#-------------------------------------------------------Follow the instructions to simulate results----------------------------------------

1. First, run .ipynb file "theta_1.ipynb" to set candidate post change parameter for different lengths of the graph. For a fixed length of the graph, post change parameter (unknown to algorithms/change detectors) is kept same (changes happen in the middle of the graph) for all the algorithms to evaluate their performances. This .ipynb file stores candidate post change parameter as "theta_1_list.npy" file.

2. Second, run .ipynb files "Epsilon_Greedy_Detector.ipynb", "Epsilon_Greedy_Detector_Theory.ipynb", "URS_Detector.ipynb", and "Oracle_Detector.ipynb" to get stopping time statistics of Epsilon-Greeedy change detector plus($\epsilon$-GCD+),Epsilon-Greeedy change detector($\epsilon$-GCD),URS change detector, and Oracle change detector respectively. 


2.After successfully running the three aforementioned .ipynb file, run "graph.ipynb" to generate histogram of stopping time, mean and standard deviation of detection delay of Epsilon-Greedy, URS, and Oracle change detectors for fixed changepoint and graph size. 
