Adam Amara, Thomas Kitching, Anais Rassat, Alexandre Refregier.
PyCosmo: Tests and Development · 2018-09-04 · PyCosmo: Tests and Development Federica Tarsitano...
1
PyCosmo: Tests and Development Federica Tarsitano Alexandre Refregier, Lukas Gamper, Adama Amara, Lavinia Heisenberg (1) @AstroFederica > Python-based framework providing cosmological model predictions > Symbolic calculation of the Einstein-Boltzmann equations - implemented with the sympy package - the symbolic representation provides the user a convenient interface to sp manipulate the equations and implement new cosmological models > Currently under test and development > Future work: CMB angular power spectrum, massive neutrinos, more sp s general models beyond flat ΛCDM… > More info at http://www.cosmology.ethz.ch/research/software-lab.html Overview Non-linear perturbations References: (1) Refregier et al., 2017, arXiv:1708.05177 - (2) Eisenstein & Hu, 1998, ApJ, 511, 5 - (3) Peacock, 1997, MNRAS, 284, 885 - (4) Smith et al., 2003, MNRAS, 341, 1311 - (5) Mead et al., 2015, MNRAS, 454 - (6) Takahashi et al., 2012, ApJ, 761, 152 - (7) Palmese et al., 2017, The Astrophysical Journal - (8) Tarsitano et al., 2018, MNRAS [email protected] Background Linear perturbations Mead + Baryons Other projects 0.0 0.5 1.0 1.5 2.0 2.5 3.0 redshift z 10 -8 10 -7 10 -6 10 -5 10 -4 Δχ(z)/χ(z) PyCosmo iCosmo CLASS > Varying cosmology, we s obtain a max relative s s s difference up to e-4, with s a dispersion up to e-6, in s comparison iCosmo, s s s and up to e-3 with s s s s dCLASS PyCosmo-iCosmo comparison Eisenstein & Hu (2) BBKS (3) Properties of GW170817 BNS in NGC 4993 (7) Galaxy Morphology catalogue DES Y1 (8) z=0 z=0 > PyCosmo can compute the non-linear power spectrum using two different s fitting functions: HaloFit and the Mead model (4)(5) > HaloFit is tested against iCosmo. The user can choose to include the spadditional corrections from Takahashi et al., 2012 (6) > The Mead model in PyCosmo is developed in python and cython. It is sptested against the original code in Fortran (HMcode). > HaloFit and Mead run at comparable speed. HaloFit Mead log log log log Exec. time: t=2.75 s Exec. time: t=2.50 s PyCosmo > In the Mead model the baryonic feedback (BF) is implemented by tuning two parameters of the model: A (min. halo conc.) and η0 (halo bloating parameter) > A and eta vary in four different cases accounting for BF: DM ONLY, REF, AGN and DBLIM > We test our python implementation against the HMcode in the four cases > We can reproduce the comparison, shown in [5], between the model+BF and the DM ONLY case. z=0.5
Transcript of PyCosmo: Tests and Development · 2018-09-04 · PyCosmo: Tests and Development Federica Tarsitano...
PyCosmo: Tests and DevelopmentFederica TarsitanoAlexandre Refregier, Lukas Gamper, Adama Amara, Lavinia Heisenberg(1)
@AstroFederica
> Python-based framework providing cosmological model predictions > Symbolic calculation of the Einstein-Boltzmann equations - implemented with the sympy package - the symbolic representation provides the user a convenient interface to sp manipulate the equations and implement new cosmological models > Currently under test and development > Future work: CMB angular power spectrum, massive neutrinos, more sp s general models beyond flat ΛCDM… > More info at http://www.cosmology.ethz.ch/research/software-lab.html
Overview
Non-linear perturbations
References: (1) Refregier et al., 2017, arXiv:1708.05177 - (2) Eisenstein & Hu, 1998, ApJ, 511, 5 - (3) Peacock, 1997, MNRAS, 284, 885 - (4) Smith et al., 2003, MNRAS, 341, 1311 - (5) Mead et al., 2015, MNRAS, 454 - (6) Takahashi et al., 2012, ApJ, 761, 152 - (7) Palmese et al., 2017, The Astrophysical Journal - (8) Tarsitano et al., 2018, MNRAS
Background
Linear perturbations
Mead + Baryons Other projects
0.0 0.5 1.0 1.5 2.0 2.5 3.0redshift z
10�8
10�7
10�6
10�5
10�4
��
(z)/
�(z
) PyCosmo
iCosmo
CLASS
> Varying cosmology, we s obtain a max relative s s s difference up to e-4, with s a dispersion up to e-6, in s comparison iCosmo, s s s and up to e-3 with s s s s dCLASS
PyCosmo-iCosmo comparison
Eisenstein & Hu(2)
BBKS(3)
Properties of GW170817 BNS in NGC 4993(7) Galaxy Morphology catalogue DES Y1(8)
z=0
z=0
> PyCosmo can compute the non-linear power spectrum using two different s fitting functions: HaloFit and the Mead model(4)(5) > HaloFit is tested against iCosmo. The user can choose to include the spadditional corrections from Takahashi et al., 2012(6) > The Mead model in PyCosmo is developed in python and cython. It is sptested against the original code in Fortran (HMcode). > HaloFit and Mead run at comparable speed.
HaloFit
Mead
log log
loglog
Exec. time: t=2.75 s
Exec. time: t=2.50 s
PyCosmo
> In the Mead model the baryonic feedback (BF) is implemented by tuning two parameters of the model: A (min. halo conc.) and η0 (halo bloating parameter) > A and eta vary in four different cases accounting for BF: DM ONLY, REF, AGN and DBLIM > We tes t ou r py thon implementation against the HMcode in the four cases > We can reproduce the comparison, shown in [5], between the model+BF and the DM ONLY case.
z=0.5
