Tuesday. No sign of computer arriving yet.
1105.3730
Pangenesis in a baryon-symmetric universe: dark and visible matter via the Affleck-Dine mechanism
Bell, Petraki, Shoemaker, Volkas
* Here we go again... what is the Affleck-Dine mechanism?
* A-D mechanism can do both baryogenesis and DM creation, hence opens the possibility of explaining the ration between Omega_m and Omega_b. In supersymmetric theories, the early universe carried baryon and lepton-number-carrying scalar fields. This interacts with the inflaton field, and CP-violating and B-violating effects can then be introduced. The scalar particles decay to fermions, the net baryon number the scalars carry can be converted to ordinary baryon excess.
If DM and ordinary matter originate via the same mechanism, then the baryonic asymmetry may be compensated by an asymmetry in the dark sector. Describe how separation of baryonic and antibaryonic number can originate in the vacuum via the Affleck-Dine mechanism, due to spontaneous symmetry breaking and a 2nd order phase transition in the post-inflationary era. Individual stability of the two sectors are guaranteed by symmetry restoration in the current epoch.
1106.0299
Dark energy and nerutrion masses from future measurements of the expansion history and growth of structure
Joudaki, Kaplinghat
* another Fisher matrix forcast?
Forecast [yup] the cosmological constraints from combining WL tomography, galaxy tomography, SNe, CMB, incorporating all cross-correlations between observables. Non-zero curvature, high-z DE. Early DE can be constrained to 0.2% of the critical universe with Planck+LSST; curvature constrained to 0.06%. But this extra degree of freedom degrade the ability to measure late-time DE and the sum of neutrino masses--roughly a factro of two degredtation in the constraints overall, compared to the case w/o early DE. Constraints similar for spaced-based missions. Bias can be up to 2 sigma with early DE. Throwing out nonlinear scales (l>1000) may not result in significant degradation in future parameter measurements, if multiple cosmological probes are combined. Including cross-correlations between the different probes result in better constraints (x2) for neutrino masses and early DE density.
1102.3237
Improved constraints on cosmological parameters from SNIa data
March, Trotta, Berkes, Starkman, Vaudrevange
Test new method based on Bayesian hierarchical model to extract constraints with the SALT2 lightcurve fitter. Better by a factor of 2-3 in statistical bias compared to the usual chi-squared approach. Includes full statistical covariance matrix.
* this is another one on SNe.
1104.2596
Optimizing weak lensing mass esimates for cluster profile uncertainty
Gruen, Bernstein, Lam, Seitz
* Very relevant for our cluster project. We need to know the cluster mass well, but for WL masses, systematics can interfere.
Measured cluster shear signal varies at fixed mass due to shape noise, intervening LSS, and cluster structure variation due to asphericity, substructure and scatter in concentrations.
Use N-body sims to derive and evaluate WL circular aperture mass with minimum variance in the estimated mass in the presence of these variabilities. The new estimator does better than the M_ap filter optimized for circular NFW profile without intervening LSS. Accounting for the variation of internal cluster structure appears to be more important than accounting for intervening LSS.
* I'm actually now going to read this paper, to figure out what the optimum estimator for a single cluster mass estimate (I'm assuming the above is not for stacked, or averaged, cluster mass).
2005MNRAS.361.1287M
Systematic errors in weak lensing: application to SDSS galaxy-galaxy weak lensing
Mandelbum, Hirata, Seljak, Guzik, Padmanabhan, Blake, Blanton, Lupton, Brinkmann
Use ratio test (using source subsamples of different mean redshift) to calibrate errors in Delta Sigma measurements. Uncertainty in Delta Sigma is 10% (with or without photoz estimates). Also check other systematics. No evidence of inconsistency among different subsets of data.
* What are the "other systematics" that they've checked?
* It's a really long paper (36 pages).
