charlottenosam / pangloss Goto Github PK

In order to match fields, we need a mlim, see #13

BUT BoRG has WFC3 filters (F125W) and my MS catalog has SDSS ugriz-band

Need to marginalise over M* and velocity dispersion as there is error in relationships in order to find the einstein radius

Tests:

• Compare the mean weak magnification of each field with dropouts and their brightness
  • Doesn't look like there's any correlation...
    
• Plot distribution of velocity dispersion for the whole sample - is it skewed to low-sigma?
  ???????????????????
  
  • This includes objects with no measured mag: mag=99 excluded
  • Only a problem for z<1: low z and low mag
    • are the magnitudes correct?
    • Is the fit off?
  • I was using log(z) not log(1+z)....
  • Now it looks like this:
    
    
• Label strong lensing with details about the dropout

• Apply any selection cuts to the catalogue (e.g. a magnitude cut)
• Bin the universe into redshift slices.
• Calculate all the mass in each slice (neglecting any mass in halos that failed selection cuts). Truncation radius ~ 3R_vir?

Need to talk to Kasper about this.

Marginalise over mu?

• Calculate kappa_slice according to (TotalMass_slice/area_slice)/sigmacrit_slice
• Add kappa_slices according to the keeton prescription, to give kappa_smooth.
• Subtract kappa_smooth from the total kappa_keeton of each line of sight.

Rob:

• Your value of tau seems quite low. Does it change if you include scatter in your velocity dispersion function? Because the Einstein Radius goes as the square of the velocity dispersion, and the area of the sky covered goes as the square of the Einstein Radius, so tau goes as sigma^4. So the area covered by the sky will not average out when taking objects at their mean, because you gain more area from high-sigma outliers than you lose from low-sigma outliers.
  • convolve inferred velocity dispersion function with a gaussian in sigma with width 20% or so (or whatever is the intrinsic scatter of your inferred Faber Jackson relation)
• You get a magnified fraction of 1% for Mstar-Mlim=-1 in Figure 4, but your optical depth for z8 sources is 0.1%, so do you get a bias of ~10 at this flux limit? The bias should approach ~3 for flux limits below Mstar. My eyeball approximations from the plots might be off, but it would be helpful to plot the bias on this plot as well (with a shared x axis) to convince the reader that you’re not overestimating the lensed fraction.
• Dropout 1233 looks like it’s only about 2_ER from the lensing galaxy. But you quote it to be about 3_ER in Table 2, which doesn’t seem to match the scale. Are the numbers in Table 2 correct? Otherwise, it could be a possible strong lens! Also, 341 is within one sigma of strong lensing and 160 looks suspiciously elongated, so they’re all intriguing strong lensing candidates. I note in the comments that using the quoted errors, there’s a 1/3 chance that one of these has mu>2, and I suspect the magnification of 1233 might be higher, so it could be interesting.
  • Redo the postage stamps - I used ER = 1.8'' for some reason...

• Need calculate ER
  • sigma^4 ~ L (Faber-Jackson relation) -- need to account for z evolution of luminosity (distance galaxies are brighter for fixed sigma)
    • bin by z
    • Treu+ 05:
      • z<0.5: F606W
      • 0.5<z<1: F850LP ~ Y-band
    • Belli+ have higher z:
      • z>1: F160W
    • fit planes...
  • angular diameter distances
  • uncertainty in L
  • uncertainty in z (from photo-z pdfs)
• Identify lensing galaxies near sources
  • red? -- can't do this as we only have IR filters
  • z <3
  • massive?
• fit gaussians to pdfs
• When sources are within 5ER_ul of these galaxies, that field needs strong lens pdf

• Get nice results for p(mu) delta function
• p(mu) is 1 gaussian
• p(mu) is 2 gaussians
• Run with lookup tables
  • delta fn
  • gaussians

• Journal abbreviations

• SMF evolution
  • same naming convention as paper
  • z scale easier to read?
• optical depth
  • normalisation
  • z<2 - blows up, pointless
• Mult probs versus M-Mlim with BoRG
• faber jackson
  • bin with redshift
  • only 200 km/s
  • make subplots neater
• postage stamps
• Overdensity
• PDFs for overdensities - neater, remove title
• Evol LF with z
  • Joey's LF
  • Schechter LF
• Observable - fix normalisation

Kappa not so crazy, why is Mu so broad? Shear must be broad?

• I know it's not going to be important, but it looks shoddy if I don't include the simulation to z>3.5 if sources are at z~8
• I need to account for cosmic variance as best as possible.
• Best to be able to compare many lines of sight that are similar to a BoRG field, not just one.

Apparently Michele is the boss of getting data off the MS http://www.mpa-garching.mpg.de/millennium/

Need to find the distribution of kappa with z:

I need to know the total mass of halos as a function of redshifts - has anyone done this before? Or should I just use the MS and do it numerically?

Wyithe formalism: Omega_m = 0.3, Omega_lambda = 0.7, H0 = 70.2
MS/Pangloss: Omega_m = 0.25, Omega_lambda = 0.75, H0 = 73

How much difference will this make the optical depth?

Am I just being thick or this a thing?

What's a strong lens? Need to define this - mass limits?
So pangloss pdf won't include high mu (>2)

Need them for z>1 #19

Trying to replicate Fig 2 in but why do my overdense los have > 0?

• Calculate overdensity in each field
• Make lightcones with the area of the field
• Find pmu in all lines of sight of simulation using pangloss
• Match lightcones by overdensity

This should be easy once I've accounted for the smooth component...

Following #8, PDF shouldn't be gaussian - what should we use?

By Monday Sept 8th

We have the area of the sky in radians^2, but how do we find the area of each redshift slice from that?

To find the proper distance we multiply r_rad * Da(z), so do we multiply by Da(z)^2?

• Calculate the mass needed to give Omega_M = 0.25 in a given area of sky
• Calculate halo masses by truncating BMO09 NFW mass profile at 5R_vir

• Field area
• Number of objects
• Overdensity
• Limiting magnitude in a given band (one we can convert to SDSS or we'll need to get new catalogs...)