A simple table reader to be implemented into AthenaK.
Every table file is constructed of an ASCII header and a binary blob. A typical header looks like the following:
<metadatabegin>
version=[version #]
endianness=[little or big]
precision=[single or double]
<metadataend>
<scalarsbegin>
mb=[baryon mass]
<scalarsend>
<pointsbegin>
nn=[# of points in n]
nYe=[# of points in Ye]
nT=[# of points in T]
<pointsend>
<fieldsbegin>
P // Pressure
s // entropy per baryon
mub // baryon chemical potential
muq // charge chemical potential
mul // lepton chemical potential
e // energy density
cs // sound speed
[other fields]
<fieldsend>
Note that the file specification does not require all these point labels or fields to be present, nor does it prohibit more from being included. In principle this means that the same format can be used for a 1D table for a zero-temperature nuclear EOS in weak equilibrium, a 3D microphysical table (as suggested by the header above), a table of temperature-dependent chemical reaction rates, etc.
The binary blob for this header would be structured as follows:
<n0><n1><n2>...<nn-1>
<Ye0><Ye1><Ye2>...<nYe-1>
<T0><T1><T2>...<nT-1>
<P(n0,Ye0,T0)><P(n0,Ye0,T1)>...<P(n0,Ye0,nT-1)>
<P(n0,Ye1,T0)><P(n0,Ye1,T0)>...<P(n0,Ye1,nT-1)>
...
<P(n0,Yen-1,T0)><P(n0,Yen-1,T1)>...<P(n0,nYe-1,nT-1>
<P(n1,Ye0,T0)><P(n1,Ye0,T1>...<P(n1,Ye0,nT-1)>
...
...
<P(nn-1,Yen-1,T0><P(nn-1,Yen-1,T1)>...<P(nn-1,nYe-1,nT-1)>
<repeat for other fields>
Both table points and table fields in the binary blob will always be specified in the order they're written in the header, and the fields will be indexed assuming this same point ordering.