Calculating a turnover frequency¶
The turnover frequency (TOF) can be calculated from the number of times a product is formed per site and second. The same procedure can be followed for the individual elementary step rates. The script analyze_tof.py provides a complete example of how the TOF can be calculated.
Assume that the event list is instantiated as
from user_events import A, B, X, Z
events = [A, B, X, Z]
Where X is the forward reaction of a step that generates the product molecule, and Z is the reverse reaction that consumes one product molecule. To calculate the system’s overall TOF, we load in the time, and events that were executed (see output)
import numpy as np
evs_exec = np.loadtxt("evs_exec.txt")
time = np.loadtxt("time.txt")
stypes = np.loadtxt("stypes.txt") # for number of sites here
Nsites = len(stypes)
TOF_global = (evs_exec[2]-evs_exec[3])/(time[-1]-time[0])/float(Nsites)
We may want a TOF for each type of site and to discard the first half of the simulation, which may be out of steady-state:
Nevents = 4
sid_ev = np.loadtxt("sid_ev.txt").reshape(-1, Nsites, Nevents)
Nhalf = int(np.round(len(sid_ev)/2.,0))
sid_ev = sid_ev[Nhalf:]
dt = time[-1]-time[int(np.round(len(time)/2.,0))]
tofs_st = np.zeros(len(list(set(stypes))))
for time_chunk in sid_ev:
for n, st in enumerate(list(set(stypes))):
sids_st = [i for i, s in enumerate(stypes) if s == st] # site-indices of stype==st.
Nst = float(len(sids_st)) # number of sites with the current stype == st.
tofs_st[n] += sum([time_chunk[i][2] - time_chunk[i][3]
for i in sids_st]) / (dt*Nst) # TOF of the stype
Here, the last half of the simulation is used, and for all unique types of sites the indices are noted. Then the number of sites with the current stype is noted, and finally the TOF is calculated.
Statistical averaging should be done to address the convergence of the TOF. This can be done by running multiple identical simulations in parallel.