====
1. Intro
- gg lensing measure at high S/N, revisit common sources of systematic error
- many sources of calibration biases, either through tangential shear or Sigma_crit
- notations and behavior of Sigma_crit
- errors in shear computation discussed in Hirata+ 2004 (psf, noise, selection bias)
- assume zlens known; calibration bias due to source redshift distribution
- systematics in computing the signal: intrinsic alignments, selection effects, sky noise, etc.
- such systematics common to both gg lensing and cosmic shear autocorrelation
- Delta Sigma is invariant; use it to test cosmology, or (down to 1%) systematics.
- Sections: data, z distributions, additional systematics, implementation, results, discussion
- cosmological models used
2. Technical apparatus
- use SDSS data
2.1 Lens catalogue
- SDSS main VAGC, north only, ~260k galaxies, z>0.02, -23 > ^{0.1}M_r > -17, 1 mag bin
- flux limited to r~17.77, redshift evolution corrected
- z distribution of six luminosity bins, luminosity relative to L* (M*=-20.44), z_eff
- flux limited sample, use weight of Sigma_crit^-2, effective values include these weights
2.2 Source catalogue
2.2.1 Construction
- SDSS photometric catalogue, Photo rerun 137, model magnitudes (higher S/N)
- different from Hirata+04 set, more galaxies, shape measurements, selection, cuts, organiz'n
- selection: star/galaxy separation, cut on resolution factor, deblended gals, r<22 and i<21.6, r & i detection, discard problematic shape measurements or images
- shape measurement: re-Gaussianization, avoids shear calibration problems
- Re-Gaussianization: perturbative PSF correction scheme: ellipse as Gaussian of covariance
- correct for non-Gaussian shape (find best-fit Gaussian)
- covariance matrix M = 'adaptive' covariance matrix, T = 'adaptive' trace, obtain ellipticity
- non-Gaussianity of the galaxy (assuming circular PSF), its correction, resolution factor R_2, generalize to elliptical PSF
- non-Gaussianity of the PSF: find best fit, find residual, approximate pre-seeing galaxy as Gaussian, construct re-Gaussianized image (what would have been observed had the PSF been Gaussian).
- re-Gaussianization scheme: exact to first order in PSF non-G; higher order methods proposed
- cuts: shapes avg of r and i; resolution R_2 > 1/3 in both bands (to avoid selection effects)
- shape error: for weighting, determining shear responsivity, determining error bars on final qtty. determined from sky and dark current noise and total flux (S/N divided by resolution factor); noise from galaxy Poisson noise itself (~1/R_2 times sqrt(1/N)), sky dominates for r>20
- removal of regions with faulty data: runs with high e_rms, <e> > 0.05, e^2 > 4, faulty astrom.
- unique observation: pick the one with better seeing. combine r and i shapes w = (S/N)^2
- photoz for r<21 galaxies w/ Kphotoz v3_2 (Blanton+2003), 4% photoz failure
- magnitude distribution of sources; RMS ellipticity in mags; resolution R_2 in mags shown.
2.2.2 Shear calibration bias
- describe shear calibration bias, estimate its magnitude
- five major sources: (1) PSF dilution--small at <5%
. analytical model from the calibration plot (conservative)
. upper bound (most gals at R_2>1/3); R_2 and e^2 uncorrelated after sigma_e^2 removed
- (2) PSF reconstruction (misestimation of PSF ellipticity or trace by PHOTO). <0.03 error.
- (3a) Shear selection bias: resolution cut favors elongated galaxies, analytical model
. estimate R_2>1/3 cutoff on mean ellipticity from the analytic function above. ~0.05-0.11 err.
- (3b) Shear selection bias: S/N>5 detection requirement, opposite as above. ~-(0.03-0.06) err.
