snapshot : which redshift snaptshot to use. 85 is the lowest z~0.

shapes_method : method used to compute galaxy shapes, allowed: inertia_tensor, reduced_inertia_tensor, spin

postprocessed : FITS file in which to store the output of step 2 above.

shapes_dir : Directory to which to write the output of step 1 above. Also tells the code where to look for the inputs to step 2. 

cuts : column names, lower then upper bounds

halo_matching : 

catalogue : Input unsymmetrised catalogue. Should be the path to a FITS file of the format ouput by step 2 above.

output : Path to a FITS file to which we want to write the symmetrised catalogue.

pivot_type : Whether to rotate about the centre of mass or the centre-most galaxy, allowed : mass, galaxy

central_type : Flag used to define the central galaxy, allowed: spatial_central, most_massive

 nsub : number of jackknife volumes to use to calculate errorbars.

 ctypes : whitespace separated list of two point correlations to calculate. 
         See Mandelbaum et al 2010 (https://arxiv.org/pdf/0911.5347.pdf)
         and Joachimi et al 2010 (https://arxiv.org/pdf/1008.3491.pdf)
         
         Allowed:
         
         ed : orientation - separation vector 3D correlation
         ee : orientation - orientation 3D correlation
         gg : position - position 3D correlation
         iiplus_3d : eq 10a of MB10
         iiminus_3d : eq 10b of MB10
         giplus_3d : eq 8 of MB10
         giminus_3d :
         iiplus_proj : 
         iiminus_proj : 
         giplus_proj : eq 12 of J10
         giminus_proj : 

errors : Whether or not to calculate jackknife errorbars, bool

shapes : location of the input shape/position catalogue

savedir : place to save the output text files

split : quantity by which to split the catalogue before calculation (for example, for central/satellite separation).

split_val : value of the above quantity about which to split the catalogue

rmin : minimum 3D separation

rmax : maximum 3D separation

nbin : number of logspaced bins to use

rpmin : minimum perpendicular separation

rpmax : maximum perpendicular separation

nrpbin : number of logspaced rp to use

pi_max : upper limit of line of sight integration for projected statistics

       id

/header.txt

the value of flag_double(header) is important. It decides the precision of
the particle properties

the value of redshift(header) is also handy. 

use SnapDir.readgroup() to read in the tab.

groupdtype = numpy.dtype([
    ('mass', 'f4'), 
    ('len', 'i4'), 
    ('pos', ('f4', 3)), 
    ('vel', ('f4', 3)), 
    ('nhalo', 'i4'),   
    ('massbytype', ('f4', 6)),  
    ('lenbytype', ('u4',6)), 
   ])

One group per entry, ordered by number of particles

one subhalo per entry, plus one entry for the contamination
For example, if the first group has nhalo = 100, then there will
be 101 subhalo entries. 

This is indicated in subhalo/type.raw (u1). If the type is 0
then entry is a subhalo. if the type is 1, the entry is a contamination.

    load('subhalo', 'type') reads in the types.

** Be aware of the contaminations for subhalo analysis **

    load('subhalo', 'tab') reads in the tab.

subdtype = numpy.dtype([
    ('mass', 'f4'),  # mass, nan for contamination
    ('len', 'i4'), 
    ('pos', ('f4', 3)), 
    ('vel', ('f4', 3)),   
    ('vdisp', 'f4'),  
    ('vcirc', 'f4'),  
    ('rcirc', 'f4'),  
    ('parent', 'i4'),  
    ('massbytype', ('f4', 6)),  
    ('lenbytype', ('u4',6)), 
    ('unused', 'f4'),    
    ('groupid', 'u4'), 
   ])
 
the luminosity of subhalos at various band are also available (described in
    next section).

