`astrohack.panel`¶

Module Contents¶

Functions¶

panel(image_name: str, panel_name: str = None, clip_type: str = 'sigma', clip_level: Union[float, dict[dict[float]]] = 3.0, use_detailed_mask: bool = True, panel_model: str = 'rigid', panel_margins: float = 0.05, polarization_state: str = 'I', ant: Union[str, List[str]] = 'all', ddi: Union[int, List[str]] = 'all', parallel: bool = False, overwrite: bool = False)

Analyze holography images to derive panel adjustments

panel(image_name: str, panel_name: str = None, clip_type: str = 'sigma', clip_level: Union[float, dict[dict[float]]] = 3.0, use_detailed_mask: bool = True, panel_model: str = 'rigid', panel_margins: float = 0.05, polarization_state: str = 'I', ant: Union[str, List[str]] = 'all', ddi: Union[int, List[str]] = 'all', parallel: bool = False, overwrite: bool = False)[source]¶

Analyze holography images to derive panel adjustments

Parameters

image_name (str) – Input holography data file name. Accepted data formats are the output from astrohack.holog.holog and AIPS holography data prepackaged using astrohack.panel.aips_holog_to_astrohack.
panel_name (str, optional) – Name of output file; File name will be appended with suffix .panel.zarr. Defaults to basename of input file plus holography panel file suffix.
clip_type (str, optional) – Choose the amplitude clipping algorithm: none, absolute, relative, sigma or noise_threshold, default is sigma
clip_level (float, dict, optional) – Choose level of clipping, can also be specified for specific antenna and DDI combinations by passing a dictionary, default is 3 (appropriate for sigma clipping)
use_detailed_mask (bool, optional) – Use a detailed aperture mask, i.e. include arm shadows for the VLA or include regions outside central circular aperture for the ngvla, default is True.
panel_model (str, optional) – Model of surface fitting function used to fit panel surfaces, None will default to “rigid”. Possible models are listed below.
panel_margins (float, optional) – Relative margin from the edge of the panel used to decide which points are margin points or internal points of each panel. Defaults to 0.05.
polarization_state (str, optional) – Select the polarization state over which to run panel, only parallel hands or stokes I should be used, default is I.
ant (list or str, optional) – List of antennas/antenna to be processed, defaults to “all” when None, ex. ea25
ddi (list or int, optional) – List of ddi to be processed, defaults to “all” when None, ex. 0
parallel (bool, optional) – Run in parallel. Defaults to False.
overwrite (bool, optional) – Overwrite files on disk. Defaults to False.

Returns

Holography panel object.

Return type

AstrohackPanelFile

Each Stokes I aperture image in the input image file is processed in the following steps:

Code Outline

Phase image is converted to a physical surface deviation image.

A mask of valid signals is created by using the relative cutoff on the amplitude image.

From telescope panel and layout information, an image describing the panel assignment of each pixel is created

Using panel image and mask, a list of pixels in each panel is created.

Pixels in each panel are divided into two groups: margin pixels and internal pixels.

For each panel:

Internal pixels are fitted to a surface model.

The fitted surface model is used to derive corrections for all pixels in the panel, internal and margins.

The fitted surface model is used to derive corrections for the positions of the screws.

A corrected deviation image is produced.

RMS is computed for both the corrected and uncorrected deviation images.

All images produced are stored in the output .panel.zarr file.

Available panel surface models:

AIPS fitting models:

mean: The panel is corrected by the mean of its samples.

rigid: The panel samples are fitted to a rigid surface (DEFAULT model).

Corotated Paraboloids: (the two bending axes of the paraboloid are parallel and perpendicular to a radius of the antenna crossing the middle point of the panel):

corotated_scipy: Paraboloid is fitted using scipy.optimize, robust but slow.

corotated_lst_sq: Paraboloid is fitted using the linear algebra least squares method, fast but unreliable.

corotated_robust: Tries corotated_lst_sq, if it diverges falls back to corotated_scipy, fast and robust.

Experimental fitting models:

xy_paraboloid: fitted using scipy.optimize, bending axes are parallel to the x and y axes.

rotated_paraboloid: fitted using scipy.optimize, bending axes can be rotated by any arbitrary angle.

full_paraboloid_lst_sq: Full 9 parameter paraboloid fitted using least_squares method, tends to heavily overfit surface irregularities.

Amplitude clipping:

In order to produce results of good quality parts of the aperture with low signal (e.g. the shadow of the secondary mirror support) a mask is defined based on the amplitude of the aperture. There are 5 methods (clip_type parameter) available to define at which level (clip_level) the amplitude is clipped:

none: In this method no amplitude clip is performed, i.e. the clipping value is set to -infinity.

absolute: In this method the clipping value is taken directly from the clip_level parameter, e.g.: if the user calls panel(…, clip_type=’absolute’, clip_level=3.5) everything below 3.5 in amplitude will be clipped

relative: In this method the clipping value is derived from the amplitude maximum, e.g.: if the user calls `panel(…, clip_type=’relative’, clip_level=0.2) everything below 20% of the maximum amplitude will be clipped

sigma: In this method the clipping value is computed from the RMS noise in the amplitude outside the physical dish, e.g.: if the user calls panel(clip_type=’sigma’, clip_level=3) everything below 3 times the RMS noise in amplitude will be clipped.

noise_threshold: In this model the cut is first set to the maximum amplitude outside the disk, a proxy for the noise maximum in amplitude, if this preserves a fraction of the aperture disk that is larger than the clip_level this is the chosen amplitude cutoff, if not, the cutoff is iteratively lowered by 10% until it preservers a fraction of the disk that is larger thatn clip_level. This heuristic was created with help from VLA operations.

The default clipping is set to 3 sigma.

Passing a dictionary for amplitude clipping:

The dictionary used for specifying amplitude clippings specific for each antenna and DDI combination must follow the following scheme:
amp_clip_dict = {
    ant1_name: {
        0: 0.3
        1: 0.5
    }
    ant2_name: {
        0: 0.45
        1: 0.21
    }
}
Where the antenna name is the usual antenna designation e.g. ea24 for the VLA or DV42 for ALMA, and the DDI number must be given as an integer.

AstrohackPanelFile Panel object allows the user to access panel data via compound dictionary keys with values, in order of depth, ant -> ddi. The panel object also provides a summary() helper function to list available keys for each file. An outline of the panel object structure is show below:

panel_mds =
{
    ant_0:{
        ddi_0: panel_ds,
         ⋮
        ddi_m: panel_ds
    },
    ⋮
    ant_n: …
}

Example Usage

from astrohack.panel import panel

# Fit the panels in the aperture image by using a rigid panel model and
# excluding the border 5% of each panel from the fitting.
panel_mds = panel(
    "myholo.image.zarr",
    panel_model='rigid',
    panel_margin=0.05
)

# fit the panels in the aperture image by using a rigid panel model and
# excluding points in the aperture image which have an amplitude that is less than 20% of the peak amplitude.
panel_mds = panel(
    "myholo.image.zarr",
    clip_type='relative',
    clip_level=0.2
)

astrohack.panel¶

Module Contents¶

Functions¶

`astrohack.panel`¶