| JPG or GIF image (click for full resolution) | Other formats | Description/caption | Credit information |
 | B0218_Merlin.eps |
MERLIN radio contour map (Biggs et
al. 2001) of the quasar we studied, called B0218+367, which lies
about 7.5 billion light years away. The galaxy containing absorbing
ammonia molecules lies about 6 billion light years away and, though it
is not seen in this radio map, gravitationally lenses the background
quasar light to produce two bright quasar images on the sky (big red
circles). The alignment of the lensing (and absorbing) galaxy is such
that it also produces a so-called "Einstein ring" of quasar light (the
larger, less bright, circular shape). The molecules we study are only
observed along the line of sight towards the lower right quasar
image. The physical size of the image (at the distance of the
absorbing galaxy) is about 19,000 light years across. |
Andy Biggs |
 | B0218A_global.eps | Very high
resolution radio map of the only lensed quasar image towards which
molecular absorption occurs (Biggs et
al. 2003). For comparison with the much larger scale image above,
this image is "only" 600 light years across (again, at the distance of
the absorbing galaxy). In this image we see the actual structure of
the quasar's radio light emitting regions. We see a core (compact red
region) and a knotty structure extending away from it to the left -
this is a jet of radio emitting material being ejected from the quasar
core. It is thought that the molecular absorption only occurs along
the sight-line to the quasar core. | Andy Biggs |
 | None | The IRAM Plateau de Bure Interferometer | Sébastien Muller |
 | None | The IRAM Plateau de Bure Interferometer | Sébastien Muller |
 | None | The Effelsberg 100-m Radio Telescope | Norbert Junkes |
 | None | The Effelsberg 100-m Radio Telescope | Norbert Junkes |
 | None | The Effelsberg 100-m Radio Telescope | Max-Planck-Institut für Radioastronomie |
 | None | Observing molecular absorption in distant galaxies using the light of bright, background quasars. As light from the quasar travels to Earth, the Universe continues to expand, stretching the light's wavelength (it gets redder the longer it travels). In our observations, the light is also gravitationally lensed (its path is bent) as it passes through an intervening galaxy; when a radio map of the field is made, two quasar images appear. However, the molecular absorption clouds are only along the line of sight to one image. Furthermore, when very high resolution images are made of that quasar image, some structure is evident - a core (the brightest part of the image) and a knotty jet extending away from the quasar core. It's only towards the quasar core that molecular absorption is thought to occur. | As on image
Telescope: S. Muller
Radio insets: A. Biggs
Intervening galaxy: NASA, ESA, STScI & W. Keel.
Quasar: NASA, ESA, STScI & E. Beckwith. |
 | None | As above | On image |
 | None | As above | Telescope: S. Muller
Radio insets: A. Biggs
Intervening galaxy: NASA, ESA, STScI & W. Keel.
Quasar: NASA, ESA, STScI & E. Beckwith. |
 | None | As above | On image |
 | None | As above | Telescope: S. Muller
Radio insets: A. Biggs
Intervening galaxy: NASA, ESA, STScI & W. Keel.
Quasar: NASA, ESA, STScI & E. Beckwith. |
 | None | As above | Telescope: S. Muller
Radio insets: A. Biggs
Intervening galaxy: NASA, ESA, STScI & W. Keel.
Quasar: NASA, ESA, STScI & E. Beckwith. |
 | None | As above | On image |
 | None | As above | Telescope: N. Junkes
Radio insets: A. Biggs
Intervening galaxy: NASA, ESA, STScI & W. Keel.
Quasar: NASA, ESA, STScI & E. Beckwith. |
 | None | As above | On image |
 | None | As above | Telescope: N. Junkes
Radio insets: A. Biggs
Intervening galaxy: NASA, ESA, STScI & W. Keel.
Quasar: NASA, ESA, STScI & E. Beckwith. |
 | None | As above | Telescope: N. Junkes
Radio insets: A. Biggs
Intervening galaxy: NASA, ESA, STScI & W. Keel.
