About the arc from 'test.fits'

[WentaoLuo]

Is there a lens galaxy or the smoothed profile of the dark matter that creates the arcs?

[linan7788626]

yes, the mass model of this lens is Non-singular Isothermal Ellipsoid, and the velocity dispersion is 320 km/s.

[linan7788626]

@WentaoLuo try this http://www.syntevo.com/smartgit/ for using github under linux.

[WentaoLuo]

Thanks, I am fine with git now after one night training. It is basically SSH.

[WentaoLuo]

I am wondering if we could combine the arc image and simulated SDSS image together and form a more realistic one?

[WentaoLuo]

And I would then simulate more image, maybe 10000. And add others to th github.

[linan7788626]

hold on, I have question, is sdssgal is the image with noise? you told me the noise should be white noise, but in this fits, it is not flat...

[WentaoLuo]

It is Gaussian. With a galaxy image in the center.

[WentaoLuo]

And also notice that the image size is not 120x120, instead. It is 119x119, or things like that due to the scaling from cosmos to SDSS by SHERA.

[WentaoLuo]

When do you think we add other people?

[linan7788626]

How do you identify the parameters of the gaussian distribution? it should be related to the dark matter halo we are using. Also, please set the images size to be 128x128. it should be easier for both of us.

When do you think we add other people? At least, after producing one simulated images. Do you have any suggestion?

[WentaoLuo]

I will go to some details later. Sure, 128x128 sounds good.

[linan7788626]

for the lensed images using toy models, you do not need to worry about the image of the lens galaxy, please just give noise and PSF.

[WentaoLuo]

The Gaussian noise is 58.7 in terms of variance (in unit of count^2, photon-electron counts), and the mean is 0.0.

[WentaoLuo]

I will update the sdss galaxy image soon with 128x128.

[linan7788626]

the PSF you give me is not normalized......

[WentaoLuo]

You mean the size is not 128x128?

[WentaoLuo]

Or I could try combine those images together?

[WentaoLuo]

The convolution between them in IDL does not require the same size for both images.

[linan7788626]

no, I mean the summary of all pixels is not one.

[linan7788626]

the values of the pixels in the sdssgal.fits is too high, what is the units of the value of these pixels?

[WentaoLuo]

Both is galaxy image and PSF image there are softbias=1000.0, before any scale. the images should subtract 1000.0. For the psf, after subtraction, scale the image using psf/total(psf).

[linan7788626]

all right, before we combine them together, the most important thing is that we have to make sure we are using the same unit in all images.....

[linan7788626]

I updated a new lensed image which is in ./output_fits/, it looks realistic, but the flux of source and lens are incorrect. And another issue, noise should be added separately, but not added with lens galaxy. otherwise, how could we convolve PSF for the lens galaxy. Anyway, please tell me what is the unit of the value of the pixels in the fits files you gave me.

[WentaoLuo]

what I use is count of photon-electron, produced by CCD. It mimics SDSS detector. The count magnitude relation is :http://classic.sdss.org/dr7/algorithms/fluxcal.html. f/f0 = counts/exptime * 10^{-0.4_(aa + kk * airmass)}, And number of photon-electron=count_gain, gainr=4.7 mag = -2.5 * log10(f/f0) SDSS exposure time 54 minutes. We simply set {-0.4(aa+kk_airmass)) to be -9.6 meaning we fix a zero-point and airmass.

Then, given an wanted SDSS magnitude MAG MAG--->f/f0--->counts.

By the way, I think the unit we use should be counts rather than photon-electron, so we do not need gain_r value.

Sorry for that.

[WentaoLuo]

Nice work, man!

[WentaoLuo]

We should definitely add the others to this project, so we could brainstorm. What do you think. I think we are ready to simulate a larger set of images, don't we?

[linan7788626]

I know we should add others to this project, how about later this week, before that we have a couple of things to fix:

• The image of lens galaxy without noise.
• Calibrating the magnitudes of source galaxies and lens galaxies.
• Modeling the surface brightness of lens galaxies according to the velocity dispersion of lenses.

If we do not figure our these problems, the images we produced are not naturally right. After this, we add them in the collaborators list. Does it make sense ?

[WentaoLuo]

Yes, you are right. We should at least fix the above three issues before we add them.

[WentaoLuo]

I believe the velocity dispersion you need of the lens galaxy should be taken as the dispersion of the dark matter halo with an SIE profile?

[linan7788626]

the velocity dispersion I am using here is the velocity dispersion of stars in the lens galaxy. It is also the velocity dispersion of the total mass distribution, not dark matter halo only.

