Difficult question about photometric redshift...

[steiner]

Im doing a project at the moment involving using redshift data measured photometrically. This is opposed to redshift data that is measured spectroscopically. Could anyone outline the main differences between photometric redshift and spectroscopic redshift?

 

Also, the photometric redshift measurements are obtained using different methods. There's a Neural Network based 'empirical' method and there are template based methods. Some of these template based methods are Bayesian.

 

Can anyone explain what this all means? I have to write a paper up which explains how the methods are different, but the explanations need to be as concise as possible (A sentence or two for each term).

 

Thanks

[Martin]

As I'm sure you know there is a lot of stuff that immediately comes up on google if you say "photometric redshift"

 

I cant do the research for your paper, or list the different versions of photometric redshift. I'll give you my intuitive take, for what it's worth, on the underlying reasoning.

 

most stars are in a normal sequence where bluer means hotter means brighter.

 

and stars typically form in some regular proportion, so much percent very massive, so much percent medium, so much small. So there is a regular power spectrum pattern of so much blue, so much green, so much red.

 

Plus there is ABSORBING of the light going on, by gas clouds in the galaxy itself.

 

But somehow there is a typical pattern that you expect----some amount blue, some green, some red.

 

and if the distance to galaxy is constant, then what it sends is what we receive, so we expect to see that typical pattern.

 

but if the distance has increased while the light was in transit, then what we receive is longer wavelength by the factor 1 +z

and so the pattern is shifted

 

what is different about photometric is that instead of a GRATING that spreads the mix of light out into all its wavelengths, you use FILTERS. bandpass. So you measure the wattage (nanowattage?) comeing thru the red and compare that with what is coming thru the green filter.

 

You might have 5 filters. You look for a pattern. and is the pattern shifted towards the red. and if so by how much is it shifted.

 

I think I am saying what you, steiner, already know. but in case somebody else is reading, or in case you have a different take on it, I want to get it said.

 

As regards your paper, I don't know enough to help. and am kind of busy with other stuff right now. Maybe somebody else here can help with details. It strikes me there is a ton of stuff on google. the work would be to print it out, go thru with a highlighter and mark the key stuff, gather it together into an organized list.

 

BTW I don't see the fundamental difference between using a grating and using filters. In some sense when you use a grating and get spectral lines you can see the brightness of the lines, so it is in some way photometric (you are gauging the amount of light at a given wavelength. so what's not photometric about using a grating?