AMOE_mult comments

  • a given exp spectrum is normalized to RMS=1. This is somewhat arbitrary, yet pragmatic and simple.

  • AMOE_mult output (in nohup.out file) shows opt_fun lines for each exp spectrum. A rms-fit value < 0.1 for the residual's rms can be considered a good fit. A value in the range 0.15 - 0.3, corresponds to an average fit, as often seen in publications. That can be reached by a few trials with parameters set by hand. It may be the starting point of an AMOE calculation. rms-theo should be close to 1 for a reasonable fit.

  • especially the parameter for threshold_corr in Xray spectra is highly nonlinear. After a first model calculation its value can easily be guessed to 1eV accuracy. This value should be used as starting value for AMOE refinements.

  • as spectrum fitting is significantly non-linear, it may be wise to start with a minimal set of parameters and see what can be achieved. The minimal parameter set is less likely to get trapped at a local optimum fit. As further parameters get introduced, one may start from the previously best to obtain even better fits. If a further parameter lowers rms-fit by less than 0.01, we may consider it as of little relevance and prefer a fit with fewer parameters.

  • apparent noise in the exp spectrum will raise the rms-fit value.

    back to auxiliary programs
    back to index