MNRAS 000, 14 (2016) Preprint 27 January 2017 Compiled using MNRAS LATEX style file v3.

Getting more out of V/Vm than just the mean

Dilip G. Banhatti,1?
1 Zeleni Trg 3A, 10000 Zagreb, Croatia

Accepted XXX. Received YYY; in original form ZZZ

ABSTRACT
Banhatti earlier set down the procedure to derive cosmological number density n(z)
from the differential distribution p(x) of the fractional luminosity volume relative to
the maximum volume, x V/Vm (0 x 1), using a small sample of 76 quasars for
illustrative purposes. The limiting volume Vm is found by using the survey limit (for
example, the faintest flux density S0 ). This procedure is applied here to a bigger sample
of 286 quasars selected from Parkes half-Jansky flat-spectrum survey at 2.7 GHz. The
values of n(z) are obtained for 8 values of redshift z from 0 to 3.5. The function n(z) can
be interpreted as the redshift distribution obtained by integrating the radio luminosity
function (P, z) over luminosities P for the survey limiting flux density S0 = 0.5 Jy. So
(P, z) can be formally derived by differentiating n(z; S0 ) relative to S0 . To actually do
this, the method must be used for many other larger values of S0 to get n(z; S0 ). This
will give meaningful results only for much larger and deeper samples than the present
one.
Key words: cosmology: miscellaneous

1 INTRODUCTION ther on calculating the mean V/Vm value or the luminosity

function via 1/Vm . An example of luminosity function cal-
The luminosity-volume or V/Vm test has traditionally been
culation is Croom et al (2009). Our paper has a different
used only through the mean and standard error of x V/Vm
purpose, and the method we describe is complementary to
for cosmological samples of objects like quasars. Here V is
such calculations.
the volume upto the redshift z of the object or quasar of flux
density S in a sample with limiting flux density S0 , and Vm On the other hand, Lynden-Bell (1971) addresses selec-
the volume upto the limiting redshift z m at which the quasar tion effects in luminosity function calculation from a small
would be as faint as possible and still be in the sample, i.e., sample, using the 3C radio sources then available as an ex-
have flux density S0 . For an unbiased sample, x is calculated ample. Further, Sokolov (1988) treats quite a different as-
for each quasar and the mean and standard error are used to pect of the space distribution of radio sources, namely their
estimate uniformity in the space density of the underlying counts at different observation frequencies relative to an as-
quasar population. sumed average luminosity, rather than a spread of luminosi-
However, Kulkarni and Banhatti (1983) showed that the ties as described by a luminosity function.
differential distributions p(x) of x and n(z) of z are directly No new scientific results are found in this methods pa-
related to each other. Banhatti (2009) illustrated how n(z) per, which merely outlines calculation of the overall redshift
is calculated from p(x) for a small sample of 76 quasars. distribution of the underlying population of radio sources
Here we use a bigger sample for a more realistic calculation. directly from the V/Vm distribution of a representative flux-
The cosmological models used in calculating V , z m and Vm limited sample. The limiting redshifts z m are explicitly used
are only illustrative (see equations (1) and (2) below). Any (in addition to the limiting volume Vm ) in the calculation,
other models could also be used. as will become clear, which gives this method its advantage
In this methods paper, calculation of the redshift distri- (see discussion at the end).
bution n(z) direcly from p(x) p(V/Vm ) is illustrated. Nor- Schmidt et al (1988) suggested use of V/Vm test to ex-
mally n(z) is calculated only from the more complex and amine the space distribution of gamma-ray bursts for homo-
difficult-to-calculate luminosity function (P, z) (see equa- geneity. The test was extensively used for a few years for this
tion (4) below). The direct relation between n(z) and p(x) purpose (e.g., Dezalay et al 1994 and references therein), but
has not been exploited, as the focus has generally been ei- the question of the location of gamma-ray bursts whether
within our Milky Way Galaxy or in other possibly distant
galaxies was not resolved from these studies. Only on dis-
? E-mail: [email protected] covering afterglows of gamma-ray bursts at lower (X-ray, op-

c 2016 The Authors
2 D. G. Banhatti
5 5 5
tical, infrared and radio) photon energies in distant galaxies 4
zm: 0 to 0.3
4
zm: 0.3 to 0.7
4
zm: 0.7 to 1.2

