Issue with some templates

[saimn]

I have several templates (for LyA, CIII) that Marz refuse to process. It doesn't show the cross-correlation and the template, though I can still manually click on the line and then I get the correct z and see the template.

So after diving in the code, I don't understand the second part of TemplateManager.prototype.shiftTemplate and the relation between the redshift of the template, z_start, z_end, and the startZIndex/endZIndex you are computing (and maybe also globalConfig.startPower, globalConfig.endPower).

Let's take an example: I have a LyA template, where:

    "start_lambda": 900.0,
    "end_lambda": 1895.769591008408,
    "redshift": 0.0,
    "z_start": 2.7,
    "z_end": 6.4,
    "shift": false,

And

globalConfig.startPower = 2.8;
globalConfig.endPower = 4.0;
globalConfig.arraySize = 32768;

When I look at shiftTemplate, it computes t.startZIndex = 31900 and t.endZIndex = null, which seems not good. If I use only these templates, I get many errors in coalesceResults because the templates .peaks is empty.
Is there some setting somewhere that should be adjusted for highz templates, or other ideas on what to look at ?

[Samreay]

Ah right, so what is happening is that all templates get rebinned to a
log-linear wavelength (array size 2^32) to far xcorr. So what you would
need to do is increase the endPower to a large enough number (I would try
4.2 and go up in 0,1 increments until it works) that the redshift range you
want is captured by the array.

Given the high redshift, it might be useful to increase the arraySize by a
factor of two, to make sure you dont lose precision with lower redshift
templates (however you may not need to do this, I have not tested the
impact of precision, array size and redshift for such high redshift ranges).

Apologies for the hassle getting this to work!

On Fri, May 20, 2016 at 2:19 PM, Simon Conseil notifications@github.com
wrote:

I have several templates (for LyA, CIII) that Marz refuse to process. It
doesn't show the cross-correlation and the template, though I can still
manually click on the line and then I get the correct z and see the
template.

So after diving in the code, I don't understand the second part of
TemplateManager.prototype.shiftTemplate and the relation between the
redshift of the template, z_start, z_end, and the startZIndex/endZIndex you
are computing (and maybe also globalConfig.startPower,
globalConfig.endPower).

Let's take an example: I have a LyA template, where:

"start_lambda": 900.0,
"end_lambda": 1895.769591008408,
"redshift": 0.0,
"z_start": 2.7,
"z_end": 6.4,
"shift": false,

And

globalConfig.startPower = 2.8;
globalConfig.endPower = 4.0;
globalConfig.arraySize = 32768;

When I look at shiftTemplate, it computes t.startZIndex = 31900 and t.endZIndex
= null, which seems not good. If I use only these templates, I get many
errors in coalesceResults because the templates .peaks is empty.
Is there some setting somewhere that should be adjusted for highz
templates, or other ideas on what to look at ?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub
#135

[saimn]

Ah thanks, I tried several things but didn't go far enough for endPower: with 4.6 it is working ! I also increased arraySize, thanks for the advice.
Don't apology, your software is really good (particularly with these highz templates now working) ! 👏
It is normal to have a few adaptations to make for different data.