Ultra-fast model emulation with PRISM; analyzing the Meraxes galaxy formation model. (arXiv:2011.14530v2 [astro-ph.IM] UPDATED)
<a href="http://arxiv.org/find/astro-ph/1/au:+Velden_E/0/1/0/all/0/1">Ellert van der Velden</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Duffy_A/0/1/0/all/0/1">Alan R. Duffy</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Croton_D/0/1/0/all/0/1">Darren Croton</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Mutch_S/0/1/0/all/0/1">Simon J. Mutch</a>
We demonstrate the potential of an emulator-based approach to analyzing
galaxy formation models in the domain where constraining data is limited. We
have applied the open-source Python package PRISM to the galaxy formation model
Meraxes. Meraxes is a semi-analytic model, purposefully built to study the
growth of galaxies during the Epoch of Reionization (EoR). Constraining such
models is however complicated by the scarcity of observational data in the EoR.
PRISM’s ability to rapidly construct accurate approximations of complex
scientific models using minimal data is therefore key to performing this
analysis well.
This paper provides an overview of our analysis of Meraxes using measurements
of galaxy stellar mass densities; luminosity functions; and color-magnitude
relations. We demonstrate the power of using PRISM instead of a full Bayesian
analysis when dealing with highly correlated model parameters and a scarce set
of observational data. Our results show that the various observational data
sets constrain Meraxes differently and do not necessarily agree with each
other, signifying the importance of using multiple observational data types
when constraining such models. Furthermore, we show that PRISM can detect when
model parameters are too correlated or cannot be constrained effectively. We
conclude that a mixture of different observational data types, even when they
are scarce or inaccurate, is a priority for understanding galaxy formation and
that emulation frameworks like PRISM can guide the selection of such data.
We demonstrate the potential of an emulator-based approach to analyzing
galaxy formation models in the domain where constraining data is limited. We
have applied the open-source Python package PRISM to the galaxy formation model
Meraxes. Meraxes is a semi-analytic model, purposefully built to study the
growth of galaxies during the Epoch of Reionization (EoR). Constraining such
models is however complicated by the scarcity of observational data in the EoR.
PRISM’s ability to rapidly construct accurate approximations of complex
scientific models using minimal data is therefore key to performing this
analysis well.
This paper provides an overview of our analysis of Meraxes using measurements
of galaxy stellar mass densities; luminosity functions; and color-magnitude
relations. We demonstrate the power of using PRISM instead of a full Bayesian
analysis when dealing with highly correlated model parameters and a scarce set
of observational data. Our results show that the various observational data
sets constrain Meraxes differently and do not necessarily agree with each
other, signifying the importance of using multiple observational data types
when constraining such models. Furthermore, we show that PRISM can detect when
model parameters are too correlated or cannot be constrained effectively. We
conclude that a mixture of different observational data types, even when they
are scarce or inaccurate, is a priority for understanding galaxy formation and
that emulation frameworks like PRISM can guide the selection of such data.
http://arxiv.org/icons/sfx.gif
