Read eBOSS fiberpos

[londumas]

For the eBOSS option of the mocks: see PR desihub/desisim#481 and PR desihub/specsim#100 and issue desihub/desisim#437.
We need a dummy eBOSS fiberpos, I created it via a copy of $DESIMODEL/data/focalplane/fiberpos.fits, keeping only the first two spectrographs, getting a random x and y position within a radius of 325. mm, within the two half-plates.
How @sbailey, do you want me to parse this file?
I was thinking of an option set to None by default, and if given read eBOSS file. Does that sound reasonable? Also where should I copy the fits file?

desimodel/py/desimodel/io.py

Line 117 in afc16d2

def load_fiberpos():

import fitsio
import scipy as sp

def main():

    radius = 325.
    outpath = 'eboss_fiberpos.fits'
    p = '$DESIMODEL/data/focalplane/fiberpos.fits'

    sp.random.seed(42)

    h = fitsio.FITS(p)
    print(h[1].read_header())

    ### Store
    cat = {}
    for k in h[1].get_colnames():
        cat[k] = h[1][k][:]

    w = cat['SPECTRO']<2
    for k in cat.keys():
        cat[k] = cat[k][w]

    theta = sp.pi*sp.random.rand(500)
    r = radius*sp.sqrt(sp.random.rand(500))
    x = r*sp.cos(theta)
    y = r*sp.sin(theta)
    cat['X'][cat['SPECTRO']==0] = x
    cat['Y'][cat['SPECTRO']==0] = y

    theta = (2*sp.pi-sp.pi)*sp.random.rand(500)+sp.pi
    r = radius*sp.sqrt(sp.random.rand(500))
    x = r*sp.cos(theta)
    y = r*sp.sin(theta)
    cat['X'][cat['SPECTRO']==1] = x
    cat['Y'][cat['SPECTRO']==1] = y

    print(cat['X'],cat['Y'])
    import matplotlib.pyplot as plt
    plt.errorbar(cat['X'],cat['Y'],fmt='o')
    plt.grid()
    plt.show()


    ### Save
    out = fitsio.FITS(outpath,'rw',clobber=True)
    cols = [ v for v in cat.values() ]
    names = [ k for k in cat.keys() ]
    out.write(cols,names=names)
    out.close()

    return

main()

[sbailey]

Could you provide some more context about why we need a fiberpos file for eBOSS when eBOSS doesn't have fiber positioners at fixed locations? i.e. specifically what doesn't work if this file doesn't exist? It smells like we're using the fiber positioners file as a convenience for interpolating x, y, z, platescale, etc. when maybe that should be parameterized differently.

[belaa]

@londumas desisim.simexp.simulate_spectra() has a specsim_config_file parameter - if that's set to 'eboss' maybe we can force it to generate these fiber positions on the fly, rather than having to add a separate file to desimodel?