We want to have a Monte Carlo test set with a grid of telescope pointings, with corresponding IRFs, from which we can later interpolate to obtain the IRFs for a given run (or good-time-interval).
Mono-analysis case
For the single-telescope case (LST1 standalone) the relevant direction-dependent quantities which affect the performance are the airmass (which grows like 1 / cos(zenith) in the planar atmosphere approximation) and the component of the geomagnetic field in the plane orthogonal to the shower axis (i.e. more or less on the plane orthogonal to the pointing direction). It seems therefore reasonable to make the grid uniform in related parameters, like cos(zenith) and sin(delta), with delta being the angle between the pointing direction and the geomagnetic field. Those would also be the parameters used for the interpolation of the three vertices of the grid cell containing a given observation.
@chaimain proposed one such grid (the orange line shows the physical limit for the ORM site, and grey points are simply those beyond 70 deg of zenith):
Translated into a zenith vs. (astronomical) azimuth map (yellow and grey circles are the grid points below and above 70 deg of zenith; color map indicates the value of sin(delta)):
Blue lines are the paths of a few sources already observed by LST1, from south to north: the Galactic Center, RS Oph, PG 1553, Crab, Mrk 421, BL Lac, LHAASO J2108+5157, 1ES 1959 (most north-wise), and small white points just show intervals of 0.5 hours of observation.
Using the inverse of cos(zenith) would give even more weight to higher zenith angles, so most of the time would be spent on sky regions with few observations, so it does not seem to be a good idea.
The same as above, in azimuthal-equidistant projection:
In all of this, I have used the orientation of the magnetic field given by https://geomag.bgs.ac.uk/data_service/models_compass/igrf_calc.html for the location of ORM, 10 km a.m.s.l., and date=2021-12-01. Magnetic declination = -4.84 deg; Magnetic inclination = -37.36 deg
The coordinates (zenith, azimuth) in degrees are below. Azimuth is astronomical azimuth, measured from geographic north clockwise (i.e. N-E-S-W). Also shown are the suggested range of impact and energy, and the viewcone for producing "ring wobble" files:
The azimuth of course has to be converted to the Corsika system, i.e.:
PHIP = mod(180 - (azimuth + 4.84), 360)
All of the above is done with single-telescope (LST1) analysis in mind...
Stereo case
For stereo analysis, besides all of the above, we must consider the projection of the array footprint on the plane orthogonal to the shower axis (hence, roughly, to the telescope pointing). Taking into account the array formed by MAGIC-1, MAGIC-2 and LST-1, we can see the changes in the minimum and maximum array inter-telescope distances:
Minimum inter-telescope distance:
Maximum inter-telescope distance:
South-wise sources seem to be relatively well covered with the "cos zenith - sin delta" grid, but this does not seem to be the case for north-wise sources, particularly on the East-side of their path.
For the next production, I would focus on the single-telescope case. The issue with stereo analysis will be easier to study when the grid is produced, and additional grid points can be produced if needed. Probably the computing time will be completely dominated by the training MC set (which includes protons), and perhaps it is not too costly to simply make a denser "cos zenith - sin delta" grid everywhere.
We want to have a Monte Carlo test set with a grid of telescope pointings, with corresponding IRFs, from which we can later interpolate to obtain the IRFs for a given run (or good-time-interval).
Mono-analysis case
For the single-telescope case (LST1 standalone) the relevant direction-dependent quantities which affect the performance are the airmass (which grows like 1 / cos(zenith) in the planar atmosphere approximation) and the component of the geomagnetic field in the plane orthogonal to the shower axis (i.e. more or less on the plane orthogonal to the pointing direction). It seems therefore reasonable to make the grid uniform in related parameters, like cos(zenith) and sin(delta), with delta being the angle between the pointing direction and the geomagnetic field. Those would also be the parameters used for the interpolation of the three vertices of the grid cell containing a given observation. @chaimain proposed one such grid (the orange line shows the physical limit for the ORM site, and grey points are simply those beyond 70 deg of zenith):
Translated into a zenith vs. (astronomical) azimuth map (yellow and grey circles are the grid points below and above 70 deg of zenith; color map indicates the value of sin(delta)):
Blue lines are the paths of a few sources already observed by LST1, from south to north: the Galactic Center, RS Oph, PG 1553, Crab, Mrk 421, BL Lac, LHAASO J2108+5157, 1ES 1959 (most north-wise), and small white points just show intervals of 0.5 hours of observation.
Using the inverse of cos(zenith) would give even more weight to higher zenith angles, so most of the time would be spent on sky regions with few observations, so it does not seem to be a good idea.
The same as above, in azimuthal-equidistant projection:
In all of this, I have used the orientation of the magnetic field given by https://geomag.bgs.ac.uk/data_service/models_compass/igrf_calc.html for the location of ORM, 10 km a.m.s.l., and date=2021-12-01. Magnetic declination = -4.84 deg; Magnetic inclination = -37.36 deg
The coordinates (zenith, azimuth) in degrees are below. Azimuth is astronomical azimuth, measured from geographic north clockwise (i.e. N-E-S-W). Also shown are the suggested range of impact and energy, and the viewcone for producing "ring wobble" files:
The azimuth of course has to be converted to the Corsika system, i.e.: PHIP = mod(180 - (azimuth + 4.84), 360)
All of the above is done with single-telescope (LST1) analysis in mind...
Stereo case
For stereo analysis, besides all of the above, we must consider the projection of the array footprint on the plane orthogonal to the shower axis (hence, roughly, to the telescope pointing). Taking into account the array formed by MAGIC-1, MAGIC-2 and LST-1, we can see the changes in the minimum and maximum array inter-telescope distances:
Minimum inter-telescope distance:
Maximum inter-telescope distance:
South-wise sources seem to be relatively well covered with the "cos zenith - sin delta" grid, but this does not seem to be the case for north-wise sources, particularly on the East-side of their path.
For the next production, I would focus on the single-telescope case. The issue with stereo analysis will be easier to study when the grid is produced, and additional grid points can be produced if needed. Probably the computing time will be completely dominated by the training MC set (which includes protons), and perhaps it is not too costly to simply make a denser "cos zenith - sin delta" grid everywhere.