MODELS FOR SOURCE-COUNTS AT SUBMM TO OPTICAL WAVELENGTHS from Rowan-Robinson M. (2001, ApJ 549, 745) The models are described in the paper sfr.ps (this is in MN format because this is easier to read) The names of the count models are of the form: pred(nnn)cnts(mmmm)....dat where nnn is the wavelength in microns (044 means 0.44 mu, ie B, 2 means 2.2 mu, ie K, 36 means 3.6 mu, 45 means 4.5 mu, 58 means 5.8 mu, 7 means 6.7 mu, 8 means 8.0 mu, 24 means 24 mu etc) mmmm = 1254, means the best Omega=1, Lambda = 0, model, with P=1.2, Q=5.4 mmmm = 032173, means the best Omega=0.3, Lambda = 0, model, with P=2.1, Q=7.3 mmmm = lam3090, means the best Omega=0.3, Lambda=0.7, model, with P=3.0, Q=9.0. The header in each case gives a table of t_i as a function of logL60 (first 7 lines) + the seds of the four source types (cirrus, starburst, A220, AGN dust torus) [for lambda < 20 mu, there is an additional sed type, corresponding to low-mass stars] as an array of 46 numbers giving log{nu Lnu} for log lambda(microns) = 3.1 (0.1) -1.4, smoothed to resolution of 0.2 dex (boxcar), normalized to 60 microns (next 20 lines). The output consists of 9 lines (11 if lambda<20mu) for each flux-density: l.1 gives P, Q, *, *, lambda, *, *, log S(Jy), log N(S), f1-f4 (or f5 if lambda<20) (ignore columns labelled * above) l.2-3 gives redshift distribution for cirrus component in z bins 0-0.2, 0.2-0.4, 0.4-0.6, 0.6-0.8, 0.8-1.0, 1-1.5, 1.5-2, 2-2.5, 2.5-3, 3-3.5, 3.5-4, 4-4.5, 4.5-5, >5 l.4-5 gives redshift distribution for starburst component l.6-7 gives redshift distribution for Arp220-sb component l.8-9 gives redshift distribution for AGN dust torus component l.10-11 (if lambda<20) gives redshift distribution for old (low-mass) star component fi gives the fractional contribution to the counts from each of the 4 (5) components N(S) means the integral counts. Differential counts can be obtained by differencing successive steps in flux-density. The names of the background models are of the form predbgspectt(mmmm)iold.dat where mmmm is as above. The header (first 37 lines) is as above. Lines 38 onwards give i, log lambda(mu), (ignore), log nu I(nu) in nW/m2/sr, next 5 columns give contributions to the background of cirrus+high-mass stars, M82 starburst, A220 starburst, AGN, low-mass stars. ERRATA There are some typos in Appendix A in the printed version of the paper. The correct version of the Appendix is given in sfr.ps The formulae used in the calculations described in the paper were correct. The typos only appeared while preparing the paper for submission. NEED TO REVISE MODEL AT 15 MICRONS ? (5/4/02) Although the model gives an excellent fit to the 15 mu data published at the time of submission, subsequent reanalysis of the ELAIS data (Lari et al, 2000, MN 325, 1173, Gruppioni et al, 2002, MN subm.) and comparison of ELAIS fluxes for stars with K-band data (Vaisanen et al, 2002, MN subm.) suggest that the 15 mu calibration used by Serjeant et al (2000) may have given fluxes too bright by a factor of 1.5-2 (see also Genzel and Cesarsky 2000). I have therefore investigated whether any simple modification to the model could accomodate these revised 15 mu counts. It also seems apprpriate to compare the predictions of the model with differential count data now available at 15 mu, and with the 12 mu counts of Clements et al (2000, AA, astro-ph/9901267). The predictions of the model at 15mu are driven by the assumed component seds. and the assumed proportions of the different components as a function of 60 mu luminosity. These assumptions were checked against the local 12 mu luminosity functions for Seyferts and non-Seyferts given by Rush et al (1993). Modifications to the component seds in the midir would change the predictions in this range without affecting the predictions in the