In this try, I would like to see how time_to_take_each next measurements are affected with preconditioning mass parameter(this is used to generate rational approximation file)
Preconditioning mass is used to optimize the speed up of generating successive measurements.
- input1_for_precoditioningmass.dat
- input2_for_precoditioningmass.dat
- input3_for_precoditioningmass.dat
- input4_for_precoditioningmass.dat
Objective is to find how the consecutive time to write gauge configuration (using RHMC) changes with preconditioning mass.
| Time (TTRAJ) for trajectory without IO | preconditioning mass | TTRAJ before equilibrium |
|---|---|---|
| 92.175s | 0.0089026 | 85.83s |
| 4min 36.735s | 0.0150824 (sqrt(m_l*m_s)) | 80.994s |
| 3min 35.697s | 0.0406366 | 62.367s |
| 2min 8.755s | 0.0783706 (m_s) | 53.474s |
- It seems like the Mean of observables except plaquette depends on preconditioning mass. Analysis for individual nodes has started:
File names Mean_CHI_UD Std deviation Jackknife correlation time
try14_3.txt 0.013398 0.001877 0.000564 6
try14_0.txt 0.060803 0.006816 0.002261 10
try14_2.txt 0.023729 0.003117 0.000831 2
combined.txt 0.036159 0.019003 0.016457 225
try14_1.txt 0.058389 0.004554 0.000929 1
File names Mean_CHI_S Std deviation Jackknife correlation time
try14_3.txt 0.199118 0.002388 0.000717 16
try14_0.txt 0.227148 0.003808 0.001263 19
try14_2.txt 0.211872 0.002507 0.000669 12
combined.txt 0.214412 0.011292 0.009779 232
try14_1.txt 0.226985 0.001725 0.000352 5
File names Mean_Plaquett Std deviation Jackknife correlation time
try14_3.txt 0.534419 0.000169 5.1e-05 2
try14_0.txt 0.532475 0.000378 0.000125 6
try14_2.txt 0.533355 0.000315 8.4e-05 4
combined.txt 0.533333 0.000767 0.000664 182
try14_1.txt 0.532622 0.000449 9.2e-05 4
File names Mean_Delta_H Std deviation Jackknife correlation time
try14_3.txt -0.054457 0.126774 0.038091 1
try14_0.txt -0.006931 0.066023 0.021898 1
try14_2.txt -0.028758 0.122742 0.032731 1
combined.txt -0.012666 0.024772 0.021453 1
try14_1.txt 0.054631 0.302814 0.061762 1
- We need to set m_s (mass of strange quark) as preconditioning mass, as it turns out that the program ratApprox is not utilising the preconditioning mass concept. If we use preconditioning mass then we shall have to find product of {psf1}{psf2}{psf3} and hence, we shall require 3x3=9 such functions(3 functions are required to calculated for rhmc, which are HB, acceptance rate and molecular dynamics, for each of these 3 functions we need to find rational approximation ; i.e. to represent each of these in form of p(x)/q(x), and thus we need 9 such rational approximation(if we use m_prec), with each rational approximation we require 2 set of coefficients (1 for denominator and 1 for numerator; which can be of degree 14 or 12)) .but with use of preconditioning mass we see that output.rat just finds 6x2=12 total_functions (for numerator and denominator of 6 functions), instead of 9 functions.
- follow try15 for further progress.
- The values of chi_ud, chi_s shall further be devided from factor of 4 as defined in definations
- find more papers like this: