Title: The Price Of Stability for Network Design with Fair Cost Allocation
1The Price Of Stability for Network Design with
Fair Cost Allocation
- Elliot Anshelevich, Anirban Dasgupta, Jon
Kleinberg, - Eva Tardos, Tom Wexler, Tim Roughgarden
- Presented by Kobi Yablonka
- (most of the slides are taken from Elliot
Anshelevichs presentation The price of
Stability for Network Design
www.cs.princeton.edu/eanshele)
2 The context
- A network Design Game
- Number of independent agents
- Each agent try to minimize the cost
-
3The Model
- Directed garph G(V,E) , with each edge e having
a nonegative cost - Each player i has a set of terminal nodes
that he wants to connect - Strategy for player i is s.t
connects all node in - All players using an edge split up the cost of
the edge equally
4The Model cont.
- vector of the
players strategies - - the number of players using edge e
- The cost to player i is
- The total edge cost of the network is
- The cost to a player is affected from the
strategies of other players
5Nash Equilibrium
- A Nash Equilibrium (NE) is a set of payments for
players such that no player wants to deviate. - When considering deviations, player i assumes
that other player payments are fixed. - Given a solution consisting of a vector of
strategies S there is no strategy for
player i s.t
6Price of stability
- The best Nash equilibrium relative to the global
optimum - Stands in contrasts to the price of anarchy
which is the ratio of the worst Nash equilibrium
to the optimum
7Congestion Games
- This is a congestion game!
- Usual congestion games have latency/delay/load
- cost per player increases as the number of
players sharing an edge increases. - Fair Connection Game has edge costs
- cost per player decreases as the number of
players sharing an edge increases.
8Related Work
- Shapley value cost sharing
- Feigenbaum, Papadimitriou, Shenker
Herzog, Shenker, Estrin - Price of anarchy in routing and congestion games
- Roughgarden, Tardos
- Potential games
- Monderer, Shapley
9Example High Price of Stability
t
1
1
1
1
1
k
2
3
k-1
. . .
1?
1
2
3
k
k-1
0
0
0
0
0
10Example High Price of Stability
cost(OPT) 1e
t
1
1
1
1
1
k
2
3
k-1
. . .
1?
1
2
3
k
k-1
0
0
0
0
0
11Example High Price of Stability
cost(OPT) 1e but not a NE player k
pays (1e)/k, could pay 1/k
t
1
1
1
1
1
k
2
3
k-1
. . .
1?
1
2
3
k
k-1
0
0
0
0
0
12Example High Price of Stability
so player k would deviate
t
1
1
1
1
1
k
2
3
k-1
. . .
1?
1
2
3
k
k-1
0
0
0
0
0
13Example High Price of Stability
now player k-1 pays (1e)/(k-1),
could pay 1/(k-1)
t
1
1
1
1
1
k
2
3
k-1
. . .
1?
1
2
3
k
k-1
0
0
0
0
0
14Example High Price of Stability
so player k-1 deviates too
t
1
1
1
1
1
k
2
3
k-1
. . .
1?
1
2
3
k
k-1
0
0
0
0
0
15Example High Price of Stability
Continuing this process, all players defect.
This is a NE! (the only Nash) cost 1
t
1
1
1
1
1
k
2
3
k-1
. . .
1?
1
2
3
k
k-1
0
0
0
0
0
1 1
2 k
Price of Stability is Hk T(log k)!
16To Show
- The Hk Price of Stability is worst case possible.
- Proof uses the idea of a Potential Game
- Monderer and Shapley.
- Extend results to many natural generalizations of
the Fair Connection Game. -
17Potential Games
- A game is a potential game if there exists a
- function ?(S) mapping the current game state S
- to a real value s.t.
- If player i moves, is improvement change in
?(S). - Such games have pure NE just do Best Response!
- The Fair Connection Game is a potential game!
- We extend analysis to bound Price of Stability.
18A Potential Function
- Define ?e(S) ce1 1/2 1/3 1/ke
- where ke is players using e in S. Hk
- Let ?(S) S ?e(S)
- Consider some solution S (set of edges for each
player). - Suppose player i is unhappy and decides to
deviate. - What happens to ?(S)?
e ? S
19Tracking Player Happiness
- ?e(S) ce1 1/2 1/3 1/ke
- Suppose player is new path includes e.
- i pays ce/(ke1) to use e.
- ?e(S) increases by the same amount.