- (4) Shear responsivity error: systematic uncertainty in e_rms (shape noise). ~0.01
- (5) Noise rectification bias: analytic estimate: -0.005 to -0.01
- other minor sources, at 0.1% level: camera shear, pixelization, atmospheric refraction. Total error (conservative): total -8 to +20 per cent
2.3 Shear estimator
- weight = 1/(shape noise^2 + sigma_e^2) * (1/Sigma_crit^2), calculate responsivity
- shear estimator equation
- ellipticity-dependent weight had minimal effect on error bars
2.4 Error determination
- determining errors in Delta Sigma
2.4.1 Analytic computation
- derived from BJ02; incorrect in the presence of spurious shear power; no easy way for errors in boost factor; does not account for correlation of radial bins
2.4.2 Random catalogues
- in absense of systematic shear, avg signal around random points should be zero; only valid on small distance scales; cannot takin into account the errors on the boost factors; can take correlation of radial bins; requires large number of random catalogs (e.g., 24 for SDSS)
2.4.3 Bootstrap resampling
- divide into 200 subregions, compute signal for each subregion with replacement, run 2500 to determine avg signal and its error. Naturally incorporates errors in boost factor and correlation of radial bins; errors not valid at scale larger than subregion size; noise in covariance matrix = chisq values do not follow chisq distribution.
3. Redshift distributions
- methods for redshift distribution determination and errors; describe reference distribution
3.1 Photometric redshifts
-
3.2 COMBO-17 distribution
3.3 DEEP2
3.4 Reference distribution: LRGs
4. Other systematic issues
4.1 Random points test
4.2 45 deg test
4.3 star-galaxy separation
4.4 Seeing dependence of calibration
4.5 R_2 dependence of calibration
4.6 Systematic differences between bands
4.7 Boosts
4.1.7 Statistical errors
4.7.2 Systematic error: non-uniformity of boost factor
4.7.3 Systematic error: magnification bias
4.8 Intrinsic alignments
4.9 Correction for non-volume limited lens sample
5. Application to the systematic test
5.1 Methodology
5.2 Cosmology dependence
6. Results
6.1 Redshift distributions and photometric redshift performance
6.1.1 Redshift distributions
6.1.2 photometric redshift errors
6.2 Error computation
6.3 General systematic tests
6.3.1 Random points test
6.3.2 45 deg test
6.3.3 star-galaxy separation
6.3.4 Seeing dependence of calibration
6.3.5 R_2 dependence of calibration
6.3.6 Systematic differences between bands
6.3.7 Boosts
6.3.8 Intrinsic alignments
6.3.9 Corrections for non-volume limited sample
6.4 Redshift systematic tests
6.4.1 Bright (r<21) sources
6.4.2 Faint (r>=21) sources
6.4.3 LRG sources
6.4.4 All
7. Conclusions
====
Rachel's e-mail:
Testing shear measurement errors
* e_rms was increasing at fainter magnitudes, based on <e^2 - sigma_e^2>, in 2005.
[can we be detecting more elongated objects? Are fainter objects more likely to be highly elliptical? not sure if this is making sense] --Rachel says we will have to trust the sigma_e values to make sure that we have the right e_rms.
* Ran test by adding SDSS-like noise to COSMOS, and check what the real sigma_e is like. These test indicated that the quoted sigma_e are too low, but also showed dependence on galaxy properties. The test is also computationally expensive.
* Use all galaxies with shapes used in the source catalog, and construct histograms of
(e1_r - e1_i) / sqrt(sigma_e,r^2 + sigma_e,i^2)
and likewise for e2. The histogram should be a Gaussian with mean 0 and sigma = 1. But:
1) the results are not quite Gaussian--it has a large tail
2) using 68 percentile as 2 sigma width (instead of std deviation), found
sigma_e,true ~ 1.45 * sigma_e
Assumes sigma_e in both bands are underestimated by the same factor.
The underestimation probably stems from a failure of the approximations behind the formula used to estimate sigma_e.
* Now estimate a new sigma_e for the source sample:
a) split into 20 mag bins (equal sized), measured sigma_e,true / sigma_e in each one.
Fraction = 3 at bright end, 1.2 at the faint end.
b) split into 20 resolution bins (equal sized). Already correcting the bias, the correction factor is not that significant, with >1 at low resolution, to <1 at high resolution.
c) split into 20 ellipticity bins; remove the trends.
No comments:
Post a Comment