Time-ordered data (TOD)

Stripeline contains functions that create the timeline of scientific data as produced by each polarimeter and distribute it across several parallel processes, potentially running through MPI.

A timeline of scientific data is generically called a TOD, and the polarimeters implemented in Strip produce a complex flow of data, as they output eight series:

Four series are called PWR series: PWR_Q1, PWR_Q2, PWR_U1, and PWR_U2;
Four series are called DEM series: DEM_Q1, DEM_Q2, DEM_U1, and DEM_U2.

Each of these series is sampled at 50 Hz and contains scientific data (in the preliminary unit tests done in 2017, the sampling frequency was 25 Hz; keep this in mind if you happen to replicate some returns of those ancient tests!); the raw instrument represents them as integer numbers, but Stripeline usually assumes they have already been calibrated to Kelvin and therefore uses floating-point numbers.

A TOD is represented by the structure StripTod, which contains the following fields:

polarimeters: a generic sequence of values, which represent each polarimeter stored in the TOD. Very common choices for this field are a range (for example, 1:49 represents all the Q-band polarimeters) or a list of strings, e.g., ["I0", "I3", "V2"].
time_range: a range representing the time of each sample in the TOD. Very common choices for this are floating-point ranges like 0.0:(1 / 50):100.0 (100 s sampled at 50 Hz) and ranges of astronomical times, created using the AstroTime package.
samples: this is the matrix containing the actual scientific samples. It has a shape of $N \times 8 \times m$, where $N$ is the number of samples along the time axis and must be equal to length(time_range), and $m$ is the number of polarimeters, which must be equal to length(polarimeters). The middle dimension is 8 because there are eight timelines produced by each polarimeter (see above).

Allocating the TOD is a matter of calling the function allocate_tod:

using Stripeline

# One minute of data sampled at 50 Hz, 10 polarimeters
tod = allocate_tod(StripTod, Float32, 0.0:(1.0 / 50.0):60.0, 1:10)

TOD(10 polarimeters, 3001 rows/polarimeter, time range from 0.0 to 60.0)

In a MPI environment, you can pass the keywords mpi_rank and mpi_size to allocate_tod and the function will only allocate the subset of the overall time range that is assigned to the running MPI process.

Once a StripTod variable is allocated, accessing the samples in the TOD can be done through the field samples, which is a matrix:

println("The shape of the TOD is ", size(tod.samples))

The shape of the TOD is (3001, 8, 10)

Accessing the right column can be cumbersome, as there are eight of them (PWR_Q1, PWR_Q2, …), so Stripeline provides the constants PWR_Q1_RANK, PWR_Q2_RANK, … for this:

# Print the minimum sample acquired by PWR_Q1 for polarimeter number #4
println(minimum(tod.samples[:, PWR_Q1_RANK, 4]))

0.0

The result is zero because allocate_tod sets all the samples to zero by default. An alternative approach is to use one of the functions pwrq1, pwrq2, pwru1, pwru2, demq1, demq2, demu1, demu2, which return a view of the original array and can thus be used both for reading and writing:

# Set one element of PWR_Q1 for polarimeter number #4
pwrq1(tod, 4)[142] = 1.36
println(
    "The maximum element of polarimeter #4 is now ",
    maximum(pwrq1(tod, 4)),
)

The maximum element of polarimeter #4 is now 1.36

Stripeline.StripTod — Type

A matrix of time-ordered PWR/DEM data for a subset of polarimeters

The fields of this structure are the following:

time_range: a range representing the time of each sample in the TOD.
samples: a N×8×m matrix, where N is the number of samples in the timeline, and m is the number of polarimeters. The middle dimension spans the three Stokes parameters I, Q, and U (in this order). The value N is equal to length(time_range)
rng: a pseudo-random number generator to be used for this TOD. It is guaranteed that the generator is uncorrelated with any other generator used for other StripTod objects created by the same call to allocate_tod.

Time-ordered data (TOD)

I/Q/U TODs