Integrating a Constant Image With Masked Values

This test shows that nothing wrong happens at mask borders. For those we want to integrate an image that is one everywhere except for the masked regions

We use the following calibration without Polarization correction and mask:

{
    "Geometry": {
        "Tilt": {
            "TiltAngleDeg": -0.56,
            "TiltRotDeg": 73.569
        },
        "DedectorDistanceMM": 1031.657,
        "BeamCenter": [
            808.37,
            387.772
        ],
        "Imagesize": [
            1043,
            981
        ],
        "PixelSizeMicroM": [
            172.0,
            172.0
        ]
    },
    "Directory": [
        "."
    ],
    "Masks": [
        {
            "PixelPerRadialElement": 1,
            "MaskFile": "data/AAA_integ.msk",
            "Oversampling": 2,
            "qStart": 0.0, 
            "qStop": 5.0
        }
    ],
    "Wavelength": 1.54
}

The image we are going to integrate is exactly the array the SAXS.openmask() returns:

(Source code, png, hires.png, pdf)

The result is constant 1 (where the intensity is not 0), save 2e-12.

(Source code, png, hires.png, pdf)

Doing the same with Fit2d,

$ fit2d -svar\#IN=mask.tif -dim2000x2000 -svar\#OUT=data/const.chi -mac../data/AAA_integ_Pilatus1M_cmd.mac
 
STATUS: The error was identified in module:
IO: Input/Output and Status
 
STATUS: Position where error condition was identified
Subroutine IO_OPEN_INMACRO V0.2
 
STATUS: The error condition has been classified as:
Failed to open file: (file not found or bad file name ?)
...

results in something similar, just with less precision, about 10e-7. Probably because of single precision arithmetics.

(Source code, png, hires.png, pdf)