Up to now there does not exist a decent documentation/tutorial for running MaxEnt. I attach my Python skript which does the job for me quite well (based on svn revision 6727). It is taylored to analytically continue imaginary time (fermionic) Green's function from the hybridization QMC (see applications/dmft/QMC), but should be easy to adopt to other QMC flavors. The type of physical system is selected with the KERNEL option, which by default is set to fermionic. If one wants to deal with bosons, set it to bosonic. The OMEGA_MAX should be set to cover the frequency range where A(w) has structure, and for bosonic kernels one should in addition OMEGA_MIN=0 (I believe). If the norm of the spectrum is different from 1 (fermionic Green's function), the parameter NORM should be set accordingly. I suggest to start with data with error bars, because the covariance matrix provided by ALPS seems to have some problems (negative eigenvalues, for example). Otherwise, the advices given in the documentation apply for choosing for example the alpha values. I do provide the script "as is" and as a rough guideline on how to use the MaxEnt. I do not want to invest too much into helping to make it run, especially when there problems with the data from simulations and so on. In the course of time I intend to provide a tutorial, but this may take quite a while. Have fun and good luck Thomas Pruschke
-- Prof. Dr. Thomas Pruschke Institute for Theoretical Physics Friedrich-Hund-Platz 1 37077 Goettingen Germany Phone +49 551 39 7683 FAX +49 551 39 9263
Am 01.02.2013 um 17:50 schrieb Viveka Singh vivekasps@gmail.com:
Dear all, I need to do the analytic continuation of my imaginary frequency / imaginary time data to the real frequency. I want to use the ALPS MaxEnt code for that, but I am not able to find an elaborate documentation on how to use the maxent code. I have looked at the "hybdoc.pdf", but not enough is being mentioned there and it is pointed there that a tutorial for the ALPS maxent code will be provided elsewhere. Can someone send me that link?
Or can someone describe the input parameters that are required for the maxent code like - what are ALPHA's and how do one decide upon FREQUENCY_GRID? Do you need to spacify the OMEGA_MAX/OMEGA_MIN, even if someone wants to do the analytic continuation of imaginary time data?
With regards, Viveka Nand Singh