Plots for SFR-Stellar Mass-Size Paper

This is an outline of the results that will go into the first paper. The first paper will focus on the $$SFR-M_{*}-Size$$ relation, where $$Size \equiv R_e(24)/R_e(r).$$



In [309]:

    
import numpy as np
from pylab import *
%matplotlib inline
import warnings
warnings.filterwarnings('ignore')

As of 1/6/16, need to make one more pass through the sample and remove galaxies that are blended with nearby companion. Not sure if people think the numbers in each panel are useful.

Galaxies that are blended with a nearby companion are:

99056 - Hercules
103927 - Coma
103933 (AGN) - Coma
140160 - A1367
143485 - MKW11
146607 - Hercules

ALSO running this from ipython and using paperplots().

Converting SDSS Magnitude Limit to Stellar Mass Cut



In [377]:

    
# Now moving on to the SFR-M*-Size analysis

%run ~/Dropbox/pythonCode/LCSanalyzeblue.py









    



WARNING: problem matching to moustakas MSTAR_50 - probably ok
WARNING: problem matching to moustakas MSTAR_50 - probably ok
did not find  99840
did not find  146607



In [357]:

    
# using John Moustakas's stellar mass estimates
figure()
plot(s.s.ABSMAG[:,4][s.blueflag2],s.logstellarmass[s.blueflag2],'bo')
xlabel('$M_r$')
ylabel('$ log_{10}(M_*) $')
# r-band limit
rlim=17.7
# distance modulus to Hercules, the furthest cluster
mM=35.97
# absolute mag limit corresponding to r=17.7
Mr=rlim-mM
axvline(x=Mr,ls='--',color='r')
axis([-20,-16,8.5,10.5])
axhline(y=minmass)









    Out[357]:





<matplotlib.lines.Line2D at 0x122e7b510>

RESULT: SDSS mag limit corresponds to a stellar mass cut of approximately $log_{10}(M_*) > 9.5$.

Select Blue Galaxies using NUV-r Color

I have limited the sample to blue galaxies only, using a NUV-r color cut:

self.NUVr=self.s.ABSMAG[:,1] - self.s.ABSMAG[:,4]
self.blueflag2=self.NUVr < 4.1

Some galaxies don't have GALEX data (JM is checking into why this is the case). For these, I require u-r < 1.8.

We have a total of 138 blue star-forming galaxies with successful GALFIT fits.

70 cluster members
51 near-field galaxies
17 field galaxies



In [373]:

    
s.plotsalimcolormag()

SFR-Mass-Size for Blue Galaxies only



In [384]:

    
# For blue galaxies only
s.plotSFRStellarmassSizeBlue(blueflag=True,plotbadfits=False)



In [375]:

    
# to compare size distributions
print 'comparing size ratios for field vs cluster'
f1=s.bluesampleflag & ~s.membflag & ~s.agnflag
m1=s.bluesampleflag & s.membflag & ~s.agnflag
t=ks(s.s.SIZE_RATIO[f1],s.s.SIZE_RATIO[m1])
print 'mean of field = %5.2f +/- %5.2f'%(mean(s.s.SIZE_RATIO[f1]),std(s.s.SIZE_RATIO[f1])/sqrt(1.*sum(f1)))
print 'mean of clust = %5.2f +/- %5.2f'%(mean(s.s.SIZE_RATIO[m1]),std(s.s.SIZE_RATIO[m1])/sqrt(1.*sum(m1)))









    



comparing size ratios for field vs cluster
KS Test:
D =   0.30
p-vale = 0.00290 (prob that samples are from same distribution)
Anderson-Darling test Test:
D =   4.54
p-vale = 0.00521 (prob that samples are from same distribution)
mean of field =  0.67 +/-  0.03
mean of clust =  0.56 +/-  0.03



In [378]:

    
s.printsizeblue()









    



$cluster$: 
 	 N_SF  =  68, size =  0.56 ( 0.54) +/-  0.03 
 	 N_AGN =  10, size =  0.40 ( 0.43) +/-  0.07 
	 N_ALL =  78, size =  0.54 ( 0.54) +/-  0.03 


$near \ field$: 
 	 N_SF  =  51, size =  0.66 ( 0.64) +/-  0.03 
 	 N_AGN =   8, size =  0.67 ( 0.72) +/-  0.08 
	 N_ALL =  59, size =  0.66 ( 0.65) +/-  0.03 


$field$: 
 	 N_SF  =  16, size =  0.71 ( 0.70) +/-  0.07 
 	 N_AGN =   2, size =  0.68 ( 0.68) +/-  0.05 
	 N_ALL =  18, size =  0.71 ( 0.70) +/-  0.06

RESULT:

The ratio of Re(24)/Re(r) increases as galaxy environment transitions from cluster, to near field, and to field. The difference in size ratios is significant at the $3\sigma$ level.

