Abstract:
We explore the possibility of constraining model parameters of the Epoch of Reionization (EoR) from 21cm brightness temperature maps, using a combination of morphological
descriptors constructed from the eigenvalues of the Contour Minkowski Tensor (CMT), Betti
numbers (count of connected regions ncon and holes nhole) and the area of structures in
the excursion set of the field. We use a three parameter model of EoR simulated using
21cmFAST, namely the ionizing efficiency of sources ζ, the minimum virial temperature Tvir
required for collapse into a halo and the maximum radius for ionizing radiation described
by Rmf p. We performed a Bayesian analysis to recover model parameters for a mock 21cm
image from SKA phase I at a redshift of z = 7.4 corresponding to a mean neutral hydrogen
fraction of ¯xHI ' 0.5. We find that in the absence of noise the average size of structures
in the field with xHI . 0.5 is smaller than regions with xHI & 0.5 and the structures are
equally isotropic when ¯xHI = 0.5 . We also find that in order to recover the input model
to within 1 − σ accuracy for a mock noisy image at a single frequency channel of 1 MHz,
for an observation time tobs < 2000 hrs, the noisy δTb map needs to be smoothed at a scale
Rs > 9.5 Mpc. Finally we show that the systematic behaviour of the statistic as ionization
progresses, enables us to obtain stringent constraints on ¯xHI (with a coefficient of variation
∼ 0.05 as compared to ∼ 0.1–0.2 for model parameter constraints), thereby making these
descriptors a promising statistic for constraining EoR model parameters and the ionization
history of the universe.