Assignment 1
Game Theory
170010033
Swapnik Jagarlapudi
Q1) a)Suppose that each person cares only about her own comfort.
Model the situation as a strategic game. Find its Nash equilibrium
(equilibria). Is it identical to Prisoner’s Dilemma?
P1\P2 Sit Stand
Sit 5,5 10,1
Stand 1,10 1,1
This game is identical to Prisoner’s Dilemma. The Nash equilibrium is the
state where both of them sit, which is also the Pareto optimal outcome.
b) In this case, the payoffs will be reversed if both of them choose different
actions.
P1\P2 Sit Stand
Sit 5,5 0,10
Stand 10,0 1,1
The Nash equilibrium is the state where both of them stand, whereas the
Pareto optimal outcome is where both of them sit.
c) In the first case, both of them care only about themselves sitting, so both
of them are comfortable when both of them are sitting.
Q2) There are two animals fighting over prey. Each prefers to be
aggressive if the opponent is passive, passive if the opponent is
aggressive; given its own stance, it prefers the outcome where the
opponent is passive to that in which its opponent is aggressive.
Considering two states A (aggressive) and P (passive), we can write
payoffs for the first animal as follows:
u1(A,P) > u1(P,P)
u1(P,A) > u1(A,A)
u1(A,P) > u1(A,A)
u1(P,P) > u1(P,A)
Combining these inequalities, we get:
u1(A,P) > u1(P,P) > u1(P,A) > u1(A,A)
Assigning values 10, 7, 5, 2 respectively, we get the game table as follows:
A1\A2 Aggressive Passive
Aggressive 2,2 10,5
Passive 5,10 7,7
This game has 2 Nash equilibria, where both of the animals have different
actions, which are also the Pareto optimal states.
Q3) There cannot be a Nash equilibrium where more than k people
contribute. As the payoff is greater for people who have received the good
and not contributed, they will be inclined to not contribute in case k people
have already contributed. So there is a natural tendency to go away from
the state where more than k people contribute. So that cannot be a Nash
equilibrium. However, the state where k people contribute will be a Nash
equilibrium. This is because exactly k contributions are needed in order to
provide the good. Suppose one person decides not to contribute. Then, the
payoff of that person will reduce. If a person who decided not to contribute
changes his decision, then his payoff will also reduce. So there is a
tendency to remain in this state.
The state in which no person contributes is also a Nash equilibrium, since if
any person decides to contribute, they will have a loss in their payoff, as
the state where they neither contribute nor get the good has greater payoff
than the state where they contribute and don’t receive the good. Thus they
will have no tendency to deviate from this state. If there are less than k but
more than 0 contributors, then the contributors in this case will have an
inclination to change their state, since that way they can increase payoff.
Thus, this state is not a Nash equilibrium.
Q4) Suppose the sums announced by P1 is x and the sum announced by
P2 is y. Then we can represent this problem as follows.
If x + y - 10 = 0 and x = y, then u1(x,y) = u2(x,y) = 5.
If x + y - 10 < 0, then u1(x,y) = x, u2(x,y) = y
If x + y - 10 > 0 and x = y, then u1(x,y) = x, u2(x,y) = y
If x + y - 10 > 0 and x =/= y, if x < y then u1(x,y) = x and u2(x,y) = 10 - x and
if x > y, then u1(x,y) = 10 - y, u2(x,y) = y
Now we look at the best response for player 1.
Suppose y <= 5. Then, P1 will get maximum payoff for x = 10 - y.
=> BR1(y<=5) = 10-y
If y > 5, then P1 will have to choose a number x such that x - y = dx (i.e. the
difference should be miniscule). This means there are an infinite number of
possible values for x. Thus there is no best response. For both players, the
only intersection is the point (x = 5, y = 5). So this is the only state which is
a Nash equilibrium.
The best response graph for P1 will look as follows:
Q5) i) f(x1,x2) = 3x1x2, c(xi) = xi2
u1(x1,x2) = 3x1x2/2 - x12
du1(x1,x2)/dx1 = 0 => max(x1) = 3x2/4
Similarly, max(x2) = 3x1/4
At Nash equilibrium, max(x1) = max(x2) => x1 = x2 = 0
ii) f(x1,x2) = 4x1x2, c(xi) = xi
u1(x1,x2) = 2x1x2 - x1
du1(x1,x2)/dx1 = 2x2 - 1
So minimum of this function will be at x1 = 0, and maximum will be at x1 = 1
Similarly, for payoff of x2, minimum of this will be at x2 = 0, and maximum
will be at x2 = 1, and where x1 = ½ or x2 = ½.
So there are 3 Nash equilibria, x1 = x2 =
0, x1 =
x2 =
1, x1 =
x2 =
½
Q6)
P1/P2 A B C
A 5,5 1,5 1,1
B 5,1 7,2 2,7
C 1,1 2,2 7,0
