Grids¶

A grid is the emission of a stellar population synthesis (SPS) or AGN model, worked out over a range of parameters: ages and metallicities for a stellar population, or black hole masses and accretion rates for an AGN, for example. Synthesizer interpolates over these to give galaxies their spectra.

Grids come in two kinds:

Incident: The model’s own emission, as its authors provide it.
Photoionised: The same emission passed through a photoionisation code, usually Cloudy, which adds the nebular continuum and emission lines from the surrounding gas.

Grids are made with syncretize, and Synthesizer’s grids documentation explains how to use them.

What’s in the file¶

A grid is an HDF5 file. If you made it with syncretize, it already has everything below. At minimum it needs its axes, and at least one of spectra, lines or log10_specific_ionising_luminosity.

Location	Contents
`axes` attribute	The names of the axes, in order.
`axes/<name>`	The values along each axis, with their `Units`. An axis marked `log_on_read` is read in log space.
`spectra/`	One dataset per kind of spectrum (`incident`, `nebular`, …), and the `wavelength` they share.
`lines/`	Line luminosities, with each line’s `id` and `wavelength`.
`log10_specific_ionising_luminosity`	Ionising photon luminosities.
`Model/`	Which model this is: `sps_name` and `sps_version` for an SPS model, or `type = "agn"` and `family` for an AGN model.
`CloudyParams/`	The photoionisation settings, including `cloudy_version`. Photoionised grids only.
Root attributes	`synthesizer_version`, `synthesizer_grids_version` and `date_created`, recording how and when the grid was made.

Naming¶

A grid’s name lists what makes it distinct, from the model down to the processing:

bpass-2p2p1-bin-chabrier03-0p1-300p0-cloudy-c23p01
│     │     │   │          │         └ photoionisation code and version
│     │     │   │          └ IMF mass range (0.1 to 300 solar masses)
│     │     │   └ IMF (Chabrier 2003)
│     │     └ variant (binary stars)
│     └ version (2.2.1)
└ model (BPASS)

An incident grid stops before the photoionisation code. A photoionised grid takes its incident grid’s name and adds the code, and anything that differs from the standard photoionisation setup goes on the end, as in …-cloudy-c23p01-resolution0p05.

Checks¶

syndex-check reads a grid’s kind from what is inside it, not from its name, with one exception.

Warning

A filename containing dust is always treated as a dust grid. Keep dust out of the name of any other grid.

Errors¶

These stop a grid being published.

Problem	Fix
The grid does not say what kind of grid it is.	Give the `Model` group `sps_name` for an SPS model, or `type = "agn"` for an AGN model.
The grid does not say what emission it holds.	Include `CloudyParams`, or reprocessed spectra (named `nebular…`, `transmitted…` or `linecont…`). A grid with only `incident` spectra is read as incident.
The grid has no axes.	Add the `axes` attribute and the `axes/` group.

Warnings¶

These are reported, but do not stop a grid being published. They are still worth fixing.

Problem	Fix
An axis name is singular.	Use the plural: `ages`, `metallicities`, `ionisation_parameters`, `hydrogen_densities`, and so on.
An axis has unexpected units.	`ages` should be a time; `metallicities`, `ionisation_parameters`, `accretion_rates_eddington` and `cosine_inclinations` should be dimensionless.
The filename and the file give different Cloudy versions.	A file named `…cloudy-c25.00…` should hold `cloudy_version = c25.00`. Correct whichever is wrong.
No model name, wavelength or `synthesizer_version` could be read.	Add what is missing. Without a model name, the grid will not appear under its model in the portal’s filters.