MHO_Passband

Purpose

MHO_Passband selects a contiguous frequency range for either inclusion or exclusion across all channels. Depending on the ordering of the two frequency limits, the operator either cuts out a specific band of spectrum (exclusion mode) or retains only a specific band and cuts everything outside (inclusion mode). This provides a simple mechanism for restricting analysis to a known clean frequency range or for removing a specific interferer band.

Control File Trigger

  • Keyword: passband

  • Category: flagging

  • Priority: 4.5

Parameters for passband

Parameter

Type

Description

value

list_real

List of exactly two frequency values (in MHz) specifying the band limits. The interpretation depends on their order: if first < second, the band [first, second] is the inclusion range (everything outside is cut). If second < first, the band [second, first] is the exclusion range (only this band is cut).

The builder (MHO_PassbandBuilder) validates that exactly two frequency values are provided and passes them to the operator’s SetPassband method. The default behavior (when neither ordering is explicitly set) is exclusion mode, which matches the legacy fourfit convention.

Input Data

This operator acts on the visibility_type container and simultaneously modifies the weight_type container (supplied via SetWeights).

Algorithm

MHO_Passband has no Initialize method; all work occurs in ExecuteInPlace.

The operator supports two modes of operation, distinguished by the fIsExclusion flag set during SetPassband:

Exclusion mode (\(f_{\rm second} < f_{\rm first}\)):

The frequency band \([f_{\rm low}, f_{\rm high}]\) is excised; everything outside is retained.

  1. Iterate over every frequency channel.

  2. Retrieve the channel’s sky frequency, bandwidth label, and net_sideband label.

  3. Compute the channel’s lower and upper frequency limits using DetermineChannelFrequencyLimits:

    \[\begin{split}[f_{\rm chan\_low}, f_{\rm chan\_high}] = \begin{cases} [f_{\rm sky},\ f_{\rm sky} + B] & \text{USB} \\ [f_{\rm sky} - B,\ f_{\rm sky}] & \text{LSB} \end{cases}\end{split}\]

  4. Use FindIntersection to check whether the exclusion band overlaps the channel interval.

  5. If there is overlap, iterate over all spectral points:

    1. Compute the absolute frequency \(f_{\rm sp} = f_{\rm sky} + s_b \cdot \Delta f_{\rm sp}\) where \(s_b = +1\) for USB and \(s_b = -1\) for LSB.

    2. If \(f_{\rm low} < f_{\rm sp} < f_{\rm high}\), zero out the visibility slice and increment the zeroed counter.

  6. Compute the used bandwidth fraction and rescaling factor, apply to weights, and store metadata.

Inclusion mode (\(f_{\rm first} < f_{\rm second}\)):

Only the frequency band \([f_{\rm low}, f_{\rm high}]\) is retained; everything outside is cut.

  1. Iterate over every frequency channel.

  2. Retrieve channel frequency information and compute the channel interval \([f_{\rm chan\_low}, f_{\rm chan\_high}]\).

  3. Use FindIntersection to check whether the inclusion band overlaps the channel interval.

  4. If there is overlap:

    1. Iterate over all spectral points.

    2. For each spectral point, if \(f_{\rm sp} < f_{\rm low}\) or \(f_{\rm sp} > f_{\rm high}\) (outside the inclusion band), zero out the visibility slice and increment the counter.

    3. Compute the used bandwidth fraction and rescaling factor, apply to weights, and store used_bandwidth_fraction on the visibility channel axis and rescaling_factor on the weight container’s channel axis.

  5. If there is no overlap (channel is entirely outside the inclusion band):

    1. Zero out the entire channel (all spectral points, all polarization products, all accumulation periods).

    2. Multiply all weight entries for the channel by zero.

    3. Store used_bandwidth_fraction = 0.0 and rescaling_factor = 0.0.

Effect on Data

In exclusion mode, spectral points whose absolute frequency falls within the specified band [f_low, f_high] are zeroed. In inclusion mode, all spectral points outside [f_low, f_high] are zeroed, and channels with no overlap with the inclusion band are entirely zeroed (both visibility and weight data). In both modes, the weight container is rescaled per channel using 1/frac where frac is the fraction of retained spectral points. used_bandwidth_fraction is stored on the visibility channel axis, and rescaling_factor is stored on the weight container’s channel axis.