This page contains the benchmark data sets used by Ann Campbell and Barrett Thomas in their paper "Models for the Probabilistic Traveling Salesman Problem with Deadlines ." The paper was submitted for publication in the fall of 2006 . The datasets are derived from the sets first presented in Dumas et al (1995)1. The tar file contains two different instances for each of the n=20, 40, 60, and 100, time window width 20 sets for the Dumas sets. The "_mixed" sets have the probability settings of either 0.1 or 1 for each customer. The sets without "_mixed" have the range of probabilities described in the paper. To get the homogeneous sets of 0.1 and 0.9, simply hard code the appropriate probabilites using either data the "_mixed" or not "_mixed." You can similarly generate the "earlier" and "later" deadlines by selecting the appropriate deadline in the datasets when reading in the datafiles.