Markov Chain States

[evanberkowitz]

In the state of the markov chain we need to keep

• states of RNGs
• field variables
• NSL::Tensor results of observables (like action, ...)
• Markov Chain metrics (see #27)
• anything else?

These states will be good for evaluating Actions (#30) and to be of any use at all require an I/O interface.

[Marcel-Rodekamp]

• Configuration - Fields
• Action Value - Type
• Weights - map<string,Type>
• state of master RNG - int,int,int
• Number of threads - int
• Number accepted configurations - int
• Number configurations - int
• min(1, exp(- Delta Re(H))) - double
• Markov Time = Number Configuration / Nsave_freq - int