Integration channel jumping

[Juergen Reuter]

WHIZARD 2 seems to jump from result to another one whenever its finds a new or discards some old channel. See this example:

  16     716410  3.9351784E+00  1.32E+00   33.63  284.68*   0.00
  17     716410  4.3737785E+00  1.27E+00   29.03  245.73*   0.01
  18     716410  4.8151479E+00  1.25E+00   25.90  219.20*   0.00
  19     716410  4.8899320E+00  1.24E+00   25.41  215.05*   0.00
  20     716410  2.6357781E+00  6.80E-01   25.81  218.43    0.00
  21     716410  2.6142407E+00  5.61E-01   21.47  181.74*   0.00
  22     716410  2.7138862E+00  5.55E-01   20.43  172.95*   0.00
  23     716410  2.4408394E+00  4.72E-01   19.34  163.68*   0.00
  24     716410  2.2756128E+00  4.26E-01   18.73  158.51*   0.00
  25     716410  2.2692842E+00  4.10E-01   18.08  153.01*   0.00
  26     716410  4.6708383E-01  1.12E-01   23.95  202.73    0.00
  27     716410  4.7044695E-01  1.12E-01   23.77  201.20*   0.00

Here you can see the jumps after iterations 19 and 25. This makes one completely unsure whether one is in the correct ballpark for a result!!! Especially its unclear how many iterations one really needs. This seems to have been better in WHIZARD 1.

[kilian]

The given output clearly shows an unstable result, visible in the accuracy (in the hundreds) and efficiency (below 1e-4). This run should be discarded anyway.

The real question is how to generate a phase space parameterization so that convergence is possible. This is ticket #389. I close this as invalid or duplicate.