paolo From mazzali@circe.oat.ts.astro.it Sat Aug 10 08:57:06 2002 To: "Koji S. Kawabata" cc: nomoto@astron.s.u-tokyo.ac.jp, , , Dear all, it seems the referee is strongly opposed to the jet idea, yet i cannot see why. but should we now really say that we can rule out a jet? my point about the radio stuff is what is the definition of relativistic. we only ever claimed to see stuff moving at 0.1c at most. now, that's not really relativistic to me. i think they are confused. but should we really go into that? cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 August 12 Hi Paolo: If you seen the draft reply to the editor, then you seen my argument why asymmtry does make a jet tenable. Do you agree with that argument? I'm not a radio expert, but it seems unchallengeable to me. Anyway Berger et al. I think are just hiding too many cards. They rule out your values, but not using your model. Maybe their idea is right that the Matzner and McKee results give the true analytic outside of the ejecta---but obviously only in spherical symmetry. So they seem to be arguing for asymmetrical ejecta on the grounds that a spherically symmetric model is right. Another point on which I'm unclear is this. Say that we were looking at the fast axis of SN 2002ap and that accounts for the high velocity lines. How much would changing to say a prolate model reduce the amount of relativistic ejecta that Berger et al. can't tolerate: factor of 2, 3, 10, but not 100, I'd guess. Now they don't actually say what they can or can't tolerate. They just imply outer ejecta would have to orders of magnitude lower than the values they get using your parameters with someone else's model. I don't see that invoking asymmetry in optical modeling is going to reconcile the fast lines and their radio modeling if that is the case. What a long haul to get a simple little paper published. Regards David From mazzali@circe.oat.ts.astro.it Tue Aug 13 04:18:12 2002 To: "David J. Jeffery" Subject: Re: your mail Hi David, i do agree with you. i am really upset about this. i would like to know who this idiot is. as i said, go for it! i was in something like this last year, i fought hard and i won! cheers paolo From mazzali@circe.oat.ts.astro.it Tue Aug 13 03:58:12 2002 Dear David, Koji, i agree with your remarks. Go for it! We should ask Berger et al to say just what they mean when they talk about our work... cherrs paolo David J. Jeffery Department of Physics New Mexico Tech 2002 August 13 Hi Paolo: Thanks for the encouragement. When the dust is settled Berger et al. may well be right that people won't classify 02ap as a hypernova. And they may be right that the jet does not exist: we know that it is a speculative hypothesis---but one motivated by our data. But SN 2002ap is not an ordinary Ib: it does have very fast lines which may indeed be caused by high asymmetry. Branch et al. (2002) from an admittedly small sample show high velocity Ib's whatever their cause are at least rarish. (02ap is a Ic!!! So the Branch et al. argument is no good, except insofar as Ib/c's are the same critters.) Berger et al. themselves say 02ap is a factor of 20 less radio luminous than typical Ib/c's: given it's high velocity lines, there is clearly something unusual. Just to repeat myself, I believe geometry matters a lot to radio modeling. Unless there is some demonstration to the contrary, I don't see why I should change my view. To my knowledge (admittedly small) radio observations of SNe---not remnants of course---have never been able to diagnose large-scale asymmetry in ejecta or CSM: I know that they do invoke clumpiness at times in filling-factor parameterization. Rather than radio modeling being independent of geometry, I suspect radio modelers just lose track of it in their parameterizations. Of course, that's no different from us optical people. In any case, there is no going back for the paper and the jet model. It is already published at astro-ph, and so, right or wrong, it's already on our rap sheet. If a second referee is of the same mind as the first, then I'd say throw in the towell on refereed publication, and have Koji write a new paper just reporting the observations for refereed publication. Regards David From mazzali@circe.oat.ts.astro.it Wed Aug 14 04:17:11 2002 To: "David J. Jeffery" Subject: Re: your mail Hi David, yes, we have to think about 2002ap. i still think those high velocities are unusual however, but i need to compare with more Ibc's. i'll try to access Alex Filippenko's database for that. by the way, i have been thinking about a polarization code in 3d after hearing what the LBL people are doing. i think we could develop this quickly. i have the 3D stuff, you have the polarization stuff... you should come and visit when you have a