fitting isn't so robust (around best-fit model), but here's a representative "best-fit" model (I did a little steppar fitting on h_inf and didn't find a better fit): ======================================================================== Model windprof<1> + powerlaw<2> Source No.: 1 Active/On Model Model Component Parameter Unit Value par comp 1 1 windprof q 0.0 frozen 2 1 windprof taustar 1.88012 +/- 5.16815 3 1 windprof u0 0.731160 +/- 0.173937 4 1 windprof h 1.17914 +/- 2.91800 5 1 windprof tau0star 0.0 frozen 6 1 windprof beta 1.00000 frozen 7 1 windprof betaSob 0.0 frozen 8 1 windprof numerica 1 frozen 9 1 windprof anisotro 1 frozen 10 1 windprof rosselan 1 frozen 11 1 windprof expansio 0 frozen 12 1 windprof thick 0 frozen 13 1 windprof waveleng "A" 6.18220 frozen 14 1 windprof shift "mA" 0.0 frozen 15 1 windprof velocity "km/s" 2250.00 frozen 16 1 windprof verbose 0 frozen 17 1 windprof norm 7.81813E-06 +/- 2.16579E-06 18 2 powerlaw PhoIndex 2.00000 frozen 19 2 powerlaw norm 7.62000E-04 frozen ________________________________________________________________________ C-statistic = 130.45 using 128 PHA bins and 124 degrees of freedom. XSPEC12>goodness 1000 nosim 48.60% of realizations are < best fit statistic 130.45 "error 1. 2" choked - claimed to find better fit, but parameters it listed weren't better. Based on the isoporous fit (and reasonable expectations), it seems that taustar (and h_inf) are basically unconstrained. Yes, taustar is definitely unconstrained (taustar = 100 gives an acceptable fit, with very big h_inf) ======================================================================== Model windprof<1> + powerlaw<2> Source No.: 1 Active/On Model Model Component Parameter Unit Value par comp 1 1 windprof q 0.0 frozen 2 1 windprof taustar 99.9934 +/- 939.459 3 1 windprof u0 0.780711 +/- 0.114415 4 1 windprof h 9.00205 +/- 34.4125 5 1 windprof tau0star 0.0 frozen 6 1 windprof beta 1.00000 frozen 7 1 windprof betaSob 0.0 frozen 8 1 windprof numerica 1 frozen 9 1 windprof anisotro 1 frozen 10 1 windprof rosselan 1 frozen 11 1 windprof expansio 0 frozen 12 1 windprof thick 0 frozen 13 1 windprof waveleng "A" 6.18220 frozen 14 1 windprof shift "mA" 0.0 frozen 15 1 windprof velocity "km/s" 2250.00 frozen 16 1 windprof verbose 0 frozen 17 1 windprof norm 7.88960E-06 +/- 2.27915E-06 18 2 powerlaw PhoIndex 2.00000 frozen 19 2 powerlaw norm 7.62000E-04 frozen ________________________________________________________________________ C-statistic = 130.12 using 128 PHA bins and 124 degrees of freedom. Warning: Cstat statistic is only valid for Poisson data. ***Warning: New best fit found, fit parameters will be set to new values. Apparent non-monotonicity in statistic space detected. Current bracket values 49.081226, 48.793422 and delta stat 0.153429, 1.286101 but latest trial 48.855651 gives 0.152486 Suggest that you check this result using the steppar command. ***Warning: Parameter pegged at hard limit: 100 2 48.937324 0.000000 (-51.062564,-99.999888) XSPEC12>goodness 1000 nosim 44.30% of realizations are < best fit statistic 130.12 M ade two confidence grids; but running "error" is problematic. Steppar instead; though taustar and h_inf are clearly unconstrained, individually, when anisoporosity is allowed. So, we'll just test u_o and norm with steppar. XSPEC12>steppar 3 0.5 0.6 10 C-Statistic Delta u0 C-Statistic 3 134.42 4.2508 0 0.5 134.12 3.9553 1 0.51 133.82 3.6549 2 0.52 133.52 3.3537 3 0.53 133.23 3.0595 4 0.54 132.94 2.7728 5 0.55 132.66 2.494 6 0.56 132.39 2.2232 7 0.57 132.13 1.9615 8 0.58 131.88 1.7142 9 0.59 131.65 1.483 10 0.6 XSPEC12>steppar 3 0.6 0.65 10 C-Statistic Delta u0 C-Statistic 3 131.65 1.4853 0 0.6 131.54 1.3727 1 0.605 131.44 1.2681 2 0.61 131.34 1.1682 3 0.615 131.24 1.073 4 0.62 131.15 0.9828 5 0.625 131.11 0.93683 6 0.63 131.02 0.84799 7 0.635 130.96 0.78727 8 0.64 130.93 0.76396 9 0.645 130.81 0.64614 10 0.65 XSPEC12>steppar 3 0.85 0.95 10 C-Statistic Delta u0 C-Statistic 3 130.59 0.41963 0 0.85 130.73 0.55744 1 0.86 130.9 0.72799 2 0.87 131.01 0.83963 3 0.88 131.13 0.9623 4 0.89 131.26 1.0934 5 0.9 131.39 1.224 6 0.91 131.52 1.3544 7 0.92 131.48 1.3154 8 0.93 131.48 1.307 9 0.94 131.63 1.4587 10 0.95 XSPEC12>steppar 17 6.6e-6 6.7e-6 10 C-Statistic Delta norm C-Statistic 17 131.52 1.3467 0 6.6e-06 131.49 1.3243 1 6.61e-06 131.47 1.3022 2 6.62e-06 131.45 1.2803 3 6.63e-06 131.43 1.2585 4 6.64e-06 131.41 1.237 5 6.65e-06 131.38 1.2155 6 6.66e-06 131.36 1.1943 7 6.67e-06 131.34 1.1733 8 6.68e-06 131.32 1.1525 9 6.69e-06 131.3 1.1319 10 6.7e-06 XSPEC12>steppar 17 6.7e-6 6.8e-6 10 C-Statistic Delta norm C-Statistic 17 Fit Parameter: 2 has pegged 131.3 1.1324 0 6.7e-06 131.28 1.1118 1 6.71e-06 131.26 1.0914 2 6.72e-06 131.24 1.0713 3 6.73e-06 131.22 1.0514 4 6.74e-06 131.2 1.0317 5 6.75e-06 131.18 1.0121 6 6.76e-06 131.16 0.99276 7 6.77e-06 131.14 0.97363 8 6.78e-06 131.12 0.9547 9 6.79e-06 131.11 0.93595 10 6.8e-06 XSPEC12>steppar 17 9.0e-6 9.1e-6 10 C-Statistic Delta norm C-Statistic 17 Fit Parameter: 2 has pegged Fit Parameter: 2 has pegged 131.01 0.84795 0 9e-06 Fit Parameter: 2 has pegged 131.02 0.86001 1 9.01e-06 Fit Parameter: 2 has pegged 131.04 0.87445 2 9.02e-06 Fit Parameter: 2 has pegged 131.05 0.88981 3 9.03e-06 Fit Parameter: 2 has pegged 131.07 0.90573 4 9.04e-06 Fit Parameter: 2 has pegged 131.09 0.92198 5 9.05e-06 Fit Parameter: 2 has pegged 131.1 0.93843 6 9.06e-06 Fit Parameter: 2 has pegged 131.12 0.95508 7 9.07e-06 Fit Parameter: 2 has pegged gsl: qag.c:248: ERROR: roundoff error prevents tolerance from being achieved Default GSL error handler invoked. Abort (core dumped)