Much of the work in this thesis was done by his code, without which I would probably still be languishing in Robinson's basement. We present the results of observations of the cosmic microwave background (CMB) with the Cosmic Background Imager (CBI), a sensitive 13-element interferometer located high in the Chilean Andes.
Origin of the Microwave Background
The reason why the measurement of CMB anisotropies is of such importance is because the angular power spectrum of the anisotropies contains a wealth of detailed information about the properties and evolutionary history of the universe. Once the earliest spectrum is set (such as during inflation), the evolution of the fluctuations does not depend on exotic and uncertain physics.
Power Spectrum Basics
At Φℓm, ℓ more or less corresponds to the wavelength of the mode and is similar to its directivity. Under these assumptions, all the information contained in the CMB is contained in the set of coefficients Cℓ, so that.
Cosmological Effects on the Power Spectrum
The power at small scales is also reduced due to the finite thickness of the final scattering surface. The curvature of the universe does not affect the physical structure at the surface of the last distribution, since the universe was very matter + radiation dominated then.
Microwave Background Observations
DASI (Kovac et al., 2002) first measured the polarization power spectrum, although the spectrum is too noisy to have any cosmologically useful information. Shortly thereafter, WMAP measured the cross-correlation spectrum of the large-scale polarization and total intensity anisotropies (Kogut et al., 2003).
Interferometers
For the case of a single dish, there will be a single-aperture diffraction pattern on the sky, which is the Fourier transform of the collecting aperture. The power pattern in the sky is called the primary beam and is the Fourier transform of the square of the dish's electric field response.
The Cosmic Background Imager
As a result, we have very dense UV coverage (see Figure 3.1 for a sample CBI UV coverage). The first two years were devoted to measurements of the total strength of the power spectrum, and in the fall of 2002 the CBI switched to predominantly polarization observations.
Uncorrelated Likelihood
In log-likelihood space, the joint log-likelihood is the sum of the individual probabilities. Note that the definition is χ2i x2i/V, so the maximum of the probability is the point at which the mean value of χ2 equals one.
Correlated Power Spectrum
This is the standard expression for the probability of a theory under a particular data set that starts most microwave background analysis papers. Thei, the jth element of the product of two matrices is the multiplication of the first times the jth column of the second.
Likelihood Gradient
Recall the formula for the derivative of the probability of uncorrelated data under these assumptions, Equation 2.8. The trace is the sum of the diagonal elements of the matrix and has a nice.
Likelihood Curvature
An early suggestion in Bond et al. (e.g. 1998) was to note that at the maximum probability the first term in (2.35) is approximately zero, and so we can approximate the curvature matrix by. With reasonably high accuracy, the error opqB for most experiments is simply that of the Gaussian approximation of the probability surface, FBB−1 (see, for example, Press et al., 1992).
Band Power Window Functions
If we are at the maximum, then the ∆∆T −C part of the gradient is zero, and we are left with the expected gradient as a result of the new signal. For example, we can calculate the expected contribution to the power spectrum from a population of weak radio signals. point sources that are statistically isotropic.
Early Observations
The ring configuration also provided a fairly uniform distribution of baseline lengths in the UV plane. As the CBI rotates the deck, a uniform distribution along the length also leads to fairly uniform sampling in the UV plane.
Ground Spillover
Since the observing pairs are observing the ground in identical ways, the ground signal should be identical in. Although critical to rejecting the ground signal, the cost of the difference is a factor of two in time.
Analysis
Interferometer Response to a Random Temperature Field
A is the square of a receiver's response to the electric field (the primary beam), and the axis position on the sky relative to the pointing center. Therefore, we need to understand the response of the interferometer to a plane wave on the sky, which is most conveniently done by taking the Fourier transform of (3.5).
Visibility Window Functions
The scaling of the primary beam in the coefficient is a bit unusual at first because we expect the total variance to be proportional to the total area of the beam, namely σp−2. We are now in a position to choose a parameterization of the power spectrum, which specifies S(w).
Complex Visibilities
This is a very reasonable expression - basically, a two-element interferometer is sensitive to modes in the sky that have the same wavelength as the separation of the elements, and the sensitivity from this peak drops off as the primary beam Fourier transform. This is a set of four relations, since each of the two equations must hold for both its real and imaginary parts.
Power Spectrum
The set of relations can be solved for the covariances between the real and imaginary parts of the visibilities as follows: However, if both Ci∗j and Cij are nonzero, then the symmetry is broken and the real and imaginary parts of the visibilities are no longer statistically equivalent, and therefore should be treated separately.
Interpretation and Importance of Spectrum
My main contribution to the first year papers was extracting the power spectrum from the mosaics using CBIGRIDR/MLIKELY. Finally, in Section 4.6, I describe the final power spectrum of the first-year mosaics and cosmological results from the spectrum.
Noise Statistics
Fast Fourier Transform Integrals
For the first term, we can convolve all wi fori > 1 to get a new variable, sayq. The values plotted are x, where the correction factor applied to the variance is of the form 1 + d.o.f.x, where the d.o.f.
