toltec-astro / macana2 Goto Github PK

This is a continuation of #2 and #7 . Introducing a schema based config parsing library would make it much easier to manage the config file validation and processing. An example of such library is https://www.codesynthesis.com/products/xsd

Due to an incorrect size specification of 'fitParams' and 'PreviousFitParams' in 'beammap.cpp' the errors are not being written to the nc beammap files. Also, the loop to populate these from 'fitToGaussian' does not loop over all 14 parameters.

it seems to take longer time to write out the beammaps

The current cmake script does not create the unit test executable. This should be fixed in order to let users without Qt/qmake to run the tests.

Added some fitting region limits to GaussFit.cpp as well as the option to give a guess vector with offsets from the bolostats file in beammap.cpp.

Also added a function writeSensToNcdf to Observation.cpp to write the new sensitivity calculations to the NetCDF file along with the FCF.

Finally, fixed a bug in writeFitParamsToNcdf that was resulting in the incorrect bolo names being written to the NetCDF file.

Need to implement the sensitivity calculation from the beammap fitted parameters.

We need to write tests for all the core algorithms that we are going to port to citlali. The same tests will be ported along with the covering functions to ensure correctness in the new citlali code.

Note that these tests should include both unit tests and feature tests.

Follows are a list of functionalities to be covered:

• Gaussian fitting (added with efed56f)
• PCA clean
• Time stream filtering
• Time stream sensitivity calculation
• Detector-level despiking
• Array-level despiking
• Telescope Pointing calculations
  see https://github.com/toltec-astro/macana2/blob/master/Observatory/Telescope.cpp
• Source Sky Coordinates calculations
  see https://github.com/toltec-astro/macana2/blob/master/Observatory/TimePlace.cpp
• kernel timestream generation
  see https://github.com/toltec-astro/macana2/blob/master/Observatory/Detector.cpp :1148
• all vector_utilities.cpp functions (or maybe these just get replaced by Eigen?)
• all astron_utilities.cpp functions (these must exist elsewhere!)

Test/simulated data creation and handling:

• netcdf format
• MUSCAT data handling
• simulated time-stream data
• ...

The current constructors for beammap and science map analparams are selected by a flag parameter and there are a lot of repeating in code.

use median weight of all time stream weights for those affected by bright sources

When running cmake, the linker is unable unable to find the OpenMP library:

CMake Error at CMakeLists.txt:73 (message):
  Cannot find OpenMP, abort

Here is the output of the homebrew installation of libomp:

==> Reinstalling libomp
==> Downloading https://homebrew.bintray.com/bottles/libomp-6.0.1.mojave.bottle.tar.gz
Already downloaded: /Users/paulhorton/Library/Caches/Homebrew/downloads/b4a4b5583565a13f5cbf05a8a1ab0d4c6953e1cc0cc6c944c0fc955812de055d--libomp-6.0.1.mojave.bottle.tar.gz
==> Pouring libomp-6.0.1.mojave.bottle.tar.gz
==> Caveats
On Apple Clang, you need to add several options to use OpenMP's front end
instead of the standard driver option. This usually looks like
  -Xpreprocessor -fopenmp -lomp

You might need to make sure the lib and include directories are discoverable
if /usr/local is not searched:

  -L/usr/local/opt/libomp/lib -I/usr/local/opt/libomp/include

For CMake, the following flags will cause the OpenMP::OpenMP_CXX target to
be set up correctly:
  -DOpenMP_CXX_FLAGS="-Xpreprocessor -fopenmp -I/usr/local/opt/libomp/include" -DOpenMP_CXX_LIB_NAMES="omp" -DOpenMP_omp_LIBRARY=/usr/local/opt/libomp/lib/libomp.dylib
==> Summary
🍺  /usr/local/Cellar/libomp/6.0.1: 12 files, 1.2MB

I have attempted to reinstall via homebrew but it appears that the issues is certain flags not being set in the cmake file that allow the library to be discoverable.

Any advice would be greatly appreciated!

Need to implement the PCA clean method for removing atmosphere in the time stream data.