Quasar: NASA, ESA, STScI & E. Beckwith. |
 | None | As above | On image |
| JPG or GIF image (click for full resolution) | Other formats | Description/caption |
 | fit08_all_vmu.eps |
Figure 1: Spectra of the molecular transitions used in this study registered to
a heliocentric velocity scale centered on z=0.68466. The
nominal observed frequencies are noted in each panel. The data,
normalized by fits to their continua, are plotted as black
histograms. Tick-marks above the spectra show the positions of
velocity components in our fiducial 8-component fit (solid line
following the data). Note that the HCN and NH3 transitions
have complex hyperfine structure reflected in each velocity component;
the tick marks show the position of the strongest hyperfine component
in LTE. Residuals between the fit and data, normalized by the
(constant) error array, are plotted above the spectra, bracketed by
horizontal lines representing the +/- 1σ level. The fit contains
57 free parameters: an optical depth for each component in each
transition (5x8 parameters) plus a Doppler width and redshift for each
component (8+8 parameters) and a single value of
Δμ/μ. The fitted line parameters are tabulated in the
Supporting Online Material.
|
 | da_vs_Ncomp.eps |
Figure 2: Variation in Δμ/μ and χ2 per degree of
freedom, χ2ν, of different velocity
structures characterized by the number of fitted absorption
components. χ2ν is defined as
χ2/ν=ΣNdj[dj-m(j)]/σ2j
for dj the jth data value with variance
σ2j and model value m(j). The
sum is over all Nd=223 data points;
ν=Nd-Npar for
Npar free model parameters. Our fiducial 8-component
(Npar=57) result is highlighted with square points.
Different components were added/removed to/from the fiducial fit to
form each initial velocity structure and VPFIT was run again to
minimize χ2 by varying all free parameters. Two
different initial fits with 6 components and three fits with 7
components were possible; the different results are offset in the plot
for clarity in these cases. Large χ2ν
values for <=6 components indicate that those fits are not
statistically acceptable. Of the remaining fits, the 8-component fit
has the lowest χ2ν. Note that the
9-component fit has a smaller χ2 (because more
parameters are being fitted) but a marginally higher
χ2ν, indicating that it is less
statistically preferred than the 8-component fit. Only statistical
error bars on Δμ/μ are shown; see text for discussion
about systematic errors.
|
 | chisq_vs_da.eps |
Figure S1: χ2 curve for the fiducial 8-component model
fit. Note its (required) smoothness and (expected) near-parabolic
shape. The right-hand vertical scale shows
χ2ν which takes a minimum value of
χ2ν,min=1.07125. The left-hand
vertical scale shows
Δχ2/χ2ν,min,
thereby allowing the 1-σ error in Δμ/μ to be
immediately read off the graph (dotted black lines). In practice, we
measured the position at which χ2 takes its minimum
and the curve's width by fitting a parabola to the central points; the
fit, marked by the black solid line, demonstrates how close to
parabolic that part of the curve is. The black circle and error-bar
represent the result from this fit, Δμ/μ=(+0.75 +/-
0.45)x10-6, which closely matches our fiducial value
represented by the red/grey square and error-bar.
|
 | fit04-3478_all_vmu.eps
fit05-348_all_vmu.eps
fit06-34_all_vmu.eps
fit06-48_all_vmu.eps
fit07-3_all_vmu.eps
fit07-4_all_vmu.eps
fit07-8_all_vmu.eps
fit09+9_all_vmu.eps
|
Figure S2: The different velocity structures attempted in search of the
statistically preferred one. The values of Δμ/μ and
χ2ν corresponding to each fit are
represented by the black points in Fig. 2. Components were
removed/added from/to the fiducial fit to form each initial velocity
structure and VPFIT was run again to minimize χ2. For
the 9-component model, the additional component was added redwards of
all other components. The residual spectra show how poor the fits are
with <=6 components.
|
 | compfit_rot.eps |
Figure S3: Independent fits to the two rotational transitions studied
here. Layout similar to Fig. 1. The data, tick-marks and residuals for
HCN(1-2), marked in lighter grey lines, are offset below those for
HCO+ for clarity. The formal 1-σ statistical error in
the redshift of each component (converted to velocity) is represented
by the horizontal bars across the tick marks. The error bar for each
component is plotted higher than the one to its left for
clarity. There is broad agreement between the positions of the
velocity components. The bluest component in HCN(1-2) is not
statistically required by the HCO+(1-2) data alone, similar
to the simultaneous fit in Fig. 1.
|
