hi Bartek, Agnieszka, cosmo-torun,
On Sat, 29 Jul 2006, Bartosz Lew wrote:
Hi Boud, Agnieszka,
Well, sorry to hear about the grant results.
thanks.
my answers are:
* referee 1: score 4 * referee 1: report - not possible to reject the hypothesis, so not worth fundingright, maybe polish science is more oriented on applied science and projects that must give positive results. Our project from the beginnig said "maybe". Of course we couldn't have said anything beyond that. While the referee seem not to get too deeply into details of the projetc itself - not mentioning the field related - since he doesn't understand that the simulations for proper statistical significance analysis are computationally consuming, apart from the search of the circles, and apart from the fact that the planck data details are already known but - he makes one good point - that we do not mention that the results will be e.g. PhD. Thesis and me and Agnieszka are Ph.D. students, so from his point of view it might be suspicious that maybe we will not be as involved as we should. I give him that.
Hmmm one of you is on one already published topology paper and the other is on a submitted paper on cosmic topology (theory) which is now just waiting for the referee to check out the revised version. Expecting us to write "this will be a part of X's PhD thesis" seems to me pretty silly. It's not your main topic (for either), but that's no reason why you cannot substantially contribute or be part of the group.
The fact that he's not interested in topology is also obvious.
* referee 2: score 7 * referee 2: report - the project will not give a final answer to the topology of the Universewell, this report is more moderated and ballanced in words but to me it's just general mambo-jubmo just to say NO, however the score 7 is considered to be "very good".
Overall we got ranked 23rd and only the first 10 got accepted. So even a mean score of 7 is probably not high enough...
* referee 3: score 6.5 * referee 3: report - researching the PDS model no longer has any sense because of the [unrefereed] article AstroPh:0604616well, I didn't have time to fully understand the seriousness of the m-weightning "mistake" - I just looked quickly through the paper, but my impression is that the "proper" weightning only recudes the signal.
The m-weighting increases contributions from high-l signals. It's certainly a good idea for a search strategy - with perfect data - but it's not in itself an argument that our detection is insignificant.
In particular, Aurich et al. astro-ph/0510847 argue that the ILC (and also tegmark et al.) map have not much power at high-l. As i understand it, this is not a problem for C_l spectra we normally see because they make statistical corrections. On the other hand, it's unclear if the high-l part of these maps is the right thing to use for circles testing.
Of course "badanie modelu PDF nie bardzo ma sens" is more less consistent with the conclusions from the paper of Shapiro et al. (however they rather
Sure, but any serious astronomer knows that the conclusions of any single published, refereed paper are not necessarily to be taken that seriously - it's the hard stuff in the middle - data and analysis - which is most important. People often publish conclusions because they need to get funding and to get past referees - try looking at all the refereed papers proving that Omega_matter = 1 \pm 0.2 or so from the 1980's and early 1990's. Yes, Omega_matter, not Omega_total.
say that he detection might be not possible) but the most annoying thing is that the authors of the paper
a) make remakrs on "*infamous* PDS model" and I take it that it straigt referes to our (Boud's) work.
This is an emotional appeal to the reader. Not terribly scientific. (But i think it also refers to Luminet et al, Aurich et al, Gundermann).
b) they say "detection" and "claim" which is obviously NOT what was written in our paper - it was merly a "hint" and "naturally it's too soon to blablalba.... etc." in the conclusions. So to me it's a kind of attack.
Well, at least it's a lack of precision.
The fact that the detection as well as our whole paper lack ANY statistical significance estimation of the detection is obvious - and it's true and it's bad. It was not a big deal to do some estimation of how strong that at 10deg. feature is , just from a bunch of gaussian simulations. And I agree with referees and shapiro et al. that this is an obvious drawback. Why didn't we do that anyway ? :)
Of course this has to be done, it's in progress :). As for why we didn't do it, there was a certain distraction from people confused about what copyright means and academic freedom etc. And i wanted to get some nice theoretical papers done. Anyway, i totally agree it has to be done. But very carefully.
