The RoboCup simulator system is composed of:
- A soccer server, a soccer monitor, and the individual player programs.
- A soccer server and all the individual player programs.
- A world model, a soccer server, and all the individual player programs.
- A world model, a soccer server, a soccer monitor, a coach server, and all the individual player programs.
- A soccer server, and a player server.
The soccer server communicates with the players using:
In the RoboCup soccerserver when a player kicks a ball that is
already moving, what happens? Assume that the ball is in the
kickable area. Pick best.
- The vector representing the kick is added to the the velocity vector representing the ball's velocity.
- The ball is moved one step forward in the direction of the kick.
- The ball is moved to a random location constrained by the vector of the kick.
- The vector representing the kick is assigned to the ball.
- The player's facing direction is added to the ball's velocity vector.
In the RoboCup soccerserver the player gets information about
the world in a series of see
commands. These
commands contain:
- The position of the things the agent sees relative to its current position and facing direction.
- The position of the things the agent sees in absolute coordinates.
- The player's stamina as well as the position of the things the player sees.
- The position of the ball in relative coordinates to the player.
- The position of the other players in relative coordinates to the player.
The Biter architecture uses the concept of independent
activities (or behaviors). The most important member functions
for these activities are:
canHandle
, handle
, inhibits
override
, handle
belief
, desires
, intentions
act
extend
, provide
, eliminate
The main contribution of Biter's WorldModel
is
that it provides:
- The absolute coordinates of everything the agent sees.
- The location of every object on the field.
- The time of the last movement in the field.
- The relative location of every object on the field, with respect to the agent.
- The location of every fixed object on the field.
N worker ants can carry a weight proportional to:
The ants' Long Range Recruitment strategy consists of:
- The ants going back to the nest while secreting a chemical trail.
- The ants secreting a chemical agent which disperses in the air.
- The ants going back to the nest and recruiting other nestmates there.
- The ants secreting a pheromone which causes other nearby ants to secret the same pheromone and so on.
- The ants running around in ever-expanding circles.
Stigmergy is
- The coordination of activities through indirect interactions.
- The coordination of agents via protocols.
- The use of reactive agents to achieve coordination.
- The use of communications to achieve coordination.
- The implementation of coordination in the presence of selfish agents.
In the initial experiments with robotic box-pushing the box
would sometimes not move. This problem was solved by:
- Having the robots detect stagnation and randomly re-allocate themselves.
- Removing the nails that held the box firmly to the floor.
- Adding more robots.
- Using stigmergy on the robots.
- Starting the robots in random locations and adding a behavior that made them repel each other.
The task model used by the robots in the box-pushing experiments most resembles:
- A subsumption architecture.
- A BDI architecture.
- The Biter architecture.
- A goal-driven architecture.
- A finite-state machine.
Game theory was first introduced by
- John Von Neumman
- John Nash
- Oskar Morgenstern
- Richard Dawkins
- Alan Turing
The fact that I know that you know that I know (ad infinitum)
about P, is referred to as:
- Common Knowledge
- Nash equilibrium
- Infinite Knowledge
- Shared Knowledge
- Social Equilibrium
In game theory, an individually rational agent acts so as to
- Maximize the utility it receives
- Maximize the utility it receives and minimize the one the opponent receives.
- Minimize its losses.
- Maximize the difference between the utility it receives and the utility the opponent receives.
- Maximize the sum of its utility and those of the other agents its cooperating with.
A strategy S is said to be Pareto optimal if
- There is no other strategy S' where one agent is better off in S' and no agent is worse off in S' than in S.
- It is the one that maximizes the sum of everyone' payoffs.
- For all agents i, S(i) is i's best strategy given that all the other players will play the strategies in S.
- No matter what all the other players do, all players want to play S.
- All actions are chosen with a probability of 1.
A strategy S is said to be a Nash equilibrium if
- There is no other strategy S' where one agent is better off in S' and no agent is worse off in S' than in S.
- It is the one that maximizes the sum of everyone's payoffs.
- For all agents i, S(i) is i's best strategy given that all the other players will play the strategies in S.
- No matter what all the other players do, all players want to play S.
- All actions are chosen with a probability of 1.
A strategy S is said to be the social welfare strategy if:
- There is no other strategy S' where one agent is better off in S' and no agent is worse off in S' than in S.
- It is the one that maximizes the sum of everyone's payoffs.
- For all agents i, S(i) is i's best strategy given that all the other players will play the strategies in S.
- No matter what all the other players do, all players want to play S.
- All actions are chosen with a probability of 1.
The Prisoner's Dilemma is a dilemma because:
- The Nash equilibrium is not the same as the Pareto Optimal solution.
- Moral questions are always a dilemma.
- The agents are forced to act selfishly.
- The agents are forced to act selflessly.
- It has two Nash equilibriums.
What is the Nash Equilibrium of this game matrix?
|
Alice |
Bob |
|
A |
B |
A |
9,16 |
0,0 |
B |
6,6 |
14,8 |
- (A,A) and (B,B)
- (A,A)
- (A,B) and (B,A)
- (B,B)
- (B,A)
What is the Pareto Optimal equilibrium of this game matrix?
|
Alice |
Bob |
|
A |
B |
A |
9,16 |
0,0 |
B |
6,6 |
14,8 |
- (A,A) and (B,B)
- (A,A)
- (A,B) and (B,A)
- (B,B)
- (B,A)
What is the social welfare solution of this game matrix?
|
Alice |
Bob |
|
A |
B |
A |
1,16 |
0,0 |
B |
6,6 |
8,8 |
- (A,A) and (B,B)
- (A,A)
- (A,B) and (B,A)
- (B,B)
- (B,A)
The Borda count seems more fair because
- It obeys reflectional and rotational symmetry.