Noise Correction Using Monte Carlo
The reason is that individual UV points are not independent, but rather are correlated due to the primary beam. As such, maximum likelihood combines, with weights, several different UV points to create independent estimators of the CMB.
GRIDR/MLIKELY Speedups
In this case we regularized to the value from the unregularized spectrum, so the only effect is the small error bar. The acceleration of the hybrid mesh occurs both in CBIGRIDR, as each visibility is gridded onto fewer estimators, and in the linear algebra part of the pipeline, MLIKELY.
Source Effects in CBI Data
Source Effects on Low-ℓ-Spectrum
This leaves significant uncertainties in the residual flux from point sources that are difficult to estimate (since the statistics of weak sources at 30 GHz are poorly known), which can add a significant amount of energy. However, the question remains as to what the maximum likelihood is actually doing when it projects the sources and what the expected effects are in the spectrum.
Two Visibility Experiment
In this simple case, it would also be correct to think of maximum likelihood using a long baseline to measure the current from the source and subtract it. This works because the long baseline is only sensitive to the current from the source, which is also true for a pure source brightness measurement.
Sources in a Single Field
In the general case, however, there is no such pure measurement, and thus there is no estimate of the source flux to subtract. The short baseline is sensitive to both the CMB and a foreground point source, while the long baseline is only sensitive to the source.
Source Effects in the First-Year Mosaics
The total signal available is just the sum of the eigenvalues in the window matrix after a matrix transformation that transfers the noise matrix to the identity matrix. So the source will not really be projected from the mosaic spectrum even though it has disappeared from the deep spectrum.
First-Year Data
First-Year Results
Power Spectrum
The band power window functions, which describe the sensitivity of the CBI bands to the CMB power at a givenℓ, are shown in Figure 4.12. Also note how much the CBI expands the range over which the CMB power spectrum is measured.
Cosmology with the CBI Spectrum
-h - measurement of the Hubble cash from the HST key project of 72±8, as found in Freedman et al. The CBI is not very sensitive to the spectrum below ℓ∼400, so these cosmological results are basically independent of the first acoustic peak.
Compression
To calculate the compression matrix, we need to diagonalize the blocks along the diagonal of the covariance matrix. Another useful feature of compression is that it (usually) only needs to be done once.
Mosaic Window Functions
General Mosaic Window Functions
Because only the data itself changes between different realizations and not the statistical properties, the only compression required is again that of the data vector.
Gaussian Beam
The terms involving A and B do not have any θ dependence, so they can be drawn from the integral. So instead of making separate calls to different Bessel functions, we can add the sum of the products of the Bessel functions and normalize them at the end, making the whole integral only marginally more work than two calls to Bessel routines from high order.
Comparisons with Other Methods
BOOMERANG (described in Hivon et al., 2002), as they require a fast spherical harmonic transformation of the data to calculate the window matrices. See Table 5.2 for a summary of the comparison statistics, and Figures 5.8 and 5.9 for CBISPEC and CBIGRIDR fits to the first (high signal) and last (high noise) bins, respectively.
Foreground with CBISPEC
Measuring the Spectral Index
The red dots are the expected variances for the same Cℓ spectrum, with an applied frequency spectral index, forα= 0. The addition of the 125 cm baseline has broken the degeneracy between the flat,ν0 spectrum and the ℓ−6.4, Planck spectrum.
The Spectral Index Measured by CBI
See Figure 5.12 for the histogram of the best fit values for α for the individual simulations. The blue crosses are the four contaminated spots at the northern end of the 02 hour mosaic.
Future Improvements
The CBI is a highly sensitive interferometer operating at 30 GHz optimized for observations of the CMB in the multipole range 500 < ℓ <3500. Initially, I helped build the CBI, including mounting and testing the CBI receivers.
Statistical Basics
Variance of a Product
We need to know the variance of the product of two independent random variables (not necessarily identically distributed). It is also worth noting explicitly that if the expected values of the variables are zero, the variance of the product is the product of the variances: Var(xy) = Var(x)Var(y).
Expectation of f(x)
We can break the expectation into different terms as the sum expectation is the sum of the expectations. If we set the reference valuex0 to be the expectation ex, then the second term goes to zero, since hx−x0i=hxi −x0= 0 ifx0=hxi.
Some Relevant Distributions
The general expectation relationship is established by integrating parts and comparing the resulting integral with the expectation value of the underlying order. It is the distribution of the ratio of two empirically determined variance estimates when they come from samples with the same intrinsic variance.
Combining Two Identical Data Points
However, if the points are widely spaced, they must actually give a better estimate of the mean. So if we want the expected variance of V to be the same as the expectation of our estimate, we need to scale by a factor.
Combining Many Identical Data Points
As cluster size decreases, longer and longer baselines are expected to persist. Essentially, exploring a larger region allows us to better characterize the behavior of the CMB below the cluster.