Cut data when equilibrium begins

I am simulating the traveling samlsman problem(TSP) whit Metropolis (500000 sweeps) and i want to cut my data when it at the "equilibrium"

[Geoff Hayes]

Perhaps elaborate on what you mean by cut my data when it is at the "equilibrium". Do you just want to exclude all data that is beyond a certain number of sweeps? Or when the tour length no longer varies so much?

[Kevin Grønberg Poulsen]

I am sorry for openness of the question. To specify the question i am not interested in the large fall in tour length in the start of the data, but only in the part where it fluctuates around a specific tour length in this case 100, i want to cut off start data and save the part where it fluctuates around specific tour length.

I hope this more specific.

[Geoff Hayes]

If tourLength is the vector of all tour lengths, then you could just find within it all indices for elements that are within a certain window around 100:

 wdwLen = 10;
 idcs   = find(tourLength>(100-wdwLen) & tourLength<(100+wdwLen));
 data   = tourLength(idcs);

data are all the tour lengths that fluctuate around 100 (within 10 units). The catch here is that idcs would not have any index that refers to an element whose tour length is just outside of the window. So data will not be complete in the sense that it may be missing some tour lengths… A check for missing data could be

 if (idcs(end)-idcs(1)+1) ~= length(idcs)
   % then at least one element within the start and end of idcd 
   % falls outside of our window around 100
 end