- Likewise, if player i leaves an edge e,
- ?e(S) exactly reflects the change in
is payment.
ce1 1/2 1/ke
e
i
e
ce1 1/2 1/ke
20Tracking Player Happiness
- ?e(S) ce1 1/2 1/3 1/ke
- Suppose player is new path includes e.
- i pays ce/(ke1) to use e.
- ?e(S) increases by the same amount.
- Likewise, if player i leaves an edge e,
- ?e(S) exactly reflects the change in
is payment.
ce1 1/2 1/kece/(ke1)
e
i
e
ce1 1/2 1/ke -ce/ke
21Bounding Price of Stability
- Consider starting from OPT (central optimum).
- From OPT, players will settle on some Nash NE.
22Bounding Price of Stability
- Consider starting from OPT (central optimum).
- From OPT, players will settle on some Nash NE.
- ?(NE) lt ?(OPT)
_
23Bounding Price of Stability
- Consider starting from OPT (central optimum).
- From OPT, players will settle on some Nash NE.
- ?(NE) lt ?(OPT)
- for any S,
- cost(S) lt ?(S) lt Hk cost(S).
- So cost(NE) lt ?(NE) lt ?(OPT) lt Hk cost(OPT).
_
_
_
_
_
_
24Extensions
- Take a fair connection game with each edge having
a nondecreasing concave cost function ce(x),where
x is the number of players using edge e. Then the
price of stability is at most Hk - The proof is analogous to the previous proof.
25Extensions
- All results hold if edges have capacities.
- Incorporate distance
- cost to player i ci(Pi)
length(Pi) - Utility function of player i can depend on both
cost and the set Si picked by i - cost to player i S ce(ke)/ke fi(Si)
- PoS is still within log(k) if ce is concave
e ? Si
26More Questions
- Cost and Latency
- Only Latency
- Nash exist (same potential argument)
- Best NE costs at most OPT w/ twice as many
players. - Best Response Dynamics
- Can construct games with k players so that a
certain ordering of moves takes 2O(k) time. - Weighted Game
27Adding Latency
- What if we want to model congestion?
- marginal cost increases, so not buy-at-bulk.
- Every edge has increasing delay function de(ke).
- Cost of edge e for player i is
-
ce(ke)/kede(ke). - Total cost of edge is
- ce(ke) ke?de(ke).
-
28Cost Latency
- From earlier proof, we know that if for all S,
- cost(S) lt A??(S) lt AB?cost(S),
- then the price of stability is lt AB.
- if ce is concave, de is nondecreasing for all e,
and - for all e and xe then the price of stability is
at most AHk (separate cost and latency) - E.g. if ce is concave, de is polynomial with
degree m, - then Price of Stability is lt (m1)?log(k).
-
-
-
29Latency
- In this case Nash Equilibria can be computed.
Convert all edges
d(1)
All edges capacity 1
d(x)
d(2)
d(3)
- Claim A min cost flow corresponds to a NE.
- Idea Since d is increasing, flow will use d(1),
then d(2), etc, mirroring a potential function. - Fabrikant, Papadimitriou,
Talwar
30Latency
- Results (with single source)
- Nash exist (same potential argument)
- Best NE costs at most OPT w/ twice as many
players.
31Best Response Dynamics
- How long before players settle on a NE?
- In games with 2 players, O(n) time,
- since shared segment grows monotonically.
- Can construct games with k players so that a
certain ordering of moves takes 2O(k) time. - Can 3-player games run for exponential time?
- Can k-player games be scheduled to be polytime?
32Weighted Game
- If some player has more traffic, should pay more
- In a weighted game, player i has weight w(i).
- Players pay for edges proportionally to their
weight. - No potential function exists. Do NE always
exist? - Best Response converges for single commodity.
- Games with at most 2 players per edge have NE.
- If NE do exist, Price of Stability will be gtgt
log(k)
33Games with at most 2 players per edge have NE
For edge e used by players i and j
- For edge with one player F(s) wici if I uses e
0 otherwise - When player i moves the change in F(S) is equal
to the change in player's i payment scaled up by
wi
34Best Response converges for single commodity
- All players have the same source and sink
- For every s-t path P the marginal weight is
- Order the paths of the players in tuple in
according their marginal weight(lexicographic
order) - Show that in every step the lexicographic size of
the tuple decrees - If the move is P1 -gt P2
- P group of paths that have edges in either P1
or P2 - C(P2) the cost of P2 after the move
- show that
35If NE do exist, Price of Stability will be gtgt
log(k)
t
1
1
1
1
2
2
2
1/2
2
. . .
1
1
2
3
k
k-1
0
0
0
0
0
36Thank you.