Interestingly, the cluster and field galaxies are both consistent with the star-forming main sequence, even though the star-formation is more compact on average in the cluster galaxies. Thus, the size of the star-forming region is an important parameter to add to the SFR-Mass analysis when looking at galaxy properties as a function of environment.

Starburst Galaxies



In [383]:

    
s.calc_size_starburst()









    



mean (median) size of starburst galaxies =  0.38 ( 0.35) +/-  0.05
mean (median) size of non-starburst galaxies =  0.64 ( 0.62) +/- 0.0216

 CLUSTER ONLY 

mean (median) size of starburst galaxies =  0.38 ( 0.36) +/-  0.07
mean (median) size of non-starburst galaxies =  0.56 ( 0.55) +/- 0.0300

 FIELD ONLY 

mean (median) size of starburst galaxies =  0.38 ( 0.35) +/-  0.05
mean (median) size of non-starburst galaxies =  0.69 ( 0.70) +/- 0.0252

The Impact of Coma



In [382]:

    
# comparing sizes for sample with Coma removed
nc.plotSFRStellarmassSizeBlue(blueflag=True,plotbadfits=False)
nc.printsize()
ncf1=nc.bluesampleflag & ~nc.membflag & ~nc.agnflag
ncm1=nc.bluesampleflag & nc.membflag & ~nc.agnflag
t=ks(nc.s.SIZE_RATIO[ncf1],nc.s.SIZE_RATIO[ncm1])









    



Cluster: 
 	 N_SF  =  53, size =  0.59 ( 0.56) +/-  0.04 
 	 N_AGN =   5, size =  0.30 ( 0.15) +/-  0.10 
	 N_ALL =  58, size =  0.57 ( 0.55) +/-  0.03 


Near Field: 
 	 N_SF  =  50, size =  0.66 ( 0.65) +/-  0.03 
 	 N_AGN =   8, size =  0.67 ( 0.72) +/-  0.08 
	 N_ALL =  58, size =  0.66 ( 0.67) +/-  0.03 


Field: 
 	 N_SF  =  11, size =  0.66 ( 0.68) +/-  0.08 
 	 N_AGN =   2, size =  0.68 ( 0.68) +/-  0.05 
	 N_ALL =  13, size =  0.66 ( 0.68) +/-  0.07 


KS Test:
D =   0.24
p-vale = 0.06511 (prob that samples are from same distribution)
Anderson-Darling test Test:
D =   1.37
p-vale = 0.08740 (prob that samples are from same distribution)

RESULT:

The difference in size ratios between cluster and field galaxies is not as significant once Coma is removed from the sample. The difference between the field and cluster galaxies is at the $2\sigma$ level. The conclusion is that Coma is important or unique among the clusters in the sample. Perhaps environmental effects are stronger in the more X-ray luminous environment, or maybe the sample size gets too small once Coma is removed (weak, I admit).

Checking for Other Hidden Systematics

Need to make sure that we are not seeing the effect of some parameter that is linked with environment. For example, B/T is strongly correlated with environments.

Check:

stellar mass
B/T
sSFR
B/A



In [380]:

    
# stellar mass

print 'comparing stellar mass for field vs cluster'
f1=s.bluesampleflag & ~s.membflag & ~s.agnflag
m1=s.bluesampleflag & s.membflag & ~s.agnflag
t=ks(s.logstellarmass[f1],s.logstellarmass[m1])

# B/T

print ''
print 'comparing B/T for field vs cluster'
f1=s.bluesampleflag & ~s.membflag & ~s.agnflag & s.gim2dflag
m1=s.bluesampleflag & s.membflag & ~s.agnflag & s.gim2dflag
t=ks(s.s.B_T_r[f1],s.s.B_T_r[m1])

