The BEAST uses classes to abstract the specific implementation from the details of storage in memory and on disk.
The two main classes to use to access information in spectral and SED grids are:
SEDGrid: Provides the interface to grids of extinguished SEDs composed of band integrated fluxes. Used fairly extensively in the BEAST to interface with the core physicsmodel grid.
SpectralGrid: Provides the interface to grids of spectra. Generally used as part of generating the physics grid. Provides functions to apply dust extinction and compute the band integrated fluxes from the spectra.
The core attributes of both classes are the same.
lamb: wavelengths of the seds.
seds: fluxes at the wavelengths
grid: astropy Table giving the parameters of each sed (mass, age, etc.)
keys: column names of the grid (can be present w/o grid existing: e.g., backend=”disk”)
nbytes: the memory used by the class
SpectralGridfor more attributes
Both grids can interface with files via backends on disk with differing levels memory usage.
MemoryBackendreads contents of the file into memory. Largest memory usage and fastest access to the information after reading is done.
CacheBackendreads the contents into memory when that specific information is first accessed.
DiskBackendreads the data from disk when it is accessed. This supports reading only a portion of the data. This allows spectral and SED grids larger than can fit into memory.
The current file formats that are supported by the backends are:
fits: Supported by memory and cache backends (disk backend support TBD)
hdf: Supported by memory, cache, and disk backends