This study compares the structural evolution of galaxies from observations in the Cosmic Assembly Near-infrared Deep Extragalactic Legacy Survey (CANDELS) to cosmological simulations with and without radiative pressure (RP) feedback. Axis ratio distributions are obtained from simulated galaxies of different stellar masses, sizes, and redshifts and compared to observed distributions from CANDELS. Preliminary results show axis ratios decrease towards higher redshifts for simulated galaxies viewed face-on, similarly to CANDELS observations, and simulations with RP feedback generally have lower axis ratios than without at z > 1.5. More analysis is needed to determine how RP feedback affects galaxy structure and morphology over cosmic time.

1. Structural Evolution of Galaxies from Cosmic Assembly Near-
infrared Deep Extragalactic Legacy Survey and Cosmological
Simulations
Vivian Tang1, Yicheng Guo2, Joel Primack3
1 Department of Astronomy and Astrophysics, University of California, Santa Cruz
2 UCO/Lick Observatory, Dept. of Astronomy and Astrophysics, University of California, Santa Cruz
3 Department of Physics, University of California, Santa Cruz
Introduction Methods
Results
Conclusion
Future Work
Acknowledgement
The main motivation for this project is to systematically compare projected axis ratio distributions under random
viewing angles in simulated galaxies to those of the observed galaxies which have unknown viewing angles. In
particular, we focus on galaxies in the range of redshift 1 to 3 in both simulation and observation.
In both observations and simulations, wealth of data allow us to do statistical studies. By looking at how the
distributions change over cosmic time, we can statistically determine how the shape and formation of galaxies
evolve.
Study done previously by van der Wel et al. have shown that observed galaxies in the lower-mass range from
CANDELS are elongated, rather than disk-like or spheroidal, and the axial ratio is skewed for higher redshift
galaxies.
To investigate this observation, we obtain the axis ratio distributions from simulated galaxies as a function of
redshift, stellar mass, size, and Sersic index with and without RP feedback and compare to those found by van
der Wel et al.
We suspect that simulated galaxies with RP feedback represent observation more correctly, thus we will also
compare the axis ratio distributions of simulated galaxies with and without RP feedback to see how RP feedback
affect each of the parameters. A subset of 35 different simulations with and without RP feedback is used for
analysis.
III. The Project: Observations vs. Simulations
T o u n d e r s t a n d
s t r u c t u r a l a n d
m o r p h o l o g i c a l
e v o l u t i o n o f
galaxies, we need to
look far back in time
and compare the
differences between
older and younger
galaxies.
Early star forming
g a l a x i e s h a v e
distorted, irregular
appearance while
present day star
forming galaxies
tend be flat disks
a n d h a v e s p i r a l
arms.
I. Observation: CANDELS
II. Cosmological Simulations: With and without
Radiative Pressure (RP) Feedback
References
I. CANDELization
High resolution images of simulated galaxies need to go
through a process called CANDELization before the analysis.
To CANDELized an image, the Sunrise radiative transfer code
is applied, as well as the Point Spread Function for the
relevant Hubble Space Telescope observations. Noise is also
added and the end result is an image with the same resolution
to those of the observed galaxies from Hubble Space
Telescope.
It is necessary for simulated images to be CANDELized since
in reality we do not observe distant galaxies with high
resolution.
II. GALFIT
After CANDELizing images, GALFIT, a data analysis algorithm that fits
2D analytic functions to galaxies, is used for image analysis. Each
function in GALFIT corresponds to a component that creates a model
for image fitting. For this project we have chosen to use a single
component fit.
Output from GALFIT contains best fit parameters which can be used
for statistical studies. These include the Sersic index, axis ratio, and
effective radius.
The Sersic index describes radial distribution of light; the effective
radius is defined as containing half of the total luminosity from the
galaxy, and the axis ratio is simply the minor axis divided by major
axis.
FIGURE 3: Model shape parameter defined by colour [4]
Determine whether simulations with RF is a more accurate representation
than those without RF by comparing findings to the axis ratio distributions
obtained previously by van der Wel et al.
To obtain axis ratio as a function of stellar mass and redshift of simulated
galaxies, in those that have been analyzed for this project as well as other
simulations by Daniel Ceverino et al. to see whether lower-mass galaxies
are elongated as observed by CANDELS.
Galaxies used for this project were selected based on acceptable
uncertainties from GALFIT. We will compare those uncertainties for the face-
on and edge-on galaxies to determine whether it has an effect on our
previous selections.
CANDLES is the largest project in history of Hubble Space Telescope with 902
assigned orbits of observing time. Equivalent to observing 4 months
consecutively, it captures images of galaxies far in deep space at high redshifts
as well as those in the local universe.
In astronomy, redshift is used to determine how far objects are in the skies. By
measuring the wavelength of emitted light from stars and galaxies, astronomer
can determine their ages, speeds, and distances away from us. The higher the
redshift, the further away they are from us.
From observation, galaxies from early epochs were smaller than present day.
They were bluer than older galaxies as the stars were younger and hotter than
older galaxies, which tend to be redder.