# ssfr

print ''
print 'comparing sSFR for field vs cluster'
f1=s.bluesampleflag & ~s.membflag & ~s.agnflag 
m1=s.bluesampleflag & s.membflag & ~s.agnflag 
t=ks(log10(s.ssfr[f1]),log10(s.ssfr[m1]))

# B/A

print ''
print 'comparing B/A for field vs cluster'
f1=s.bluesampleflag & ~s.membflag & ~s.agnflag 
m1=s.bluesampleflag & s.membflag & ~s.agnflag 
t=ks((s.s.SERSIC_BA[f1]),(s.s.SERSIC_BA[m1]))

# ir surface brightness 

print ''
print 'comparing $L{IR}/R_e(24)^2$ for field vs cluster'
f1=s.bluesampleflag & ~s.membflag & ~s.agnflag 
m1=s.bluesampleflag & s.membflag & ~s.agnflag 
t=ks(log10(s.sigma_ir[f1]),log10(s.sigma_ir[m1]))

# size

print ''
print 'comparing Re(24)/Re(r) for field vs cluster'
f1=s.bluesampleflag & ~s.membflag & ~s.agnflag 
m1=s.bluesampleflag & s.membflag & ~s.agnflag 
t=ks((s.s.SIZE_RATIO[f1]),(s.s.SIZE_RATIO[m1]))









    



comparing stellar mass for field vs cluster
KS Test:
D =   0.18
p-vale = 0.18238 (prob that samples are from same distribution)
Anderson-Darling test Test:
D =  -0.21
p-vale = 0.43278 (prob that samples are from same distribution)

comparing B/T for field vs cluster
KS Test:
D =   0.20
p-vale = 0.15736 (prob that samples are from same distribution)
Anderson-Darling test Test:
D =  -0.27
p-vale = 0.46176 (prob that samples are from same distribution)

comparing sSFR for field vs cluster
KS Test:
D =   0.19
p-vale = 0.16298 (prob that samples are from same distribution)
Anderson-Darling test Test:
D =  -0.25
p-vale = 0.45440 (prob that samples are from same distribution)

comparing B/A for field vs cluster
KS Test:
D =   0.16
p-vale = 0.29447 (prob that samples are from same distribution)
Anderson-Darling test Test:
D =  -0.67
p-vale = 0.70785 (prob that samples are from same distribution)

comparing $L{IR}/R_e(24)^2$ for field vs cluster
KS Test:
D =   0.20
p-vale = 0.12293 (prob that samples are from same distribution)
Anderson-Darling test Test:
D =   0.25
p-vale = 0.26780 (prob that samples are from same distribution)

comparing Re(24)/Re(r) for field vs cluster
KS Test:
D =   0.30
p-vale = 0.00303 (prob that samples are from same distribution)
Anderson-Darling test Test:
D =   4.54
p-vale = 0.00521 (prob that samples are from same distribution)

RESULT:
Difference in size ratio between cluster and field galaxies can't be explained by systematic differences in stellar mass, B/T, sSFR, or B/A.

IR Surface Brightness



In [352]:

    
figure(figsize=(12,5))
subplot(1,2,1)
subplots_adjust(wspace=.5)
pcolor=s.s.CLUSTER_LX
pcolorlabel='$log_{10}(L_X)$'
#pcolor=s.s.SIGMA_5
#pcolor=sqrt(s.s.DR_R200**2 + s.s.DELTA_V**2)
#pcolor=10.**s.logstellarmass
#pcolor=s.massdensity
#pcolor=s.s.B_T_r
#pcolorlabel='$log_{10}(B/T)$'

f=s.bluesampleflag & ~s.agnflag & s.membflag 
scatter(log10(s.sigma_ir[f]),log10(s.ssfr[f]*1.e9),c=log10(pcolor[f]),s=60)
f=s.bluesampleflag & ~s.agnflag & ~s.membflag 
scatter(log10(s.sigma_ir[f]),log10(s.ssfr[f]*1.e9),c=log10(pcolor[f]),s=60,marker='^')
#f=s.bluesampleflag & ~s.agnflag & s.fieldflag
#plot(log10(s.sigma_ir[f]),log10(s.ssfr[f]*1.e9),'ks',mfc='None',label='Field')
colorbar(fraction=.08,label=pcolorlabel)
xlabel('$log_{10}(\Sigma_{ir}) $')
ylabel('$sSFR/Gyr$')