chance. cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 August 14 Hi Paolo: With your Monte Carlo code, implementing polarization shouldn't be difficult: the coding might be tedious and spoil the elegance. Your modern Monte Carlo requires building up the source functions and then formal integration for the emergent spectrum. That can work for polarization too. The way I did it in my old code---that I dare not run in polarization mode any more---is to create electron scattering lines. In your code what could be done is to create these electron scattering lines between real lines. One just treats them as pure scattering with complete redistribution in frequency: this is somewhere vague between coherent scattering and their real redistribution. Then one can treat them just like ordinary obolev lines, except that they have angular redistribution determined by the Rayleigh phase matrix. This separation of angular and frequency redistribution is most valid for 90 degree scattering which is most important for polarization (Mihalas 1978, p. 420). For full 3-d, one has to build 10 source function coefficients. The formalism for this is in Jeffery (1990, ApJ, 352, 267). This pseudoline approach is, of course, approximate. The real photon redistribution on electron scattering is rather complex (Mihalas 1978, p. 420) and not amenable to pseudoline treatment. It is also true that electron scattering can cause photons to jump across lines in frequency space to the red and the blue---although we never worry about that in SNe. In a straight Monte Carlo calculation in which you count outgoing packets to assemble the spectrum no redistribution approximations are needed for electrons. But then you will have to count a lot of packets in 3-d to build up sufficient accuracy. But this is one of those cases where my radiative transfer intuition tells me the pseudoline approach should be OK. It captures the main polarizing and angular effects. As one generally knows complex redistributions often end up washing themselves out. In fact, probably I should do a project to test pseudolines against full Monte Carlo in spherical symmetry. Or maybe I should just think about it until the analytic justification is clear. Why don't you get a grad student to work on polarization. Maybe someone at the Sternwarte if you are still on good terms with Adi and Achi. I wonder why Rolf didn't take them to Hawaii---ha, ha, just a little joke. Regards David PS: Sorry, I accidently sent you our whole recent correspondance. From mazzali@circe.oat.ts.astro.it Wed Aug 14 15:45:29 2002 To: "David J. Jeffery" Subject: Re: your mail kind of a strange email, with all the emails we exchanged about 02ap etc... how did you manage that? p From mazzali@circe.oat.ts.astro.it Wed Aug 14 15:48:43 2002 From: Paolo Mazzali To: "David J. Jeffery" hi David, yes, that was a strange email... anyway, what you say may be a possibility, and so may be using my other MC code, that does light curves. that has been set up in 3d by a student in tokyo, so we just would need the formal integral. rather than scattering lines we could just use electron scattering maybe... cheers paolo David J. Jeffery Department of Physics New Mexico Tech 2002 August 14 Hi Paolo: Now that my mind is returning to 10 years ago. I understand why pseudolines should work nearly perfectly for SNe as long as one just considers continuum radiative transfer or if one doesn't want to get really intricate with how electron partial redistribution (PRD) in frequency affects lines. But PRD must affect lines at some level. Electron PRD changes the coupling of the radiation field to lines. Consider an optically thick line (i.e., an opaque line). If electron scattering is treated as coherent, all the photons to the blue of the line must interact with it and none of the photons to the red. If PRD is turned on there can be photon jumps across the line. Say J_blue>J_red, then the net jump rate is to the red and the line sees a weaker field than with coherent redistribution since only some of the net jumpers to the red jump back to the blue: the line is less coupled to the local field; effectively it's optical depth is reduced. Whether the line emits more or less depends on detailed NLTE. But one might guess it will emit less. Say the line can be divided cleanly into scattering and NLTE parts. If the NLTE absolute emission and fractional absorption arn't affected by reduced input to the line (a big if), then net emission is reduced since the output of scattering is reduced and the NLTE absolute absorption is reduced. Overall, UV lines in SN atmospheres tend to be net absorbers (J_blue>J_red) and so with electron PRD they will overall absorb less. More flux will leak through the UV than with coherent electron scattering. Now say J_blue