Here are the runs I have run:

workingDir = 'run2011112_123'

#Angular Correlation

oversample = 5.

corrBins = 10.0 #should be one more than you want it to be

mincorr = 0.3

maxcorr = 3.0

convo = 180./pi

#Auto Correlation

boxside = 2200.0 #in Mpc/h

xicorrmin = 5.0

xicorrmax = 50.0

nxicorr = 40

#make redshift bins

#boxside = 1000.0 #in Mpc/h

rbinmin = 0.0 #in Mpc/h

rbinspace = boxside/10 #in Mpc/h

rbinmax = boxside + 1.0 #in Mpc/h

rbins = makeRbins(rbinmin,rbinmax,rbinspace)

nrbins = int(rbins.size - 1)

rbins

array([ 0., 220., 440., 660., 880., 1100., 1320., 1540., 1760., 1980., 2200.])

Reconstruction Plot

correlation functions:

workingDir = 'run2011113_051'

#Angular Correlation

oversample = 5.

corrBins = 10.0 #should be one more than you want it to be

mincorr = 0.3

maxcorr = 3.0

convo = 180./pi

#Auto Correlation

boxside = 2200.0 #in Mpc/h

xicorrmin = 5.0

xicorrmax = 50.0

nxicorr = 40

#make redshift bins

#boxside = 1000.0 #in Mpc/h

rbinmin = 100 #in Mpc/h

rbinspace = boxside/10 #in Mpc/h

rbinmax = boxside + 1.0 #in Mpc/h

rbins = makeRbins(rbinmin,rbinmax,rbinspace)

nrbins = int(rbins.size - 1)

rbins = array([ 100., 320., 540., 760., 980., 1200., 1420., 1640., 1860., 2080.])

Reconstruction Plot, same binning as above

Correlation Functions

workingDir = 'run2011113_055'

# Create file names

photofile,specfile,corr2dfile,corr3dfile,photo2dfile,photoCDfile,spec2dfile,\

spec3dfile,argumentFile,runConstantsFile,wpsMatrixFile,xiMatrixFile=makeFileNamesData(workingDir)

#Angular Correlation

oversample = 5.

corrBins = 10.0 #should be one more than you want it to be

mincorr = 0.3

maxcorr = 3.0

convo = 180./pi

#Auto Correlation

boxside = 2200.0 #in Mpc/h

xicorrmin = 5.0

xicorrmax = 50.0

nxicorr = 40

#make redshift bins

#boxside = 1000.0 #in Mpc/h

rbinmin = 100 #in Mpc/h

rbinspace = boxside/12 #in Mpc/h

rbinmax = boxside + 1.0 #in Mpc/h

rbins = makeRbins(rbinmin,rbinmax,rbinspace)

nrbins = int(rbins.size - 1)

rbins = array([ 100., 283.33333333, 466.66666667, 650. , 833.33333333, 1016.66666667, 1200., 1383.33333333, 1566.66666667, 1750., 1933.33333333, 2116.66666667])

Reconstruction Plot, same binning as above

workingDir = 'run2011113_110'

#Angular Correlation

oversample = 5.

corrBins = 10.0 #should be one more than you want it to be

mincorr = 0.3

maxcorr = 3.0

convo = 180./pi

#Auto Correlation

boxside = 2200.0 #in Mpc/h

xicorrmin = 5.0

xicorrmax = 50.0

nxicorr = 40

#make redshift bins

#boxside = 1000.0 #in Mpc/h

rbinmin = 100 #in Mpc/h

rbinspace = boxside/8 #in Mpc/h

rbinmax = boxside + 1.0 #in Mpc/h

rbins = makeRbins(rbinmin,rbinmax,rbinspace)

nrbins = int(rbins.size - 1)

rbins

array([ 100., 375., 650., 925., 1200., 1475., 1750., 2025.])

workingDir = 'run2011113_119'#Angular Correlation

oversample = 5.

corrBins = 10.0 #should be one more than you want it to be

mincorr = 0.3

maxcorr = 3.0

convo = 180./pi

#Auto Correlation

boxside = 2200.0 #in Mpc/h

xicorrmin = 5.0

xicorrmax = 50.0

nxicorr = 40

#make redshift bins

#boxside = 1000.0 #in Mpc/h

rbinmin = 400 #in Mpc/h

rbinspace = boxside/10 #in Mpc/h

rbinmax = boxside + 1.0 #in Mpc/h

rbins = makeRbins(rbinmin,rbinmax,rbinspace)

nrbins = int(rbins.size - 1)

rbins

array([ 400., 620., 840., 1060., 1280., 1500., 1720., 1940., 2160.])

Reconstruction

Hi Jess,

ReplyDeleteThis isn't quite exactly what I meant for you to do, but I think we can learn a few things from what you did. So you have definitely confirmed that the reconstruction is sensitive to the binning that you pick for measuring the correlation function (in addition to being sensitive to the binning that you pick for the reconstructed phi(chi)). This does not surprise me, because I have always believed that the measurement of the 3D corrfn is noisy, and therefore, when we spline/interpolate it, we are getting "features" that are actually noise.

The information for the plot I intended for you to make is possibly all here, but if not, it should be easy for you to adapt what you have to make it. I am actually interested in comparing the 3D (and the 2D) correlations functions that you measure, for slight variations of the SAME BIN. So what I would like you to do is pick bin somewhere in the middle distance (not the farthest or the nearest). Measure xi and wps. Now change the bin slightly: 1) test a variety of different widths (say between 15% larger to 15% smaller). 2) test a variety of bin centers (say bin centers between 5% closer and 5% farther).

Also, one more thing that was puzzling me: When I was doing this, I usually found that the 3D corrfns were way noisier than the 2D ones, and this was because the 2D ones had waaaay more objects, because the photometric sample was so much bigger. I notice you are seeing the opposite trend, and this really worries me. What are the sizes of the data samples that you are using? Do you have any idea why the 2D functions are looking so bad? I would really not expect to be able to do any reconstruction at all with correlation functions that poor, I'm sorta surprised that it EVER works.

P.S. I notice there were no plots for xi and wps for the two cases where there was significant bouncing in the reconsruction. Can you add those plots? I'd like to see it there are any qualitative differences between cases that "work" and those that clearly fail.

ReplyDelete