For easying the access, we have performed some joint queries and provide
the following extra properties of subhalos:

 ( use with SnapDir.load('subhalo', propertyname) )

subhalo/type.raw  : int8 array of type of the entry: 
                    0 for subhalo 
                    1 for contamination
subhalo/sfr.raw  : float32 array of total star formation of each entry
subhalo/bhmass.raw : float32 array of total bhmass of each entry
subhalo/bhmdot.raw : float32 array of total bhmdot of each entry
subhalo/bhmassive.raw : property of most massive bh of each entry
subhalo/bhluminous.raw : property of most luminous bh of each entry

the type of entries in bhmassive and bhluminous are:
    bhdtype = numpy.dtype([
        ('pos', ('f4', 3)),
        ('vel', ('f4', 3)),
        ('id', 'u8'),
        ('bhmass', 'f8'),
        ('bhmdot', 'f8')])
    bhmass is NaN if there is no bh in that subhalo or contamination.

use SnapDir.load(particletype, blockname)
   [optionally give g=subhalotab or g=grouptab to access groups/subhalos], see doc.

Particle types are 
   0: Gas
   1: Dark mater    
   2&3: unused
   4: Star
   5: BH

The following blocks have fixed data type:
    ('pos', ('f4', 3)),
    ('vel', ('f4', 3)),
    ('mass', 'f4'),
    ('id', 'u8'),
    ('type', 'u1')

The other blocks, currently including 
    0/sfr.raw 
    4/met.raw 
    5/bhmass.raw
    5/bhmdot.raw

are floating numbers, width of witch decided by 
the value of flag_double in header.txt.
(simply match the flag_double line)
float64 : if flag_double is 1
float32 : if flag_double is 0

One particle per entry in any of the blocks.
The particles are ordered in chunks; one chunk per subhalo/contamination.
Look at the lenbytype value in subhalo tab or group tab.

For example, 
if the first group has lenbytype == [50000, 100000, 0, 0, 50000, 100],
and the second group has lenbytype == [10000, 20000, 0, 0, 10000, 10],

0:50000 in 0/*.raw are gas in the first group,
0:100000 in 1/*.raw are dark matter in the first group,
0:50000 in 4/*.raw are star in the first group,
0:100 in 5/*.raw are black hole in the first group.

50000:60000 in 0/*.raw are gas in the second group,
100000:12000 in 1/*.raw are dark matter in the second group,
50000:60000 entries in 4/*.raw are star in the second group,
100:110 entries in 5/*.raw are black hole in the second group.

The full SED spectra of subhalos is mocked according to Wilkins et al. 2013.
use SnapDir.readSED to read the SED for a given subhalo. Note that
if the subhalo has no stars there is no SED and the function returns 0
the returned SED has nebula lines.

We also have the band luminosity of the subhalos, integrated from the SED.

For subhalos with no stars, and for the subhalo entry corresponding to
contamination, all bands are zero.

Files are named by for example sbuhaloHST.ACS.raw.
Refer to readsubhalo.py and bandfilterdtype for all the available bands.

Two categories of filters `RfFilter' for restframe, and 
`ObsFilter' for observed frame.

Subhalos:
--------
The band luminosity can be loaded like other 'subhalo' variables, eg, 

    SnapDir.load('subhalo', 'RfFilter/SDSS.SDSS/i')

    (saved  under RfFilter/SDSS.SDSS/i , in 'f4')

    They are in f4 of and of unit 1e28 Erg/s/Hz/h.

    L_band = nu_mean * L_nu_mean / Delta mu

    The magnitude of the band is given by:
    
    M = -2.5 * numpy.log10(L) + 34.1  # L in SI, Joule

Star particles:
----
The band luminosity of individual star particles can be loaded via

    (eg)
    SnapDir.loadstarLband('RfFilter', 'SDSS.SDSS/r')

IGM absoprtion and nebula lines are included.

The bands are saved in a compressed format:
 1. The filtered band luminosity per unit mass of 260 time/metallicity 
    bins are calculated and saved in SED/SEDtableXXXX.
 2. 4/SEDindex.raw saves the index of the table entry that saves the
    band luminosity per unit mass.   (Erg/s/Hz/[1e10Msun])
 3. Multiply the entry by the mass of the star particle
    gives its true band luminosity.

 See Snapdir.loadstarLband for more details.