and thereafter measuring redshifts of the parent galaxies did

p1(x)

p2(x)

p3(x)
3 3 3
2 2 2
it become clear that they form a cosmological population. 1 1 1
Use of V/Vm distribution as outlined in this paper will be 0 0 0
5 5 5
rewarding for a sufficiently large well-defined unbiased sam- 4 zm: 1.2 to 1.5
4 zm: 1.5 to 1.8
4 zm: 1.8 to 2.2

ple of gamma-ray bursts. It should be possible to construct

p4(x)

p5(x)

p6(x)
3 3 3

such a sample from the results of space and ground-based 2 2 2

1 1 1
gamma-ray telescopes by carefully taking into account de- 0 0 0

tection methods and thresholds. 5 5

zm: 2.8 to 4.0
5
zm: > 4.0
4 zm: 2.2 to 2.8 4 4

p7(x)

p8(x)

p9(x)
3 3 3
2 2 2
1 1 1
0 0 0
2 SAMPLE OF QUASARS USED AND THE 0.2 0.4 0.6 0.8
x
1 0.2 0.4 0.6 0.8
x
1 0.2 0.4 0.6 0.8
x
1

WORLD MODEL
Figure 1. Frequency polygons for the 9 z m -bins
Drinkwater et al (1997) define the survey and list the prop-
erties of 323 quasars from which 286 can be used for cal-
culating x V/Vm . The sample used is thus 89% complete
relative to the survey, which covers 3.90 sr in the sky. The
small incompleteness is not expected to affect the prelimi- binning are given in Table 1. Also listed are numbers pro-
nary results presented, in any case, only as an illustration portional to the cosmological number densities n(z j ) corre-
in this methods paper. Using the limiting flux density S0 = sponding to the bin mid-points z j . The procedure for cal-
0.5 Jy at 2.7 GHz, the limiting redshift z m is calculated for culating n(z j ) is outlined later. Table 2 presents, for com-
each quasar from its redshift z, = 2.7 GHz flux density S , parison, the same results for the smaller sample of 76 from
and spectral index (defined by d(logS )/d(log), or Wills and Lynds (1978) used by Banhatti (2009), although
equivalently, S ). Although larger samples at deeper the world model used for those calculations is (von Hoerner
mJy levels are available, the sample used has roughly equal (1974)) (q0 , 0 , k, 0 ) = (1/2, 1/2, 0, 0), for which the functions
populations in equal redshift bins (see below). So it is used ` (z) and v(z) are different (see Banhatti 2009).
illustratively in this methods paper. The world model with
the parameters (q0 , 0 , k, 0 ) = (1, 1, 1, 0) as defined by von
Hoerner (1974) is used for the functions of z needed, viz,
the monochromatic luminosity distance ` (z) and volume
3.2 Differential distributions pi (x) of x V/Vm for
v(z). Here q0 is the (normalized) deceleration parameter,
the nine z m -bins
0 the (normalized) density parameter, k the (normalized)
curvature parameter and 0 the (normalized) cosmological For each of the 9 bins, indexed by j = 1 to 9, frequency
constant. See von Hoerner (1974) for details of the normal- polygons pi (x) are plotted (see Figure 1). These polygons
ization and a full specification of the model. The functions are derived from histograms with x = 0.2 from x = 0 to
` (z) and v(z) are: 1, making five x-bins over the range [0, 1] of x. Each fre-
quency polygon has vertices at x = 0.1, 0.3, 0.5, 0.7 and 0.9,
(H0 /c) 2 `2 (, z) = z 2 /(1 + z) (1) , (1) which are the mid-points of the histogram bins. Each poly-
gon is extrapolated to x = 1 with a slightly higher slope.
Cosmological number density n(z j ) is then calculated from
(H0 /c) 3 v(z) = (3/2)(sin1 f (z) f (z) [1 f (z) 2 ]), (2) the formula (Banhatti 2009):
where f (z) = z/(1 + z). Here, c/H0 speed of light / Hubble
constant, defines the linear scale. Note that the (monochro- 9
X
matic) luminosity P is related to the (monochromatic) flux (/3)(c/H0 ) 3 n(z j ) = (Ni /v(z i ))pi (x i j ), (3)
density S via P = S `2 (, z). i= j

where x i j = v(z j )/v(z i ), and is the survey solid angle.