On Dec 8, 2015, at 3:56 PM, Wentao Luo notifications@github.com wrote:

I believe the velocity dispersion you need of the lens galaxy should be taken as the dispersion of the dark matter halo with an SIE profile?

— Reply to this email directly or view it on GitHub https://github.com/linan7788626/Arc_Finding_with_CMU/issues/2#issuecomment-163030639.

[WentaoLuo]

Or you mean the "the velocity dispersion of the lenses" are the dispersion of the dark matter halo? And then using Faber-Jackson relation, we could estimate the luminosity of the galaxy reside in the halo?

[WentaoLuo]

I see.

[linan7788626]

do not be struggling in dark matter halo or what, let’s use total mass distribution here.

On Dec 8, 2015, at 4:00 PM, Wentao Luo notifications@github.com wrote:

I see.

— Reply to this email directly or view it on GitHub https://github.com/linan7788626/Arc_Finding_with_CMU/issues/2#issuecomment-163031806.

[WentaoLuo]

I am on it. velocity dispersion----->luminosity(Faber-Jackson relation)----> SDSS CCD counts.

[WentaoLuo]

Alright, I will add the according calibration python code to our repository.

[linan7788626]

how you model the surface brightness distribution? use 1/4 law please , if you can.

[WentaoLuo]

Sure, n=1/4. typical elliptical galaxy profile, a.k.a de Vaucouleurs profile.

[linan7788626]

yes

On Dec 8, 2015, at 4:06 PM, Wentao Luo notifications@github.com wrote:

Sure, n=1/4. typical elliptical galaxy profile, a.k.a de Vaucouleurs profile.

— Reply to this email directly or view it on GitHub https://github.com/linan7788626/Arc_Finding_with_CMU/issues/2#issuecomment-163033182.

[WentaoLuo]

I hope your computer will be fine. I decide to make generate the lens galaxy profile using the following steps: 1, Fundamental plane: given a velocity dispersion(no scatter first)--> R_e, miu_e, luminosity 2, de Vaucouleurs profile with certain R_e *3, We finally need to take the scatter of the fundamental plane into consideration, with assigning a random Number from Normal distribution with the same scatter of a certain scaling relation, e.g. miu_e vs R_e, or vel_dis vs miu_e, e.t.c.

The attached is an example of fundamental plane,

/home/wtluo/Desktop/figure2.jpg

[WentaoLuo]

[linan7788626]

it looks great, everything sounds reasonable, but I have one question: how we identify the parameter I_0 of the de Vaucouleurs profile profile? we need to use I_0 and R_e to describe the profile fully.

[WentaoLuo]

Do not worry, in that profile, the I_0 is actually I_e. People build the fundamental plane to make thins easier rather than more complex. See the link :https://en.wikipedia.org/wiki/De_Vaucouleurs'_law. The only thing is the empirical number: 7.669, which change the I_0 to I_e.

Once we have the fundamental plane, everything is good, except for the scatter.

[linan7788626]

Great, I am looking forward to your result of the lens galaxies, please do not forget the ellipticity and orientation, if possible just set them as input parameters as well. here e = 0.8 , phi = 45 degree.

On Dec 10, 2015, at 11:03 AM, Wentao Luo notifications@github.com wrote:

Do not worry, in that profile, the I_0 is actually I_e. People build the fundamental plane to make thins easier rather than more complex. See the link :https://en.wikipedia.org/wiki/De_Vaucouleurs'_law https://en.wikipedia.org/wiki/De_Vaucouleurs'_law. The only thing is the empirical number: 7.669, which change the I_0 to I_e.

Once we have the fundamental plane, everything is good, except for the scatter.

— Reply to this email directly or view it on GitHub https://github.com/linan7788626/Arc_Finding_with_CMU/issues/2#issuecomment-163688422.

[WentaoLuo]

Sorry, I_o is not I_e. But can be scaled to I_e with some weird number 7.669.

[WentaoLuo]

Sure, I will take ellipticity into consideration.

[linan7788626]

could you Point me to the reference for identify I_e R_e and I_0 , please. Thanks

On Dec 10, 2015, at 11:06 AM, Wentao Luo notifications@github.com wrote:

Sorry, I_o is not I_e. But can be scaled to I_e with some weird number 7.669.

— Reply to this email directly or view it on GitHub https://github.com/linan7788626/Arc_Finding_with_CMU/issues/2#issuecomment-163689141.

[WentaoLuo]

AS3011_5.pdf

I_e=I_0exp(-7.669)

[WentaoLuo]

We need to find out a paper that describe this relation. The PDF is only from a lecture.

[WentaoLuo]

An example of Poisson noise

[WentaoLuo]

An example of Gaussian noise.