calibration fails with additional parameters

[alexlib]

using the old fashion test_cavity case https://github.com/OpenPTV/test_cavity the full calibration (or fine orientation) works only when no additional parameters are allowed. If any of the additional parameters is marked - the calibration does not converge. The errors in some cases are small, but still larger than the CONVERGENCE value. What seems to happen is that there is no improvement, e.g. two of the beta array values got stuck into some values,

failing beta[0] = -0.015985
failing beta[1] = 0.011083
failing beta[2] = -0.026192
failing beta[3] = 0.000019
failing beta[4] = 0.000018
failing beta[9] = -0.001284
failing beta[0] = 0.015985
failing beta[1] = -0.011083
failing beta[2] = 0.026191
failing beta[3] = -0.000019
failing beta[4] = -0.000018
failing beta[9] = 0.001284
failing beta[0] = -0.015985
failing beta[1] = 0.011084
failing beta[2] = -0.026191
failing beta[3] = 0.000019
failing beta[4] = 0.000018
failing beta[9] = -0.001284
failing beta[0] = 0.015984
failing beta[1] = -0.011085
failing beta[2] = 0.026192
failing beta[3] = -0.000019
failing beta[4] = -0.000018
failing beta[9] = 0.001284

@yosefm - any idea how to approach it? When the evolutionary algorithm was developed - did it work for cases with additional parameters?

[alexlib]

in some cases the calibration finds a solution, but for some reason it is very difficult now. Most of the additional parameters fail it. any idea why this is so? @yosefm ?

[yosefm]

Sorry for the long response time. the evolutionary calibration was developed because the linear regression didn't handle additional parameters gracefully.

[alexlib]

got it @yosefm thanks