In this formula, Ni are the bin populations of the 9 bins.
3 DERIVING n(z) FROM p(V/Vm ) p(x) The frequency polygons are used for estimating the pi (x i j )
values needed for calculating the sums on the RHS. Details
3.1 Binning the z m -values
of n(z j ) calculation are shown in Table 3. Examples are given
The quasars are first sorted out in increasing order of below.
(N /v(z i ))pi (x i7 ) = (N7 /v(z7 ))p7 (x 77 ) +
P9
z m . The limiting redshift is numerically calculated for each n(z7 ) i=7 i
quasar using Newton-Raphson iteration (Rajarevathi 2007). (N8 /v(z8 ))p7 (x 87 ) + (N9 /v(z9 ))p7 (x 97 ).
Iteration is carried out to five significant places, compati- The p7 (x 77 ), p7 (x 87 ) & p7 (x 97 ) values are interpolated
ble with the accuracy of the available input quantities, es- from the p7 (x) frequency polygon. Thus, for n(z1 ) calcu-
pecially z. The z m -bins are then selected, so as to have lation, there are 9 terms to sum (many of which happen
roughly equal numbers of sources (about 30) each, which to be 0 due to p1 (x i1 ) being 0). For n(z2 ) there are 8
is good enough to derive the differential distribution of terms, and so on. Finally, for n(z9 ) there is only one term:
x V/Vm (0 x 1) for each of the bins. Details of this n(z9 ) (N9 /v(z9 ))p9 (x 99 ) = (31/2.112)2.65 = 38.9 39.

MNRAS 000, 14 (2016)

 More from V/Vm 3

Table 1. Limiting redshifts, their bins, mid-points & populations plus derived cosmological number densities using 286 quasars over 3.90
sr in the sky

z m -bin 0 to 0.3 0.3 to 0.7 0.7 to 1.2 1.2 to 1.5 1.5 to 1.8 1.8 to 2.2 2.2 to 2.8 2.8 to 4.0 > 4.0

z j (bin mid-pt) 0.15 0.5 0.95 1.35 1.65 2.0 2.5 3.4 300*

Bin pop. 29 34 34 32 33 32 30 31 31

j (bin no.) 1 2 3 4 5 6 7 8 9

n(z j ) 48770. 4717. 1560. 1167. 865. 642. 464. 194. 39.

l og[n(z j )] 4.69 3.67 3.19 3.07 2.94 2.81 2.67 2.29 1.59

Table 2. Results of earlier similar calculation for a sample of 76 quasars

z m -bin 0 to 0.8 0.8 to 1.6 1.6 to 2.4 2.4 to 3.2

z j (bin mid-pt) 0.4 1.2 2.0 2.8

Bin pop. 19 31 16 10

j (bin no.) 1 2 3 4

n(z j ) 1307. 255. 67. 22.

l og[n(z j )] 3.12 2.41 1.83 1.34

Table 3. Calculation of n(z(j)). Values in rows labelled i = 1, i = 2 & so on are p(x(ij)).

j 1 2 3 4 5 6 7 8 9

Ni 29 34 34 32 33 32 30 31 31

zj 0.15 0.50 0.95 1.35 1.65 2.00 2.50 3.40 300*

v (z j ) 0.002231 0.03835 0.1251 0.2126 0.2773 0.3492 0.4436 0.5890 2.112

i=1 1

i=2 0.058 1

i=3 0.018 0.307 1

i=4 0.010 0.180 0.588 1

i=5 0.008 0.138 0.451 0.767 1

i=6 0.006 0.110 0.358 0.609 0.794 1

i=7 0.005 0.086 0.282 0.479 0.625 0.787 1

i=8 0.004 0.065 0.212 0.361 0.471 0.593 0.753 1

i=9 0.001 0.018 0.059 0.101 0.131 0.165 0.210 0.279 1

n(z j ) 48770. 4717. 1560. 1167. 865. 642. 464. 194. 39.