- It gives everyone only one vote.
- It allows minorities to be equally represented.
- It never results in a tie.
- It does not require individuals to order their preferences.
Arrow's Impossibility Theorem tells us that
- There cannot exist a voting mechanism that can satisfy the 6 intuitive rules of fairness.
- There cannot exist a voting mechanism that is better than the Borda count.
- There will always be a game matrix for which no solution can be found.
- There will always be a set of preferences for which no appropriate voting mechanism exists.
- Any voting mechanism invented that can be invented will always neglect the minority vote.
The Clarke Tax is
- A method that encourages voters to vote their true preference.
- A method that distributes the blame for a decision evenly among the players.
- A mechanism which will find the equilibrium solution in a voting scenario.
- A method for determining the true utility of an agent.
- A method for generating revenue that is proportional to the needs of the society.
There is no equilibrium solution in voting.
A common value action is one where
- An agent's value depends entirely on other agents value.
- The value of the good depends only on the agent's own preference.
- The value depends partly on the agent's preference and partly on others.
- The goods sold are well-known to all the participants.
- An action where the agents do not know the value of the good.
An English auction is a:
- First-price open-cry auction.
- First-price sealed-bid auction.
- Second-price open-cry auction.
- First-price common-value auction.
- First-price share-value auction.
There is no "shared value".
In a private value auction with risk-neutral bidders, which of the following mechanism will get the seller more money.
- They are all the same.
- English auction.
- First-price sealed-bid auction.
- Vickrey auction.
- Dutch auction.
In a correlated value auction with risk-neutral bidders, which of the following mechanism will get the seller more money.
- They are all the same.
- English auction.
- First-price sealed-bid auction.
- Vickrey auction.
- Dutch auction.
Which types of auctions self-enforce bidder collusion deals?
- English and Vickrey
- English
- Vickrey
- First-Price sealed-bid
- Dutch
The reason the Vickrey auction is not used by humans much is that:
- The dominant strategy is to reveal one's true preference.
- It is too complicated.
- It generates less revenue for the buyers.
- The dominant strategy is too complicated to calculate for humans.
- There is no dominant strategy.
Inefficient allocation arises in interrelated auctions when
- Items are auctioned in sequence.
- The players cheat.
- The players do not use the Borda count.
- The auctions are private value.
- The auctions are correlated value.
In the TAC competition there were two main strategies used by
the players, they were:
- Optimal allocation of all goods and treating each customer individually.
- Heuristic search of the space and the use of learning algorithms.
- Game theoretic and Economic-based systems.
- Selfish and Cooperative.
- Bid from the beginning and waiting until the end.
The winner's curse can only happen if the auction is
- Common or correlated value.
- Private value.
- English.
- Vickrey.
- Dutch.
In Vickrey auction with private value a risk-neutral agent could
be encouraged to counterspeculate if
- It is uncertain about its own valuation.
- It will never be encouraged to speculate.
- The agent has lied in the past.
- There is no Nash equilibrium.
- There are many Nash equilibriums.
In Axiomatic Bargaining Theory
- No equilibrium is assumed.
- An equilibrium concept is used.
- There is no way to describe a good solution.
- There is no such thing as Axiomatic Bargaining Theory
- The first player always gets a better deal.
If two players are trying to split $100 using the Nash
Bargaining solution. Player 1 has utility U1 = X,
where X is the amount of money it gets. Player 2 has
U2 = 2*Y, where Y is what 2 gets. The fallback deal
has zero utility for both. The deal that the Nash Bargaining
solution finds will be the one that maximizes (where X is what
player 1 gets).
- 2*X*(1-X)
- X2
- X/2
- X*(1-X)
- X/(1-X)
Two agents are engaged in bargaining for a good of initial value
1 but whose value is decreased by multiplying it by .9 at each
time. The game can be summarized with the table below (as seen
in class). If the final time is t=5, what will agent 1 get if a
deal is struck at time 2 (in absolute terms, not in
percentage of the total value)?
Round |
1's share |
2's share |
Total value |
Offerer |
t-3 |
.819 |
.181 |
.9t-4 |
2 |
t-2 |
.91 |
.09 |
.9t-3 |
1 |
t-1 |
.9 |
.1 |
.9t-2 |
2 |
t |
1 |
0 |
.9t-1 |
1 |
Offerer's maximal acceptable claims in a finite game.
The task allocation problem consists of a set of tasks that must
be allocated to a set of agents where:
- There is a cost associated with doing every subset of tasks.
- No agent can do more than one task.
- The agents are constrained in the order in which they carry out the tasks.
- There is no possible allocation where all the agents are happy.
- Some tasks depend on other tasks.
OCSM contracts allow contract net to, eventually, find the
optimal task allocation because:
- They allow agents to go from any task allocation to any other task allocation in one step.
- They sound very complicated and, therefore, must be magical.
- They align the agents' interests with the interest of the community.
- They allow agents to cheat.
- They encourage agents to choose the Nash equilibrium.
In a coalition formation problem with A agents, we can bound the
utility of the best possible coalition structure to within
A*S', where S' is the best one we have found so far by:
- Searching all coalition structures of one or two sub-coalitions.
- Searching all coalition structures of A or A-1 sub-coalitions.
- Searching all coalitions that involve more than two agents.
- Searching all coalitions that involve two or fewer agents.
- It cannot be done.