sir3d

Introduction

SIR3D is a parallel synthesis code for the synthesis of LTE lines based on SIR.

Installation

Download the code and type:

python setup.py install

If you want to work on the code it is much comfortable to install a development version by typing:

python setup.py develop

You need a working GFortran compiler available.

Features

It uses MPI in a master-agent topology for the fast calculation of Stokes profiles in 3D cubes
It can be easily adapted to read the output of different 3D MHD codes. Currently we allow for MuRaM cubes.
It uses a modern equation of state to compute the electron pressure from the temperature and total pressure.
It uses a modern continuum opacity package to compute the optical depth axis needed for the synthesis.
It interpolates the model to a fixed optical depth scale to output isotau surfaces.
It scales linearly with the number of available agents if not dominated by I/O processes.
It can also be used to spectrally and spatially degrade synthesis.

Example

Non-parallel

import sir3d

iterator = sir3d.Iterator(use_mpi=False)

mod = sir3d.Model('rempel.ini', rank=iterator.get_rank())
iterator.use_model(model=mod)

iterator.run_all_pixels()

You first instantiate an Iterator with use_mpi=False to work in serial mode. Then instantiate the Model passing the configuration file. Finally, we make the iterator use the model and finally synthesize all pixels.

Parallel using MPI

iterator = sir3d.Iterator(use_mpi=True, batch=256)

mod = sir3d.Model('rempel.ini', rank=iterator.get_rank())
iterator.use_model(model=mod)

iterator.run_all_pixels()

You first instantiate an Iterator with use_mpi=True to work in parallel mode. You also need to pass the batch. Note that since synthesizing a single pixel is quite fast, you can be dominated by I/O processes if batch is very small so we recommend to use a sufficiently large value (which also depends on the number of available agents). Then again instantiate the Model passing the configuration file. Finally, we make the iterator use the model and finally synthesize all pixels.

Subcubes

iterator.run_all_pixels(rangex=[0,100], rangey=[0,100])

Subcubes can be synthesized by providing the rangex and rangey arguments to the run_all_pixels method. If any of those are absent or None, the full range along the specific direction is used.

Configuration

Here is an example of a configuration file, which should be easy to understand.

# SIR3D configuration File
[General]
Stokes output = '/scratch1/3dcubes/output_stokes.h5'
Interpolated model output = '/scratch1/3dcubes/output_model.h5'
Interpolate tau = 0.0,-0.5,-1.0,-1.5,-2.0,-2.5,-3.0
EOS = 'SIR'

[Spectral regions]
    [[Region 1]]
    Name = '6301-6302'
    Wavelength range = 6300.8521, 6303.3119
    N. wavelengths = 458
    Spectral lines = 200, 201

[Atmosphere]
# Remember that "y" is the vertical direction in MURAM simulations but it is the vertical in this code
Type = 'MURAM'
Dimensions = 1536, 128, 1536
deltaz = 8e5
Maximum tau = 2.0
Tau delta = 0.1
Temperature = '/scratch1/3dcubes/rempel/spot_32x16x32km_ng/temp.float'
Pressure = '/scratch1/3dcubes/rempel/spot_32x16x32km_ng/pres.float'
Density = '/scratch1/3dcubes/rempel/spot_32x16x32km_ng/dens.float'
vz = '/scratch1/3dcubes/rempel/spot_32x16x32km_ng/vely.float'
Bx = '/scratch1/3dcubes/rempel/spot_32x16x32km_ng/magz.float'
By = '/scratch1/3dcubes/rempel/spot_32x16x32km_ng/magx.float'
Bz = '/scratch1/3dcubes/rempel/spot_32x16x32km_ng/magy.float'


    [[Multipliers]]
    Bx = 1.0
    By = 1.0
    Bz = 1.0
    vz = 1.0

Here is a decription of each parameter:

