This is a list of mistakes in the collection of solved exercises and exam samples.
It is based on the e-mails sent by other students; I have had no time yet to revise them.

A.2.4 (esercizio 2):

The graph lacks the constraint x1 <= 2, that requires to keep on the left of the vertical line
crossing the horizontal axis in x1 = 2. The optimal solutions are all the points of the segment
AB with A = (0,3) and B = (2,1).

B.2.6 (Exercise 3):
 For \epsilon_2 < -2, there is no feasible solution
 For \epsilon_2 = -2, the only feasible solution is A = (2,0)
 For -2 < \epsilon_2 < 0, the optimal solution gradually moves from A to B = (0,0)
 For \epsilon_2 >= 0, the optimal solution is always B


B.2.14 (Exercise 7):
The second row is dominated by the first row (not by the third one)


B.3.12 (Exercise 6):
The final table is correct, but unclear.
The four values reported are the conjoint probabilities \pi(y,\omega).
They allow to build the absolute probabilities of the experiment outcomes \pi(y) = [ 0.69 0.31 ]
and the conditional probabilities

\pi(\omega|y) = 63/69  6/69
                 7/31 24/31

which can be used to label the decision tree and obtain (by backward induction)
a benefit equal to  0.63 * 12 + 0.31 * 3 = 8.49 for the choice of performing the test.


B.4.12 (Exercise 6):
The value of the experiment is (0.5 * 66 + 0.5 * 62) - 52 = 11, not (66 + 62) - 2*52 = 22.


B.4.14 (Exercise 7):
After the row player has removed the third row, the second column dominated the first one (not the third one).
Tables b and c make no sense; in fact, there rows and the columns containing equilibria cannot be strictly dominated.