Check linearity of detector response

[graeme-winter]

Look at data set recorded with range of transmissions, assess whether the response appears linear e.g. as we increase transmission, does spot integrated intensity increase the same way?

[graeme-winter]

from matplotlib import pyplot

from dials.array_family import flex

all = flex.reflection_table.from_file("scaled.refl")

all = all.select(all.get_flags(all.flags.scaled))

all = all.select(all["intensity.prf.value"] > 3 * flex.sqrt(all["intensity.prf.variance"]))

bits = { }

for j in range(5):
    bits[j] = all.select(all["id"] == j)

# treat last / strongest as reference

r = bits[4]
pyplot.xscale("log")
pyplot.yscale("log")
ratios = ["1%/20%", "2%/20%", "25%/20%", "10%/20%"]
for j in range(4):
    b = bits[j]
    m1, m2, d = r.match(b)
    ir = r.select(m1)["intensity.prf.value"]
    isr = r.select(m1)["intensity.prf.value"] / flex.sqrt(r.select(m1)["intensity.prf.variance"])
    ib = b.select(m2)["intensity.prf.value"]
    pyplot.scatter(isr.as_numpy_array(), (ib / ir).as_numpy_array(), s=0.2, label=ratios[j])
pyplot.legend()
pyplot.ylim(0.01, 2)
pyplot.show()

Suitable DIALS / Python scipt

[graeme-winter]

Looks reasonable

[graeme-winter]

 2023-06-22-11-59-24_insulin3_1pcT_scan_360deg/insulin3_1pcT_scan_360deg_0_0.h5
 2023-06-22-12-00-19_insulin3_2pcT_scan_360deg/insulin3_2pcT_scan_360deg_0_0.h5
 2023-06-22-12-00-57_insulin3_5pcT_scan_360deg/insulin3_5pcT_scan_360deg_0_0.h5
 2023-06-22-12-03-08_insulin3_10pcT_scan_360deg/insulin3_10pcT_scan_360deg_0_0.h5
 2023-06-22-12-04-48_insulin3_20pcT_scan_360deg/insulin3_20pcT_scan_360deg_0_0.h5

These data

[graeme-winter]

Ratios are not working out perfectly, but there is a small chance that the sample was moved so we shoot a "fresh" bit of crystal (which would change the illluminated volume and hence scale)