All simulated galaxies with and without RP feedback by Daniel Ceverino et al. [3] were done using the ART code
with resolution of 17-34 pc. In these simulations, all the input physics are known and galaxies can be viewed
under different orientations.
In galaxy simulations, an effect of RP feedback is to suppress star formation. Those that were simulated without
the mechanism of RP feedback tend to overproduce stars, which does not match well with observations. And
those that were simulated using RP feedback produced less stars, by about a factor of 2 at all redshifts.
FIGURE 1: Hubble Sequence; dividing galaxies into groups based on visual appearance. [1]
FIGURE 2: Relationship between
redshift and age of galaxies. [2]
FIGURE 4: van der Wel et al. have shown that lower-mass galaxies at higher
redshift are more elongated [4]
FIGURE 5: Before (left) and after (right) CANDELization
FIGURE6: Elliptical like
structure corresponds to a
Sersic index value of n=4.
Spiral, bulgeless disk like
structure corresponds to a
Sersic index value of n=1. [5]
FIGURE7: Axis ratio=b/a. [6]
I. Comparing Simulation with and without RP feedback
II. Face-on viewing angle
VELA04 VELA04MRP
We have chosen images of galaxies with face-on orientation which
would yield a higher axis ratio if they are disk-like or spherical, and a
lower axis ratio if they are elongated.
Images of face-on galaxies (VLEA04 and VELA04MRP) where the
axis ratios were measured to be the lowest.
For the effective radius, each of the three pairs of simulations (with and
without radiative pressure feedback) provides different indications as to
how the size evolves over time.
The axis ratios are generally lower for the RP simulations for z > 1.5; this
might be relevant to comparison with the van der Wel et al. observations.
For the face-on viewing angle, axis ratios decrease towards higher
redshifts. Although this is similar to van der Wel et al. observations, more
analysis are needed for verification.
More data analysis between simulations with and without RP are required
in order to determine how much RF changes the structure and morphology
of galaxies.
FIGURE 8: Axis ratio distributions of star forming galaxies from CANDELS. [4]
0 0.5 1 1.5 2 2.5
0
1
2
3
4
5
6
7
VELA04
VELA04MRP
Z
SersicIndex
0 0.5 1 1.5 2 2.5
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
VELA04
VELA04MRP
Z
AxisRatio
0 0.5 1 1.5 2 2.5
0
2
4
6
8
10
12
14
VELA04
VELA04MRP
Z
EffectiveRadius
0 0.5 1 1.5 2 2.5 3 3.5 4
0
1
2
3
4
VELA15
VELA15MRP
Z
SersicIndex
0 0.5 1 1.5 2 2.5 3 3.5 4
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
VELA15
VELA15MRP
Z
AxisRatio
0 0.5 1 1.5 2 2.5 3 3.5 4
0
2
4
6
8
10
12
VELA15
VELA15MRP
Z
EffectiveRadius
0 0.5 1 1.5 2 2.5 3 3.5 4
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Face-on
VELA02
VELA02MRP
Z
AxisRatio
0 0.5 1 1.5 2 2.5
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Face-onVELA04
VELA04MRP
Z
AxisRatio
0 0.5 1 1.5 2 2.5 3 3.5
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Face-on
VELA15
VELA15MRP
Z
AxisRatio
0 0.5 1 1.5 2 2.5 3
0
1
2
3
4
5
6
VELA02
VELA02MRP
Z
SersicIndex
0 0.5 1 1.5 2 2.5 3
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
VELA02
VELA02MRP
Z
AxisRatio
0 0.5 1 1.5 2 2.5 3
0
1
2
3
4
5
6
7
8
9
VELA02
VELA02MRP
Z
EffectiveRadius
[1] “CANDELS galaxies reveal the Hubble Sequence throughout the Universe's
history”. August 2013. ESA/Hubble Press Release (http://spacetelescope.org/
images/heic1315e/).
[2] Sandra M. Faber, Henry C. Ferguson, David C. Koo, Joel R. Primack & Trudy E.
Bell. “Staring Back to Cosmic Dawn”. June 2014. Sky & Telescope (http://
hipacc.ucsc.edu/NewsArchive/June2014-S&T-CANDELS-CoverStory.pdf).
[3] Daniel Ceverino et al. 2014, Radiative feedback and the low efficiency of
galaxy formation in low-mass haloes at high redshift, MNRAS 442, 1545.
[4] A. van der Wel 2014, Geometry of Star-forming Galaxies from SDSS, 3D-HST,
and CANDELS, arXiv:1407.4233.
[5] Chien Y. Peng. August 2003. GALFIT User’s Manual (http://
users.obs.carnegiescience.edu/peng/work/galfit/README.pdf).
[6] Amit6. “Ellipse axis.svg”. April 2009. Wikimedia Commons (http://
commons.wikimedia.org/wiki/File:Ellipse_axis.svg).
This research is supported by the Julie Packard Summer Scholarship and the
Ron Ruby Memorial Scholarship. Special thanks to postdoc Yicheng Guo
here at UCSC for his great assistance with GALFIT and also to Professor
Primack for his invaluable advice and support.
VELA simulations do not have RP feedback. VELAMRP simulations have RP feedback.
Sunday, January 18, 2015