A MeqSpigot node is associated with one interferometer. It is connected to an input visibility stream. The node is a leaf. For each matching tile in the stream, it waits for the right Request, and returns a Result object containing data from the tile. (It is up to the tree designer to ensure that each MeqSpigot has a matching MeqSink somewhere below, so that requests and results come in the right sequence.)
A MeqSpigot is controlled by the following fields of its init-record:
|
field |
default |
description |
|
station_1_index |
|
interferometer pair |
|
station_2_index |
|
|
|
input_col |
|
tile column to read |
|
corr_index |
[] |
correlation map (see below) |
|
dims |
[2,2] |
output tensor dimensions (see below) |
|
flag_mask |
0 |
see FlagManagement |
|
flag_bit |
0 |
A MeqSpigot does not necessarily have to deal with visibility data. For example, a spigot that reads weights can be created by specifying "WEIGHT" for the input column. In general, three kinds of tile columns may be read:
- one-dimensional
one double or complex value per timeslot. A single time-variable Vells is generated, of type double or complex double.
- two-dimensional
N double or complex values per timeslot (corresponding to N frequencies). A single time-freq-variable Vells is generated, of type double or complex double.
- three-dimensional
NxM double or complex values per timeslots (N frequencies and M correlations). A tensor of time-freq-variable Vells can be generated, of type double or complex double.
In the latter case, the corr_index and ndims fields come into play. Tensor element i is mapped to correlation corr_index[i] in the tile. Note that, as usual, tensor elements are decomposed in row-major order. I.e., the elements of a 2x2 tensor are numbered as follows:
|
0 |
1 |
|
2 |
3 |
If specified, dims gives the dimensions of the tensor. The product of dims must be equal to the length of corr_index. If no dims are specified, the Result of the spigot is a vector or scalar. An index of -1 (0 in Glish) produces a NullVellSet in that tensor position. If corr_index is empty (the default), a 1-1 mapping is used.
- Example 1
The tiles contain four correlations: XX XY YX YY. dims is [2,2], and corr_index is either empty, or [0,1,2,3] (or [1,2,3,4] in Glish). The Result is arranged in a 2x2 coherency matrix, as [[XX,XY],[YX,YY]].
- Example 2
The tiles contain two correlations: XX YY. dims is [2,2], and corr_index is [0,-1,-1,1] (or [1,0,0,2] in Glish). The Result is arranged in a 2x2 coherency matrix, as [[XX,0],[0,YY]].
- Example 3
The tiles contain two correlations: XX YY. dims is unset, and corr_index is [0] (or [1] in Glish). The Result is a scalar containing the XX correlation only.