Creating specgrid from own synthetic spectra grids

[spacefox-101]

Hello,

I have a bunch of synthetic spectra that I have created for my science case. I wanted to know if it is possible to create a specgrid out of those synthetic spectra. I tried the tutorial on "https://msg.readthedocs.io/en/stable/user-guide/data-files.html" and when I create the specint files the labels are registered but when I create the specgrid file from the .txt file with all the specint file list the specgrid file fails to retain the axis labels from specint files.

The commands that I have been using are:

1. for ascii_to_specint:

$MSG_DIR/bin/ascii_to_specint ~/t4530_g+3.00_z-4.00_a+0.00_v00.13_x13+0.05.mod ~/specint002.h5 --label=teff:4530 --label=logg:3.0 --label=z:-4.0

2. for specint_to_specgrid:

$MSG_DIR/bin/specint_to_specgrid manifest.txt specgrid.h5 --grid-label=teff --grid-label=logg --grid-label=z

(I tried with and without the gird-label option but to no avail)

Please let me know if there is something that I need to do/can be done with this aspect.

Thank you so much!

[rhdtownsend]

Hi --

The --grid-label option is used to set the overall label of the grid, not the labels of the individual axes.

For me to get a handle on what's going, can you post output from running the following commands:

h5dump -a ~/specint002.h5

and

inspect_grid -a specgrid.h5

cheers,

Rich

[spacefox-101]

Hello,

Thank you for the quick response! I am adding the h5dump and the inspect_grid for the respective files.

For h5dump -a ~/specint002.h5 (lower case a) it gives me an error:

h5dump -a specint002.h5

h5dump error: missing file name
usage: h5dump [OPTIONS] files
(and the rest is just explanation of the usage of h5dump)

For h5dump -A ~/specint002.h5 (upper case A):

h5dump -A ~/specint002.h5

HDF5 "specint002.h5" {
GROUP "/" {
ATTRIBUTE "REVISION" {
DATATYPE H5T_STD_I32LE
DATASPACE SCALAR
DATA {
(0): 1
}
}
ATTRIBUTE "TYPE" {
DATATYPE H5T_STRING {
STRSIZE 14;
STRPAD H5T_STR_SPACEPAD;
CSET H5T_CSET_ASCII;
CTYPE H5T_C_S1;
}
DATASPACE SCALAR
DATA {
(0): "limb_specint_t"
}
}
ATTRIBUTE "lam_max" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SCALAR
DATA {
(0): 3964.36
}
}
ATTRIBUTE "lam_min" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SCALAR
DATA {
(0): 3941.18
}
}
ATTRIBUTE "precise" {
DATATYPE H5T_STD_I32LE
DATASPACE SCALAR
DATA {
(0): 0
}
}
DATASET "c" {
DATATYPE H5T_IEEE_F32LE
DATASPACE SIMPLE { ( 308, 1 ) / ( 308, 1 ) }
}
GROUP "labels" {
ATTRIBUTE "logg" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SCALAR
DATA {
(0): 3
}
}
ATTRIBUTE "teff" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SCALAR
DATA {
(0): 4530
}
}
ATTRIBUTE "z" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SCALAR
DATA {
(0): -4
}
}
}
GROUP "limb" {
ATTRIBUTE "REVISION" {
DATATYPE H5T_STD_I32LE
DATASPACE SCALAR
DATA {
(0): 1
}
}
ATTRIBUTE "TYPE" {
DATATYPE H5T_STRING {
STRSIZE 6;
STRPAD H5T_STR_SPACEPAD;
CSET H5T_CSET_ASCII;
CTYPE H5T_C_S1;
}
DATASPACE SCALAR
DATA {
(0): "limb_t"
}
}
ATTRIBUTE "law" {
DATATYPE H5T_STRING {
STRSIZE 16;
STRPAD H5T_STR_SPACEPAD;
CSET H5T_CSET_ASCII;
CTYPE H5T_C_S1;
}
DATASPACE SCALAR
DATA {
(0): "CONST "
}
}
}
GROUP "range" {
ATTRIBUTE "REVISION" {
DATATYPE H5T_STD_I32LE
DATASPACE SCALAR
DATA {
(0): 1
}
}
ATTRIBUTE "TYPE" {
DATATYPE H5T_STRING {
STRSIZE 11;
STRPAD H5T_STR_SPACEPAD;
CSET H5T_CSET_ASCII;
CTYPE H5T_C_S1;
}
DATASPACE SCALAR
DATA {
(0): "tab_range_t"
}
}
DATASET "x" {
DATATYPE H5T_IEEE_F64LE
DATASPACE SIMPLE { ( 309 ) / ( 309 ) }
}
}
}
}

For

$MSG_DIR/bin/inspect_grid -a specgrid.h5

Output:

Type: specgrid
Label:
Axes [0] :
Wavelength: (3941.1833000000001 -> 3964.3582000000001)

P.s. Just a heads up I am creating a grid only using 1 specint file for now as a test run. I hope that is not the cause of the problem.

Thank you!

[rhdtownsend]

Using just a single specint in the grid is likely what's causing the problem -- axes where there is only a single value are considered "degenerate", and automatically removed. Try creating a grid with four teff, z values arranged in a square.

[spacefox-101]

Ahh, I see. Thank you, I'll try that out and let you know how that goes.

Quick question, what do you mean by arranged in a sqaure for the arranged values?

[rhdtownsend]

I mean that, if you have two distinct temperature values teff_a, teff_b, and two distinct metallicities z_a, z_b, then you construct a grid of four spectra with the following teff/z values:

teff_a, z_a
teff_a, z_b
teff_b, z_a
teff_b, z_b

Of course, this is just to check that things work. Your real grid will likely have many more spectra.

[spacefox-101]

Thank you so much for the help! I made a grid of 4 teff and logg values and the specgrid.h5 file now recognises those labels.

A quick question, my synthetic spectra grid files are in such a way that they include teff, log(g), z and [X/Fe]; for example:

t4530_g+3.00_z-4.00_a+0.00_v00.13_x13+0.05.mod

Here t is teff = 4530 K, g is log(g) = 3.0 dex, z = -4 dex and x13 is [X/Fe] = +0.05 dex. Do you know if the [X/Fe] value is recognised as a label as well and treated accordingly? As far as I understand the documentation it should but are [X/Fe] values considered in the math for calculating the observed flux?

Thank you so much for your patience and for all the help!

[rhdtownsend]

Yes, any and all labels will be accepted by MSG. Whether the interpolated flux depends on a given label (e.g., [X/Fe]) will depend on whether the spectra themselves depend on that label -- MSG is completely agnostic to the meaning of the label, it just treats it as an additional dimension to interpolate over.

[spacefox-101]

Hello, so I was able to make my grid of synthetic spectra and msg recognises all the labels and axis perfectly. The only problem right now is that I get the "LookupError: unavailable data" no matter what parameters I choose. I am attaching a screenshot of the axis_min and max and my parameters as an example.

Thank you.