Stokes output: HDF5 file with the output
Interpolated model output: HDF5 file with the output model interpolated to certain optical depth surfaces
Interpolate tau: logarithm of the optical depth at 500 nm where to output the surfaces
EOS: specific equation of state to use. It can be either SIR (preferred option for consistency reasons) or MANCHA (the one used in the MANCHA code)
Name: name used for each spectral region
Wavelength range: starting and ending wavelengths used for the synthesis of each spectral region
N. wavelengths: number of wavelengths samples to consider in the range
Spectral lines: indices of the spectral lines to consider. The lines are listed on the LINEAS file
Type: type of input model atmosphere. For the moment the code only synthesizes Stokes profiles in MURAM snapshots but it is easy to adapt it to other formats.
Dimensions: number of (x,y,z) samples of the cube
deltaz: vertical size of the step in the simulation in km
Maximum tau: logarithmic optical depths above this value are considered thermalized and removed from the synthesis
Tau delta: maximum optical depth step during the integration of the radiative transfer equation. Vertical columns are reinterpolated to this step
Temperature: path to the file with the temperature cube
Pressure: path to the file with the pressure cube
Density: path to the file with the density cube
vz: path to the file with the vz cube (vertical velocity)
Bx: path to the file with the Bx cube (horizontal x component)
By: path to the file with the By cube (horizontal y component)
Bz: path to the file with the Bz cube (vertical z component)
Multipliers: these numbers will be multiplied by the corresponding vz, Bx, By or Bz cubes. This is useful because some cubes do not directly give the magnetic field

Degradation

This package can also be used to degrade observations with a spectral and/or spatial point spread function, also providing the output rebinned for the desired pixel size.

import sir3d

input_stokes = 'cube_stokes.h5'
input_model = 'cube_model.h5'

output_spatial_stokes = 'cube_stokes_spat_degraded.h5'
output_spatial_model = 'cube_model_spat_degraded.h5'

output_spatial_spectral_stokes = 'cube_stokes_spat_spec_degraded.h5'

# Read the spatial and spectral PSF from external files
...

lambda = np.loadtxt('wavelength_Hinode.txt') - 6301.5080
lambda_hinode *= 1e3

hinode_pixel = 0.16 * 725.
simulation_pixel = 48.0
zoom = simulation_pixel / hinode_pixel

tmp = sir3d.psf.PSF(use_mpi=True, batch=256)
tmp.run_all_pixels_spatial(input_stokes, input_model, output_spatial_stokes, output_spatial_model, spatial_psf=psf_spatial, zoom_factor=zoom)
tmp.run_all_pixels_spectral(input_stokes, output_spectral_stokes, spectral_psf=psf_spectral, final_wavelength_axis=lambda_hinode)

You first instantiate the sir3d.psf.PSF iterator like when you do synthesis.

The method run_all_pixels_spatial of this class carries out the spatial smearing for the Stokes and model cubes. Note that, although this smearing is well defined for the Stokes parameters, it is not for the model. What the code does is convolve the monochromatic Stokes images with the spatial PSF and then rebin to the desired pixel size. Additionally, it only rebins the model without using any PSF.

input_stokes: file with the Stokes synthesis
input_model: file with the model generated during the synthesis
output_spatial_stokes: file with the output smeared Stokes profiles
output_spatial_model: file with the output rebinned model
spatial_psf: 2D numpy array with the PSF
zoom_factor: ratio simulation_pixelsize/output_pixelsize

The method run_all_pixels_spectral of this class carries out the spectral smearing of the Stokes profiles.

input_stokes: file with the input Stokes profiles
output_spectral_stokes: file with the output spectrally smeared Stokes profiles
spectral_psf: 2D numpy array with the spectral PSF. The first column is the wavelength in mA. The second one is the PSF.
final_wavelength_axis: final wavelength axis in mA

Dependencies

numpy
mpi4py
h5py
scipy
tqdm
scikit-image
pyfftw

aasensio / 3d_sir