spiemonte / leonardym Goto Github PK

Hi @spiemonte, i have a problem.

When running the program compiled with NUMCOLORS=3, Landau Gauge Fixing converges well and produces correct result.

SU(3) HMC

LandauGaugeFixing::Maximal functional at step 2970: 11716.7
LandauGaugeFixing:: convergence: 4.46612e-11
LandauGaugeFixing::Final deviation from the Landau gauge: 3.40372e-30
Delta energy in metropolis: -3.14283 - accepted!

In other cases (N=2,4,5) Landau Gauge Fixing seems to work incorrectly.

SU(2) HMC

LandauGaugeFixing::Maximal functional at step 0: 3.58455e+54
LandauGaugeFixing::Maximal functional at step 30: 3.58455e+54
LandauGaugeFixing:: convergence: 0
LandauGaugeFixing::Final deviation from the Landau gauge: 9.12747e+103
Delta energy in metropolis: nan - rejected!

SU(4) HMC

LandauGaugeFixing::Maximal functional at step 0: 3.14898e+06
LandauGaugeFixing::Maximal functional at step 30: 3.14898e+06
LandauGaugeFixing:: convergence: 0
LandauGaugeFixing::Final deviation from the Landau gauge: 38858
Delta energy in metropolis: 5.32543e+31 - rejected!

SU(5) HMC

LandauGaugeFixing::Maximal functional at step 0: 710780
LandauGaugeFixing::Maximal functional at step 30: 710780
LandauGaugeFixing:: convergence: 0
LandauGaugeFixing::Final deviation from the Landau gauge: 19740.5
Delta energy in metropolis: 1.32531e+26 - rejected!

SU(2) CM

However, if instead of the PureGaugeHMC you enable PureGaugeCM, then deviation from Landau Gauge is ok...
LandauGaugeFixing::Maximal functional at step 0: 9554.54
LandauGaugeFixing::Maximal functional at step 30: 9554.54
LandauGaugeFixing:: convergence: 0
LandauGaugeFixing::Final deviation from the Landau gauge: 4.1916e-17

All runs used the same configuration file - pure.txt

logSU2GaugeFixCM.txt
logSU2GaugeFixHMC.txt
logSU3GaugeFixHMC.txt
logSU4GaugeFixHMC.txt
logSU5GaugeFixHMC.txt
pure.txt

Such a question for you, @spiemonte. Will your code be updated in the future or is this already the final version of the code? Is it possible to add Yukawa interactions in the future? Your code is pretty well organized, and it doesn't seem like a difficult task to me. What parts of the code mostly need modification? What ideas do you have on this?

Hi @spiemonte.
I explore gauge-fixed propagators of gauge bosons of SU(5) theory, i have implemented unitary gauge in scalar action to explicitly break the gauge symmetry ( Φ' = |Φ|λ_24, |Φ| is higgs length and λ_24 is Hypercharge generator of SU(5) ), and suitable propagators. But no matter what they remain massless(for any beta or mass of higgs).
and higgs of the fundamental representation in unitary gauge, on the contrary, works as it should.
Increasing the number of adjoint scalar fields (flavors) doesn't help either.
Are you sure that the scalar action implemented correctly?

@spiemonte Hello. Re-opening the topic, because the old one has stagnated. I've been waiting for a response from you for two months, do you even read your github? But the problem didn't go away.
I don't understand why everything works for SU(3) but not for N>3, although over relaxation algorithm seems to be correct. Where is the catch?

@daschaich and me have been trying to get this code running, but we instantly get the following error after running the executable (from the included Makefile.mth) using the pure gauge configuration file:
$ ./leonardYM.exe --configfile=config_pure_gauge.cfg
terminate called after throwing an instance of 'boost::exception_detail::clone_impl<boost::exception_detail::error_info_injectorstd::logic_error >'
what(): character conversion failed
Aborted (core dumped)

The main modification that we have done (https://github.com/FelixSpr/LeonardYM/tree/master), was to rewrite the path to the Boost libraries explicitly in the Makefile (.mth and .mpi) in line 19 to enable compilation:

• $(CPP) $(CPPFLAGS) $(OBJECTS) -o ./build/leonardYM.exe -lboost_program_options

+$(CPP) $(CPPFLAGS) $(OBJECTS) -o ./build/leonardYM.exe -L/opt/gridware/depots/e2b91392/el7/pkg/libs/boost/1.60.0/gcc-5.5.0+openmpi-1.10.7+python-2.7.8/lib -lboost_program_options

We are using version 3.3.9 for Eigen and version 1.60.0 for Boost.

Do you have an idea what the problem could be and how we can fix it?
Should we use a different version of Eigen and/or Boost?