Last update: 29-09-2021

arXiv arXiv Bitbucket version GPLv3 license

Simbelmynë"How fair are the bright eyes in the grass! Evermind they are called, simbelmynë in this land of Men, for they blossom in all the seasons of the year, and grow where dead men rest."
Gandalf, The Two Towers III 6, The King of the Golden Hall

Simbelmynë [SEEM-bale-mü-nay] is a hierarchical probabilistic simulator to generate synthetic galaxy survey data. The code is publicly available on Bitbucket, and the documentation is available on Read the Docs.

Simbelmynë is written in C with Open-MP parallelisation, and includes various additional tools written in python 3. The current version (v0.x.x) is a beta-release; the code will be further developed and documented in the future.

Limited user-support may be asked by email.


Simbelmynë does not have its scientific paper yet. For the time being, if you want to acknowledge use of this software, please cite Leclercq et al. (2015b), where it was first used:

From version 0.4.0, Simbelmynë includes an implementation of the perfectly parallel algorithm for cosmological simulations described in Leclercq et al. (2020). If using this feature, please also cite:

Key features

Simbelmynë is a parallel code, which, due to its fast nature, can be used to generate a large number of realisations of synthetic galaxy survey data. As the intended use is within cosmological analyses of real survey data, Simbelmynë has been developed as a self-contained, end-to-end generative process. The focus has been placed on speed and accuracy at the relevant scales.

Current key features of the simulator are:

Simbelmynë exampleGeneration of synthetic galaxy observations with Simbelmynë. The code first generates a dark matter density field (left panel) using a non-linear gravitational model. The galaxy density field (middle panel) takes into account galaxy bias and redshift-space distortions. From this, observations are simulated (right panel) using the survey geometry and a prescription for selection effects.

Bayesian hierarchical modelling

Simbelmynë exampleA generative forward model for galaxy survey data, implemented in Simbelmynë.

As an end-to-end generative process for galaxy survey data, the algorithm implemented in Simbelmynë can be conveniently formulated in the form of a Bayesian hierarchical model, represented diagrammatically in the figure.

The set of model parameters can be defined as follows: cosmological parameters θ, a set of parameters Φ characterising the distributions of intrinsic galaxy properties, and a set of parameters Ψ characterising the survey and observational effects. The latent variables are the initial density contrast δi, the final density contrast δf, the population of galaxies Ng, galaxy bias parameters b, and survey parameters s. The data d are the observed galaxies. The model involves three deterministic blocks: a simulator of dark matter structure formation F, a simulator of galaxy biasing G, and a virtual observatory S.

As a simulator implementing this generative forward model, the Simbelmynë code is particularly suitable for use with likelihood-free inference methods.

Future versions

Future public releases of the code will include additional features that are currently being tested. These include:

  • an extended module for deterministic and stochastic galaxy bias,
  • an extended module for survey systematic effects.

Origin of the name

Simbelmynë is an Old English (c. 650 to 1066) word derived from simbel, meaning "always, continually", and mynë, meaning "memory, remembrance, memorial". It can be translated as "Evermind". In mynë, the letter y is used to transliterate and transcribe the rune ᚣ (ȳr), and it is pronounced /y/ (as in the French tu). The ë (e-diaeresis) is used to indicate that the final e is not mute. It is to be pronounced /ε/ (as in the Modern English bet).

In J. R. R. Tolkien's legendarium, Simbelmynë is used to refer to an imaginary small white flower (a variety of anemone). The name is fitting in more than the usual sense (see Robinson 2013). For the interested reader, more information is available here, here and here.