Dear Wei,
I think I found the bug. It is fixed in the latest /trunk version. Be aware that I’m not 100% of the workaround, I’m currently checking it with my collaborators.. But it seems to work.
Best, Michele
-- ETH Zurich Michele Dolfi Institute for Theoretical Physics HIT G 32.4 Wolfgang-Pauli-Str. 27 8093 Zurich Switzerland
dolfim@phys.ethz.ch www.itp.phys.ethz.ch
+41 44 633 78 56 phone +41 44 633 11 15 fax
On Jul 6, 2015, at 3:20 PM, Wei.Su xichuang_s@163.com wrote:
I was using MPS code in ALPS to calculate exited states of the fermion Hubbard model when there throwed an error
Sweep 0, optimizing sites 19 and 20 Input <MPS|O[0]> : 0 Output <MPS|O[0]> : -nan JCD used 8 iterations. Time elapsed in JCD: 0.19799719 MPS overlap: 0 Energy 1 2.341151156e-310 Exception thrown: Error in SVD!
This error vanished when the "total number" was less than 10 or great than 20 (or around) or the 'MAXSTATES' is greater than about 80.
In the attachment I deliver my python script.
import pyalps import numpy as np
parms=[] for L in [40]: for P in [12]: # for P in [55,56,57,58,59,60,61,62]: parms.append({ 'LATTICE' : "open chain lattice", 'L' : L, 'MODEL' : "fermion Hubbard", 'CONSERVED_QUANTUMNUMBERS' : 'Nup,Ndown', 'Nup_total' : P, 'Ndown_total' : P, 'mu' : 0.0, 't' : 1.0, 'U' : 0.0, 'SWEEPS' : 4, 'MAXSTATES' : 50, 'NUMBER_EIGENVALUES' : 2, 'storagedir' : 'storage', })
#write the input file and run the simulation input_file = pyalps.writeInputFiles('parm_tsc',parms) res = pyalps.runApplication('mps_optim',input_file,writexml=True)
-- -------苏威-------
<fermionhubbard.py>