Grade: 12/12

Algorithms

- All algorithms are sound and the extension meets the requirements of
  the project.

- The results are among the best seen in the project.

- An error is present in Algorithm 2, moreover this algorithm could be
  implemented more efficiently

- The heuristics 1 and 2 should be randomized. How are ties broken? 

- Improvements are possible in Addnfix, by considering pruning and look
  up tables for the delta computation.

- Pruning is not considered either in Rm1add2.

- Possibly a typing error is present in Algorithm 8

- Aspiration in Tabu Search implemented but not described. But it can be
  implemented differently, being sufficient that a new solution
  improves the overall best.

- Parameters not put in relation with the size of the instance.

Analysis:

- Computational Analysis is present where it is needed. It is also
  correct and in some cases an aggregate analysis is presented which is
  more precise.

- Although right, the average number of edges of which a vertex could be
  member in the instance is not explained. 
 
- No attempt to transform the data within the instances in order to
  remove problems due to different location and scale of each instance.

- Experimental analysis: the analysis in Figure 4 could have been made
  also for the construction heuristics and for the local search, where
  it makes more sense.

- The comments and conclusions reflect well the empirical data.


Report

- Generally well written and well structured report.

- Length within the limits. 

- Reference list and citations missing. At least one local search is
clearly borrowed by some other place.

- Missing a last run of spell checker


Code: 

- Working and results reproducible.

- Seemingly efficient.