September 2008 - Comp-phys-alps-users

Controlling the amount of output
by João Manuel Viana Parente Lopes 25 Sep '08

25 Sep '08

Dear Alps users, I'm running a simulation with loop application and an 1D Chain of Heisenberg spins. I'm interested in the correlation study between two of these spins as a function of the temperature and the interaction parameters. I need to run the simulation to 300 parameters (temperature + parameters) I really need to write in the parameter file 300 lines? Or is it possible to do it in another way? After I ran the simulation the param.task*.out.xml files have different sizes: . Files with 2.3 M: <SCALAR_AVERAGE name="Energy^2"> <COUNT>1007616</COUNT> <MEAN method="jackknife">445.73115</MEAN> <ERROR converged="yes" method="jackknife">0.00979</ERROR> <VARIANCE method="simple">58.8</VARIANCE> <AUTOCORR method="jackknife">0.321</AUTOCORR> </SCALAR_AVERAGE> . Files with 20 M <SCALAR_AVERAGE name="Energy^2"> <COUNT>1004771</COUNT> <MEAN method="simple">444.9879</MEAN> <ERROR converged="yes" method="binning">0.0233</ERROR> <VARIANCE method="simple">292</VARIANCE> <AUTOCORR method="binning">0.43</AUTOCORR> <BINNED size="1"><COUNT>1004771</COUNT><MEAN method="simple">444.9879</MEAN><ERROR method="simple">0.017</ERROR></ BINNED> <BINNED size="2"><COUNT>502385</COUNT><MEAN method="simple">444.98789</MEAN><ERROR method="simple">0.0191</ ERROR></BINNED> <BINNED size="4"><COUNT>251192</COUNT><MEAN method="simple">444.98793</MEAN><ERROR method="simple">0.021</ERROR></ BINNED> <BINNED size="8"><COUNT>125596</COUNT><MEAN method="simple">444.98793</MEAN><ERROR method="simple">0.0222</ ERROR></BINNED> <BINNED size="16"><COUNT>62798</COUNT><MEAN method="simple">444.98793</MEAN><ERROR method="simple">0.0227</ ERROR></BINNED> <BINNED size="32"><COUNT>31399</COUNT><MEAN method="simple">444.98793</MEAN><ERROR method="simple">0.023</ERROR></ BINNED> <BINNED size="64"><COUNT>15699</COUNT><MEAN method="simple">444.98792</MEAN><ERROR method="simple">0.023</ERROR></ BINNED> <BINNED size="128"><COUNT>7849</COUNT><MEAN method="simple">444.98769</MEAN><ERROR method="simple">0.0232</ ERROR></BINNED> <BINNED size="256"><COUNT>3924</COUNT><MEAN method="simple">444.98744</MEAN><ERROR method="simple">0.0232</ ERROR></BINNED> <BINNED size="512"><COUNT>1962</COUNT><MEAN method="simple">444.98744</MEAN><ERROR method="simple">0.0232</ ERROR></BINNED> <BINNED size="1024"><COUNT>981</COUNT><MEAN method="simple">444.98744</MEAN><ERROR method="simple">0.0234</ ERROR></BINNED> <BINNED size="2048"><COUNT>490</COUNT><MEAN method="simple">444.98712</MEAN><ERROR method="simple">0.0234</ ERROR></BINNED> <BINNED size="4096"><COUNT>245</COUNT><MEAN method="simple">444.98712</MEAN><ERROR method="simple">0.0233</ ERROR></BINNED> </SCALAR_AVERAGE> Is it normal to have these diversity of output? Can I choose any option to just have the first kind of files? Is there any way to select that I just want the correlation between two given spins? The output of "Spin Correlations" is the average of the whole component: Sx_i Sx_j + Sy_i Sy_j + Sz_i Sz_j or just the Sz_i Sz_j ? Thanks in advance for your kind support João Lopes

2 3

specific heat measurements
by Brian Keith 11 Sep '08

11 Sep '08

Hello, I looked at the alps tutorial pages and it says that the looper code calculates specific heat. I ran the looper code from alps version 1.3.3. I looked at the task.out files and specific heat was not listed anywhere. Am I missing something? -thanks, Brian

3 2

Finding the site at a particular set of coordinates
by Miles Stoudenmire 11 Sep '08

11 Sep '08

Hi - I'm trying to write a custom measurement in the classical spin mc code, and I'd like to be able to have the alps lattice library tell me the site that is at a given location relative to another site. More specifically, I'd like to start with pair of sites, indices i and j, then compute the vector v=(r_j-r_i) (this I already know how to do using the coordinates function). Then, I'd like to go to some other site, k, and figure out the index, l, of the site at r_k + v (i.e. I want (r_l-r_k) = (r_j-r_i)). Is there a general way to do this using the alps lattice library or will I have to write special code for the particular lattice that I'm using? Thanks! -- -=Miles Stoudenmire=- miles(a)physics.ucsb.edu miles.stoudenmire(a)gmail.com http://www.crazytheory.com/

1 0

Wang-Landau method for qm spin system
by Fritz Koermann 07 Sep '08

07 Sep '08

Hi Matthias, dear alps community, is it possible to use the qmc wang-landau method for the qm Heisenberg model as implemented in alps also in combination with more than nearest-neighbour exchange J (still unfrustrated lattice)? Thanks for the wonderful support, best, Fritz --- Max-Planck-Institut für Eisenforschung GmbH Max-Planck-Straße 1 D-40237 Düsseldorf Handelsregister B 2533 Amtsgericht Düsseldorf Geschäftsführung Prof. Dr. Jörg Neugebauer Prof. Dr. Anke R. Pyzalla Prof. Dr. Dierk Raabe Prof. Dr. Martin Stratmann Dipl.-Kfm. Herbert Wilk Ust.-Id.-Nr.: DE 11 93 58 514 Steuernummer: 105 5891 1000 -------------------------------------------------

1 0

Wang-Landau method for classical spin system
by Fritz Koermann 04 Sep '08

04 Sep '08

Dear alps users, is the Wang-Landau method also available within alps for the classical monte carlo package or is it planned to be implemented? Best, Fritz --- Max-Planck-Institut für Eisenforschung GmbH Max-Planck-Straße 1 D-40237 Düsseldorf Handelsregister B 2533 Amtsgericht Düsseldorf Geschäftsführung Prof. Dr. Jörg Neugebauer Prof. Dr. Anke R. Pyzalla Prof. Dr. Dierk Raabe Prof. Dr. Martin Stratmann Dipl.-Kfm. Herbert Wilk Ust.-Id.-Nr.: DE 11 93 58 514 Steuernummer: 105 5891 1000 -------------------------------------------------

2 2

Re: [ALPS-users] Comp-phys-alps-users Digest, Vol 29, Issue 8
by João Manuel Viana Parente Lopes 03 Sep '08

03 Sep '08

Dear Matthias Troyer and Synge Todo, I'm sorry for the late response but I was away from my computer several days! Thank's for your help!! I increased the number of SWEEPS and I have finally found the square root behaviour of the error bars as a function of the number of SWEEPS. I imagine (I didn't had time to check yet) the method of estimation of the error bars makes an underestimation of the variance. Thanks again for your time João Lopes

2 1