subplot(1,2,2)
#pcolor=s.s.CLUSTER_LX
#pcolorlabel='$log_{10}(L_X)$'
#pcolor=s.s.SIGMA_5
#pcolor=sqrt(s.s.DR_R200**2 + s.s.DELTA_V**2)
#pcolor=10.**s.logstellarmass
#pcolor=s.massdensity
pcolor=s.s.B_T_r
pcolorlabel='$log_{10}(B/T)$'

f=s.bluesampleflag & ~s.agnflag & s.membflag & s.gim2dflag
scatter(log10(s.sigma_ir[f]),log10(s.ssfr[f]*1.e9),c=log10(pcolor[f]),s=60)
f=s.bluesampleflag & ~s.agnflag & ~s.membflag & s.gim2dflag
scatter(log10(s.sigma_ir[f]),log10(s.ssfr[f]*1.e9),c=log10(pcolor[f]),s=60,marker='^')
#f=s.bluesampleflag & ~s.agnflag & s.fieldflag
#plot(log10(s.sigma_ir[f]),log10(s.ssfr[f]*1.e9),'ks',mfc='None',label='Field')
colorbar(fraction=.08,label=pcolorlabel)
xlabel('$log_{10}(\Sigma_{ir}) $')
#ylabel('$sSFR/Gyr$')









    Out[352]:





<matplotlib.text.Text at 0x12908e810>



In [353]:

    
figure()

#pcolor=s.s.CLUSTER_LX
#pcolorlabel='$log_{10}(L_X)$'
#pcolor=s.s.SIGMA_5
#pcolor=sqrt(s.s.DR_R200**2 + s.s.DELTA_V**2)
#pcolor=10.**s.logstellarmass
#pcolor=s.massdensity
pcolor=s.s.B_T_r
pcolorlabel='$log_{10}(B/T)$'

f=s.bluesampleflag & ~s.agnflag & s.membflag & s.gim2dflag
scatter(log10(s.sigma_ir[f]),log10(s.ssfr[f]*1.e9),c=log10(pcolor[f]),s=60,label='Cluster')
f=s.bluesampleflag & ~s.agnflag & ~s.membflag & s.gim2dflag
scatter(log10(s.sigma_ir[f]),log10(s.ssfr[f]*1.e9),c=log10(pcolor[f]),s=60,marker='^',label='Field')
#f=s.bluesampleflag & ~s.agnflag & s.fieldflag
#plot(log10(s.sigma_ir[f]),log10(s.ssfr[f]*1.e9),'ks',mfc='None',label='Field')
colorbar(fraction=.08,label=pcolorlabel)
xlabel('$log_{10}(\Sigma_{ir}) $')
ylabel('$log_{10}(sSFR/Gyr)$')
legend(scatterpoints=1,loc='upper left')









    Out[353]:





<matplotlib.legend.Legend at 0x12993c7d0>

RESULT:

Compactness of star-forming region is correlated with B/T. However, you would expect the two quantities to be correlated. The stellar mass density must increase as B/T increases, and stellar mass density is correlated with $\Sigma_{IR}$ because stellar mass correlated with SFR, and optical size correlates with IR size.

END OF PAPER 1 PLOTS

PAPER 1

Present analysis procedure
- simulation results
statistics on size
Size of blue galaxies vs.
- local density or $\Delta r$ - correlated
- stellar mass - not correlated
- B/T - correlated

RESULTS

size of SF region is smaller than stellar disk
size is correlated with $B/T$
size is also correlated with local density and projected radius
blue cluster galaxies have smaller size ratios, and this is not explained just with $B/T$ because field and cluster sample have similar distributions of $B/T$.
This indicates that the relative size of the SF region is linked with structural property of galaxies, but there is an additional impact of environment.

PAPER 2

Add SFRs to analysis. Focus on size, stellar mass, SFR relation

field/cluster samples are comparable in terms of stellar mass, B/T, SFR
cluster galaxies have smaller size
cluster and field galaxies are still consistent with the SF main sequence

RESULTS

blue galaxies in dense environments have sSFRs that are consistent with the SF main sequence despite having more centrally-concentrated SF disks.
starburst galaxies have more centrally-concentrated SF disks when compared with galaxies on the SF main sequence, and relative size of SF disk does not appear to vary with environment for these galaxies.
Coma has an unusually high fraction of starburst galaxies



In [ ]: