Detector-mapping alignment

Plans described in this section provide support for detector-mapping alignment on reflectometers.

The plans in this module expect:

Angle scan

API reference: ibex_bluesky_core.plans.reflectometry.angle_scan_plan

This plan takes a single DAE measurement, without moving anything.

The resulting plots & data files describe the relationship between angular position of each detector pixel, and the counts observed on that detector pixel - with the aim of finding the angle at which peak intensity occurs.

This plan returns the result of the angle fit, or None if the fit failed. The model used is a Gaussian.

The following is a full example of how angle_scan_plan may be called, configuring an appropriate SimpleDae.

Click to show example code 
from collections.abc import Generator

import numpy as np
from bluesky.utils import Msg
from lmfit.model import ModelResult

from ibex_bluesky_core.utils import get_pv_prefix
from ibex_bluesky_core.devices.simpledae import (
    PeriodGoodFramesWaiter,
    PeriodPerPointController,
    PeriodSpecIntegralsReducer,
    SimpleDae
)
from ibex_bluesky_core.plans.reflectometry import angle_scan_plan


def map_align_plan() -> Generator[Msg, None, ModelResult | None]:
    controller = PeriodPerPointController(save_run=True)
    waiter = PeriodGoodFramesWaiter(50)
    reducer = PeriodSpecIntegralsReducer(
        # Select spectrum 1 as the monitor
        monitors=np.array([1], dtype=np.int64),
        # and 2-128 inclusive as the detectors
        detectors=np.arange(2, 129),
    )

    prefix = get_pv_prefix()
    dae = SimpleDae(
        prefix=prefix,
        controller=controller,
        waiter=waiter,
        reducer=reducer,
    )

    angle_scan_result = yield from angle_scan_plan(
        dae,
        angle_map=np.linspace(-5, 5, num=127, dtype=np.float64),
    )

    if angle_fit := angle_scan_result:
        print(angle_fit.fit_report(show_correl=False))

    return angle_scan_result

Height & angle scan

API reference: ibex_bluesky_core.plans.reflectometry.height_and_angle_scan_plan

This plan scans over a provided height axis, taking a DAE measurement at each point. It then does simultaneous height & angle fits, using a single set of measurements - avoiding the need to scan height and angle separately.

It is assumed that optimum angle does not depend on height. The result of the angle scan returned by this plan is meaningless if this assumption is untrue. This may be verified visually using the generated heatmap, or by multiple invocations of angle_scan at different height settings - the optimum angle should not vary with height.

This plan produces multiple plots & data files from a single scan:

  • A heatmap visualising the 2-dimensional relationship between angle (x axis), height (y axis), and measured intensity (heatmap colour).

  • A set of plots & data files describing the relationship between integrated intensity (across every configured detector pixel, normalized by monitor), versus height.

  • A set of plots & data files describing the relationship of accumulated counts on each detector pixel, across all height points, versus angular pixel position. The only difference between this and the angle_scan plan above is that this accumulates data across multiple height points, and therefore benefits from the counting statistics of all measured height points.

This plan returns a typed dictionary, DetMapAlignResult, with the following keys:

  • "angle_fit": the result of the angle fit, or None if the fit failed.

  • "height_fit": the result of the height fit, or None if the fit failed.

Both the height & angle data are fitted using two independent Gaussian models.

The following is a full example of how angle_scan_plan may be called, configuring an appropriate SimpleDae and a BlockRw.

Click to show example code 
from collections.abc import Generator

import numpy as np
from bluesky.utils import Msg

from ibex_bluesky_core.utils import get_pv_prefix
from ibex_bluesky_core.devices.block import block_rw
from ibex_bluesky_core.devices.simpledae import (
    PeriodGoodFramesWaiter,
    PeriodPerPointController,
    PeriodSpecIntegralsReducer,
    SimpleDae
)
from ibex_bluesky_core.plans.reflectometry import (
    DetMapAlignResult,
    height_and_angle_scan_plan
)


def map_align() -> Generator[Msg, None, DetMapAlignResult]:
    # Could also be a reflectometry parameter, or any other movable
    block = block_rw(float, "some_block")

    controller = PeriodPerPointController(save_run=True)
    waiter = PeriodGoodFramesWaiter(50)
    reducer = PeriodSpecIntegralsReducer(
        # Select spectrum 1 as the monitor
        monitors=np.array([1], dtype=np.int64),
        # and 2-128 inclusive as the detectors
        detectors=np.arange(2, 129),
    )

    prefix = get_pv_prefix()
    dae = SimpleDae(
        prefix=prefix,
        controller=controller,
        waiter=waiter,
        reducer=reducer,
    )

    result = yield from height_and_angle_scan_plan(
        dae,
        block,
        5,
        15,
        num=21,
        angle_map=np.linspace(-5, 5, num=127, dtype=np.float64),
    )

    print("Height fit:")
    if height_fit := result["height_fit"]:
        print(height_fit.fit_report(show_correl=False))
    print("\n\n")
    print("Angle fit:")
    if angle_fit := result["angle_fit"]:
        print(angle_fit.fit_report(show_correl=False))
    print("\n\n")

    return result

Implementation

The following specialised components are used to implement the detector-mapping alignment workflow:

PeriodSpecIntegralsReducer

PeriodSpecIntegralsReducer is responsible for exposing arrays of DAE detector data.

It exposes numpy arrays of integer counts data from the current period, integrated over the entire time-of-flight dimension. This reducer is implemented by pulling the entire spectrum-data map from the DAE at once, so is efficient even for relatively large numbers of detectors.

This reducer cannot be used by itself in a scan, as it does not produce scalar data. It is only intended for use with downstream processors, such as DetMapHeightScanLiveDispatcher or DetMapAngleScanLiveDispatcher described below.

This reducer always accumulates data across the entire configured time-of-flight range - this is because it uses the spectrum-data map rather than reading individual spectra, and the spectrum-data map does not have access to time channel boundary information. For efficiency, if slicing data in time-of-flight or wavelength is desired, this should be done by changing DAE time-channel settings.

Warning

num_periods * (num_spectra + 1) * (num_time_channels + 1) may not exceed \(5000000\).

This is a limitation of the underlying EPICS spectrum-data map used to implement this reducer.

DetMapHeightScanLiveDispatcher

DetMapHeightScanLiveDispatcher is a bluesky LiveDispatcher which processes arrays generated by PeriodSpecIntegralsReducer by integrating counts across all detector spectra, and normalizing by integrated counts across all monitor spectra. This dispatcher therefore produces the same number of output events as input events (which is the same as the number of scan points).

For example, for the following input arrays:

event: [0, 10, 20, 10, 0]
event: [0, 1, 2, 3, 4]

The outputs would be:

event: 40
event: 10

DetMapAngleScanLiveDispatcher

DetMapAngleScanLiveDispatcher is a bluesky LiveDispatcher which processes arrays generated by PeriodSpecIntegralsReducer by accumulating counts from each scan point into an internal array. At the end of the scan, events are then emitted across this accumulated array.

This means that this dispatcher will always produce a number of events equal to the number of detectors (which is also equal to the length of the angle map), regardless of the number of scan points which were performed.

For example, for the following input arrays:

event: [0, 10, 20, 10, 0]
event: [0, 1, 2, 3, 4]

The outputs would be:

event: 0
event: 11
event: 22
event: 13
event: 4

LivePColorMesh

LivePColorMesh is the specialized heatmap plotting callback which reacts to rows of data at a time.

This callback updates live as the scan progresses. It is otherwise very similar to the existing bluesky plotting callbacks.