John You are finding B2 sims with dp1101 yield average 1-sigma on r of +0.057/-0.047. Yes. Strictly speaking, that is the 1-sigma constraint from the average of the sim bandpowers. The average of the 1-sigma errors can be slightly different. From my plots, the difference for the 95% upper limit is negligible, and I have no reason to expect the 68% interval to behave differently. Can you easily add to your posting equivalent numbers for dp0000, dp0100, dp1100? That would nicely summarize the current price we pay for deprojection, and should also offer comparison to Chris' projections of r sensitivity from last fall. Yes, they are in the posting now. Which posting was Chris's projections in? http://bmode.caltech.edu/~spuder/analysis_logbook/analysis/20121104_sim1234… If I understand correctly, Chris is not calculating a likelihood. He is calculating the RMS of rho for r=0. For dp1100 he gets sigma(rho) = 0.049. I get +0.050/-0.042 for the same deprojection; however, my results are for 3 years, and his are for 2 (there are other more subtle differences in the simsets, but I don't expect any of those to matter as much). Immanuel ----- Original Message ----- Date: Mon, 23 Sep 2013 18:30:06 -0400 From: John Kovac &lt;jmkovac(a)cfa.harvard.edu&gt; &lt;jmkovac(a)cfa.harvard.edu&gt; To: Immanuel Buder &lt;ibuder(a)gmail.com&gt; &lt;ibuder(a)gmail.com&gt; Cc: bicep2-list <bicep2-list@lists.fas.harvard.edu><bicep2-list@lists.fas.harvard.edu>, Keck Array &lt;keckarray(a)mailman.stanford.edu&gt; &lt;keckarray(a)mailman.stanford.edu&gt; Subject: Re: [Bicep2-list] New Analysis Posting: Hamimeche-Lewis Likelihood for BICEP2 and Keck Great posting, Immanuel. Interesting that both BICEP2 and Keck appear to have one or more examples among the 99 r=0 s+n sims for which the ML estimate of r is ~0.3. Presumably we have already seen these outlier realizations among the gray lines in the reduc final pagers... You are finding B2 sims with dp1101 yield average 1-sigma on r of +0.057/-0.047. Can you easily add to your posting equivalent numbers for dp0000, dp0100, dp1100? That would nicely summarize the current price we pay for deprojection, and should also offer comparison to Chris' projections of r sensitivity from last fall. J On 9/23/13 3:50 PM, Immanuel Buder wrote: I put up a posting on implementing the Hamimeche-Lewis likelihood approximation for BICEP2 and Keck: http://bmode.caltech.edu/~spuder/analysis_logbook/analysis/20130923_hl_like… I show some likelihood curves and central value/upper limit constraints on r from sims. The median upper limit from BICEP2 (auto) sims with relgain+diffpoint+ellipticity deprojection is r < 0.13 (but these sims assume r=0). I propose we discuss it briefly on the telecon. Immanuel Buder Postdoctoral Fellow Harvard-Smithsonian Center for Astrophysics (office) 160 Concord Ave., M-114C (office) 617 495 7567 (mail) 60 Garden St. MS 42 Cambridge, MA 02138 ibuder(a)cfa.harvard.edu _______________________________________________ Bicep2-list mailing list Bicep2-list(a)lists.fas.harvard.edu https://lists.fas.harvard.edu/mailman/listinfo/bicep2-list

Hi, If I understand correctly, Chris is not calculating a likelihood. He is Yes, and I think you've listed all the appropriate caveats in drawing a I actually sent an email last night but forgot to reply all. That is the correct posting, and yes, the fact that my 1-sigma r numbers are for B2 2-yr and Immanuel's are for 3-yr will hamper direct comparison, as will the fact that I am not doing likelihoods but am instead reporting the rms of rho from each r=0 simset. -Chris On Tue, Sep 24, 2013 at 9:39 AM, John Kovac &lt;jmkovac(a)cfa.harvard.edu&gt; wrote: > On 9/24/13 10:15 AM, Immanuel Buder wrote: > >> Can you easily add to your posting equivalent numbers for dp0000, >>> dp0100, dp1100? That would nicely summarize the current price we pay >>> for deprojection, and should also offer comparison to Chris' >>> projections of r sensitivity from last fall. >>> >> Yes, they are in the posting now. >> > > Of course I had forgotten that you have no signal sims without difpoint to > use for the baseline dp0000, which I'd asked for wanting a "no E->B from > deproj" case, i.e. what if we didn't need to deproject anything. > > Your entry for dp0000 includes difpoint T->P and so we can ignore it. > > > Which posting was Chris's projections in? >> http://bmode.caltech.edu/~**spuder/analysis_logbook/** >> analysis/20121104_sim1234/<http://bmode.caltech.edu/~spuder/analysis_log… >> <http://bmode.caltech.edu/%**7Espuder/analysis_logbook/** >> analysis/20121104_sim1234/<http://bmode.caltech.edu/%7Espuder/analysis_l… >> >? >> >> If I understand correctly, Chris is not calculating a likelihood. He is >> calculating the RMS of rho for r=0. For dp1100 he gets sigma(rho) = >> 0.049. I get +0.050/-0.042 for the same deprojection; however, my >> results are for 3 years, and his are for 2 (there are other more subtle >> differences in the simsets, but I don't expect any of those to matter as >> much). >> > > Yes, and I think you've listed all the appropriate caveats in drawing a > > John > > > > -- > ______________________________**______________________________**_______ > John Kovac jmkovac(a)cfa.harvard.edu > > Associate Professor, Harvard University Astronomy Department > 160 Concord Ave rm 310, Cambridge MA 02138, 617-496-0611 > ______________________________**_________________ > KeckArray mailing list > KeckArray(a)lists.stanford.edu > https://mailman.stanford.edu/**mailman/listinfo/keckarray<https://mailma… > -- ********************************************************************** Christopher Sheehy KICP Fellow, University of Chicago 5640 S Ellis Ave LASR 122 Chicago, IL 60637 email: csheehy(a)uchicago.edu phone: (773) 702-9751 **********************************************************************