The other very important thing to me about the project is what Shapiro et al. wrote: it's that the circles at 10 deg. are small, and with smoothed map to 1 or 2 deg. it's natural to me that goin' on in this process - i.e. smaller circles, bigger smoothing scale - the probability of getting some arbitrary more less flat patterns correlated in some orientation rises. Of course this is just a qualitative argument - not quantitative but it's very important point which also should be studied - and can be studied by trivial MC sims.
i agree that it's qualitative. It also seems to me that it's essentially the same argument as the lack of m-weighting - the question is how much relative importance or low-l's versus high-l's?
On the other hand, i have a feeling that what Key et al. have done is to find part of the signal - from low-l stuff (such as "flat patterns") - and remove it. Then they find that there's no signal left. IMHO this is unsurprising: if you look for the signal and then remove it, then you should expect that there's none left. The real question is what is the physical meaning of the signal either measured "in place" or by the way in which it is removed.
Well :) somehow this reminds me our conversation on the needed angular resolutions to resolve the circles at SLS. Higher resolution, less smoothing, less chance for fluke.
Certainly there should be something doable at high resolution. This has to be looked at.
Summary: unless there is a Nobel prize for discovering the topology of the Universe, research into the subject will not be funded. The astronomy community does not see fundamental research as a high priority.
yee, I see you're upset. m. :/ I don't know maybe it takes a bunch more papers with different analysis' to prove and convince referees to the projet.
:) Could be... i agree that we do need at least one followup paper.
Anyway, I have a question. Does the PDS cut the low l's for all modes similarly or only those aligned with fundamental domain (FD). I guess rather the second one. In that case (in fact in any case) I don't understand how PDS can remove power from roughly all modes but m=0 as we observrve in WMAP at l=2 and l=3 ?
i think you're taking about spherical harmonic lm modes, not PDS modes, of course.
To answer this question, you would need to compare PDS (3D curved space) eigenmodes to spherical (2D) eigenmodes and make hypotheses about the power spectrum statistics, which by definition has the problem of cosmic variance. You're welcome to try it - i can't do it in my head.
Maybe we should get jobs in a Patent Office? ;) Apparently not much has changed since 100 years ago...
well, I get my paranoia about Ph.D. - so much work - so little time - but McDonalds has so much work offers :)
;)
Anyway, it's true we need to publish our followup paper to RLCMB04. Which is in progress...
in that case don't you think you should give away some details, hm ?
Well, i'm not sure whether to be paranoid and send stuff offlist or to do it publicly on cosmo-torun... ;) After the unwillingness of A&A and MNRAS to consider the precedence evidence regarding you-know-what, it seems that mailing list precedence is unlikely to be taken seriously.
I've calculated the effect of corrections from flat geometry of PDS to spherical. - well almost - I got the formula but I don't fully understand it. Seems to be really negligible.
Good :)
It would be particularly useful to check Aurich et al. astro-ph/0510847 e.g. Table 2.
For Om_m = 0.28:
They get 0 circles for Om_tot = 1.010, while we (Fig 1 RLCMB) get 0 circles for Om_tot = 1.009 (higher precision: 1.0088?)
Hmmm, maybe the difference isn't that big...
IMHO our last paragraph RLCMB (just before Acknowledgments) is incorrect and should say 1.009\pm0.001 rather than 1.010 \pm 0.001 - at least given our plot Fig. 1.
i plan to do this correctly (with spherical geometry not flat) some time, but if you have checked this too that would be nice (less chance of us both making the same error :).
If the minimum really is closer to 1.0088, then this might be why Aurich et al. missed our signal - or it could be that using intervals in Omega_total is a bad idea if the idea is to check small circles.
pozdr. b.
ps. I have lots of ideas to consider, work out and publish. But since I have so much things do to I might need some help. I'm thinking of some company to share ideas, work, and analyze results. I'm thinking who should I potentially interest with this as well. Onyone interested ? :) Generally it's about the well,... sort of degeneracy between non-Gaussainity and non-randomness of processes. :/ i haven't worked it out yet but this thing seems to me *really* important and people seem not to see this problem. Perhaps it depends on how we put it but this is more for conference talk rather than email- email are too slow.
Well, i won't be at your conference talk, but i'm happy to make suggestions by email... :)
pozdr boud