Calibration only calibrating the first few gains in array

Joshuaalbert commented 6 months ago

Only first few real components are changing. This probably implies some index is being selected inadvertently. Also, explains why residuals are not super tight.

"gains": {
    "type": "numpy.ndarray",
    "array_real": [
      [
        [
          [
            [
              [
                0.9810193777084351,
                -0.1584736406803131
              ],
              [
                0.11337465047836304,
                1.0116183757781982
              ]
            ],
            [
              [
                1.024043321609497,
                0.004812115803360939
              ],
              [
                0.112271748483181,
                0.9224221706390381
              ]
            ],
            [
              [
                1.0020111799240112,
                -0.26801177859306335
              ],
              [
                0.3347955048084259,
                1.0057669878005981
              ]
            ]
          ],
          [
            [
              [
                1.0,
                0.0
              ],
              [
                0.0,
                1.0
              ]
            ],
            [
              [
                1.0,
                0.0
              ],
              [
                0.0,
                1.0
              ]
            ],
            [
              [
                1.0,
                0.0
              ],
              [
                0.0,
                1.0
              ]
            ]
          ],
          [
            [
              [
                1.0,
                0.0
              ],
              [
                0.0,
                1.0
              ]
            ],
            [
              [
                1.0,
                0.0
              ],
              [
                0.0,
                1.0
              ]
            ],
            [
              [
                1.0,
                0.0
              ],
              [
                0.0,
                1.0
              ]
            ]
          ],
          [
            [
              [
                1.0,
                0.0
              ],

Joshuaalbert commented 5 months ago

I think what's happening is predict code is broadcasting somewhere where it shouldn't. Hence solution is a unittest for all predict implentations.

Joshuaalbert commented 5 months ago

Seems that only first antenna is corrected

Joshuaalbert / DSA2000-Cal

Calibration only calibrating the first few gains in array #51