4 RESULTS, DISCUSSION AND CONCLUSION variation of log[n(z j )] with z j . However, these plots have
no direct physical interpretation. Thus, the quasar number
Plots (Figure 2) of numbers proportional to log[n(z j )] (listed density relative to their observed redshifts peaks around red-
in Tables 1 and 2) against z j show the following broad shifts somewhat greater than 1. The values z j refer to the
trends. For the smaller sample of 76, a straight line of falling limiting redshifts z m and not the observed ones z. The ob-
(i.e., negative) constant slope is a very good approximation. served redshifts z for quasars currently go upto about 10,
For the larger sample of 286, which starts at a significantly and in our sample upto a little above 3.7. The limiting red-
lower z value, log[n(z j )] falls more steeply than the smaller shifts z m , however, reach to much larger values as they are
sample initially, and then the slope becomes shallower (less determined in a complex way from other factors (see sections
negative) than the smaller sample. 1 and 2 above).
Figure 2 is presented to bring out the trends in the The cosmological number density n(z) is interpreted as

MNRAS 000, 14 (2016)

4 D. G. Banhatti
5 Croom, S. M. et al, 2009, MNRaS 399 1755
Sample of 76 Dezalay, J. -P. et al, 1994, AA 286 103
Sample of 286
Drinkwater, M. J. et al, 1997, MNRaS 284 85
4 Kulkarni, V. K. & Banhatti, D. G., 1983, ApJ 274 469
Lynden-Bell, D., 1971, MNRaS 155 95
log n(zj)

Rajarevathi, M., 2007, MPhil Thesis Madurai Kamaraj University

3
Schmidt, M. et al, 1988, ApJ 329 L85-7
Sokolov, K. P., 1988, SvAL 14 84
von Hoerner, S., 1974 Cosmology : Chap 13 in Kellermann, K I &
2
Verschuur, G L (eds) 1974 Springer, Galactic & Extragalactic
Radio Astronomy pp353-92
Wills, D. & Lynds, R., 1978 ApJSuppl 36 317-58
1 300*
0 0.5 1 1.5 2 2.5 3 3.5 4
zj
This paper has been typeset from a TEX/LATEX file prepared by
the author.
Figure 2. Plot of l og[n(z j )] vs z j for two samples. The point
at z j = 300 is only indicative and represents the z m bin with
z m > 4.0. It is indicated at 300, which is the geometric mean of
the smallest and the largest z m values in the bin.

the redshift distribution n(z; S0 ), which is the integral of the

radio luminosity function (P, z) over all luminosities present
in the sample as determined by the flux density limit. Thus,
Z
n(z; S0 ) = (P, z)dP. (4)
S0 ` 2 (, z)

This interpretation of n(z) [derived from the differential dis-

tribution p(x) p(V/Vm )] as the redshift distribution n(z; S0 )
needs to be explored further and utilized in deriving the cos-
mological evolution of the source population (here quasars).
For a suffciently large sample, different values of S0 may be
used to directly calculate n(z) as outlined, and the differ-
ential n(z) relative to S0 should then give the luminosity
function (P, z). This can be formally seen by differentiating
equation (4) with respect to S0 . This is evident when the
lower limit of the integral in equation (4) is identified with
P0 , the limiting luminosity corresponding to the limiting flux
density S0 .
The advantage of deriving n(z) first and then (P, z)
is that a much larger redshift than the largest observed
one is reached. This is essentially because more informa-
tion from the data (namely, the limiting redshifts) is used
in this method. Of course, much larger samples are needed
for reliable results from this method. Such large samples are
now available, and need to be exploited by using the method
described in this paper.

ACKNOWLEDGEMENTS
The work reported was done at School of Physics, Madu-
rai Kamaraj University, Madurai, India. I thank University
Grants Commission, New Delhi for support. I also thank
Vasant Kulkarni for introducing me to V/Vm test and lumi-
nosity functions.

REFERENCES
Banhatti, D., G., 2009, arXiv 0902.1139, 0903.1903, 0903.2442,
0903.2549. The last and briefest version was presented at 27th
Meeting of Astronomical Society of India at Indian Institute
of Astrophysics, Bengaluru.

MNRAS 000, 14 (2016)