According to a well-known effect, the chi_2 minimization returns fitted values that are systematically shifted with respect to the data point when the full covariance matrix is used in the fit. So i want to to know how to fit data with only diagonal errors for MINUIT? More detailed, please. Thanks.

Hi,

I am not sure I have understood you. Which covariance matrix are you referring to ? For the data points ?

In that case the default fit in ROOT uses always uncorrelated errors (i.e. diagonal covariance matrix for the points)

Best Regards

Lorenzo

i use Minuit based on RooFit to get the minimum s=-lnL, L is the likelihood value. the full convariance matrix is referred to the data point, which is default to do the fit. But i want to use the diagonal convariance matrix to do the fit, could you help me? Thanks!

[quote=“moneta”]Hi,

I am not sure I have understood you. Which covariance matrix are you referring to ? For the data points ?

In that case the default fit in ROOT uses always uncorrelated errors (i.e. diagonal covariance matrix for the points)

Best Regards

Lorenzo[/quote]

Hi,

By default in ROOT and RooFit we use always uncorrelated errors for the data points. For correlated data points, you need to write your-self the likelihood function.

Lorenzo