optical, near-ir, far-ir and submm. On examination of the predicted 12 mu luminosity functions for the different components (Fig 7 of Rowan-Robinson 2001) it became clear that the luminosity function of the cirrus component could be shifted significantly (0.1-0.3 dex) without violating the fit to the Rush et al data. I have therefore explored the effect of changing the cirrus component sed so that the dust emission spectrum is reduced by 0.2 dex for lambda < 20 mu. Physically this corresponds to reducing the surface brightness of the illuminating radiation field. The results of this are shown for the lambda = 0.7 cosmology in Fig7rev.ps (12 mu luminosity functions), Fig17rev.ps (15 mu integral counts), newfig26.ps (15 mu differential counts) and newfig27.ps (12 mu integral counts). The corresponding models are pred15cntslam3090tt12ifoldrev2.dat and pred12cntslam3090tt12ifoldrev2.dat Discussion: If the contributions of cirrus and starburst components are added, the predicted 12 mu luminosity function for non-Seyferts will agree well with the Rush et al data. In Fig17rev.ps, the integral counts given by Gruppioni et al (2002) have been added as open circles. The Serjeant et al ELAIS data (hatched region) has been shifted by a factor of 1.5 towards lower fluxes. The revised model (solid line) is a reasonable compromise between the IRAS data of Verma (2000), for which a conversion S(15)=1.1 S(12) was assumed, and the ELAIS data. Models can no doubt be devised which fit all the data perfectly, but they will have to be very different in philosophy from my model. There may still be issues of calibration to be sorted out between IRAS and ISO. The original model given in ApJ 549, 745 is shown asa broken line in this and following figures. In newfig26.ps, the differential counts of Verma (2000, her Fig 2.11) are shown as filled circles, the differential counts of Gruppioni et al (2002) are shown as open circles, and the differential counts summarised by Elbaz et al (1999) are shown as filled squares (HDF-N and HDF-S) and crosses (Marano, Lockman and A2390 surveys). Again the revised model is a reasonable compromise. The 12 mu differential counts by Verma (2000) appear to be the most careful compilation to date from the IRAS FSS. They are significantly lower (by a factor ~ 1.4) than the Rush et al counts (mainly because the latter used ADDSCAN fluxes). The Verma counts can be summarised numerically as: log S(12), Jy log{-dN/dS x S^{2.5}} (Jy^{1.5}/sr) 0.70 1.30 +- 0.20 0.48 1.00 +- 0.22 0.30 1.12 +- 0.13 0.08 1.14 +- 0.09 -0.12 1.20 +- 0.07 -0.30 1.14 +- 0.05 -0.52 1.19 +- 0.03 -0.70 1.27 -0.92 1.32 In newfig27.ps, the integral counts of Verma (2000) are shown as filled circles and those of Clements et al (2000) as open circles. The ISO data of Clements et al again prefer the revised model, whereas the IRAS data prefer the original model. Changing the rate of evolution would worsen the fit of the model at other wavelengths. Some other figures in RR(2001) would also need modification, eg Figs 5, 8, 24, but the changes are not very significant. This revised model will be discussed in more detail in a subsequent paper. A paper by King and Rowan-Robinson (2002, MN submitted) will discuss refinement of the model at optical and near infrared wavelengths. The four templates used are given in iretemplates2.dat The file consists of 147 values of log10(lambda(microns)) 147 values of log nu S(nu) for the cirrus template 147 values of log nu S(nu) for the M82 template 147 values of lg nu S(nu) for the A220 template 46 values of lg nu S(nu) for AGN dust torus, at lg lambda = 3.1 (0.1) -1.4 The seds have been smoothed in the wavelength range 3-30 mu with a boxcar filter of width 0.12 dex (to correspond approximately with ISO and SPITZER bandpasses) [This page last revised on 5/4/02]