Scans.jl

AbstractExec

Abstract supertype for scan execution modes.

BatchExec(Nbatches, batch)

Execution mode to divide the scan into Nbatches chunks and run only the given batch.

CondorExec(scriptfile, ncores)

Execution mode which submits a scan to an HTCondor queue system claiming ncores cores.

LocalExec

Execution mode to simply run the whole scan in the current Julia session in a for loop.

QueueExec(nproc=0, queuefile="")

Execution mode to run a scan using a file-based queueing system. Can be run in multiple separate Julia sessions, or can spawn nproc subprocesses which then take items from the queue to run.

Possible values for nproc are:

• 0: run only in the current Julia process
• n > 0: spawn n subprocesses and run on these
• -1: spawn as many subprocesses as the number of logical cores on the CPU (Base.Sys.CPU_THREADS)

If queuefile is given, the queuefile is stored at that path. If omitted, the queuefile is stored in Utils.cachedir(). Note that the queuefile is deleted at the end of the scan.

RangeExec(range)

Execution mode to run a subsection of the scan, given by a UnitRange, in the current Julia session.

SSHExec(localexec, scriptfile, hostname, subdir)

Execution mode which transfers the scriptfile file to the host given by hostname via SSH and executes the scan on that host with a mode defined by localexec. subdir gives the subdirectory (relative to the home directory) where scans are stored on the remote host. A subfolder with automatically chosen name will be created in subdir to store this scan.

Scan(name; kwargs...)
Scan(name, ex::AbstractExec; kwargs...)

Create a new Scan with name name and variables given as keyword arguments. The execution mode ex can be given directly or via command-line arguments to the script. If given, command-line arguments overwrite any explicitly passed execution mode.

If neither an explicit execution mode nor command-line arguments are given, ex defaults to LocalExec, i.e. running the whole scan locally in the current Julia process.

addvariable!(scan, variable::Symbol, array)
addvariable!(scan; kwargs...)

Add scan variable(s) to the scan, either as a single pair of Symbol and array, or as a sequence of keyword arguments.

chunks(a::AbstractArray, n::Int)

Split array a into n chunks, spreading the entries of a evenly.

Examples

julia> a = collect(range(1, length=10));
julia> Scans.chunks(a, 3)
3-element Array{Array{Int64,1},1}:
 [1, 4, 7, 10]
 [2, 5, 8]
 [3, 6, 9]

makefilename(scan, scanidx)

Make an appropriate file name for an HDF5Output or prop_capillary output for the scan at the current scanidx.

Examples

scan = Scan("scan_example"; energy=collect(range(5e-6, 200e-6; length=64)))
runscan(scan) do scanidx, energyi
    prop_capillary(125e-6, 3, :He, 0.8; λ0=800e-9, τfwhm=10e-15, energy=energyi
                   filepath=makefilename(scan, scanidx))
end

runscan(f, scan)

Run the function f in a scan with arguments defined by the scan::Scan. The function f must have the signature f(scanidx, args...) where the length of args is the number of variables to be scanned over. Can be used with the do block syntax.

The exact subset and order of scan points which is run depends on scan.exec, see Scan.

Examples

scan = Scan("scan_example"; energy=collect(range(5e-6, 200e-6; length=64)))
runscan(scan) do scanidx, energyi
    prop_capillary(125e-6, 3, :He, 0.8; λ0=800e-9, τfwhm=10e-15, energy=energyi)
end