Dear all,

I have got a question about the calculation of the covariance matrix of a vector observable (RealVectorObservable type, for example):

Indeed I am able to compute the (jacknife-corrected) simple covariance of two given scalar observables in this simple way

--------------- string filename = "simulation.task1.out.h5";

mcdata<double> obs1;

mcdata<double> obs2;

obs1.load(filename, "simulation/results/" + alps::hdf5_name_encode("Observable1"));

obs2.load(filename, "simulation/results/" + alps::hdf5_name_encode("Observable2"));

cout << "Covariance: " << obs1.covariance(obs2) << "\n\n"; ---------------

However it is not entirely clear to me how to proceed when one instead has a single vector observable recorded in the hdf5 file and wants to compute to covariance matrix of its entries. In that case mcdata cannot be used any longer and digging into the code source I did not manage to find a method "covariance" implemented for other than mcdata-like scalar observables.

Is there a handy solution already implemented for that?

Thank you in advance!

Regards,

Giuseppe