Fairness-Efficiency Tradeoffs in Dynamic Fair Division David Zeng, - - PowerPoint PPT Presentation

β–Ά
fairness efficiency tradeoffs in dynamic
SMART_READER_LITE
LIVE PREVIEW

Fairness-Efficiency Tradeoffs in Dynamic Fair Division David Zeng, - - PowerPoint PPT Presentation

Fairness-Efficiency Tradeoffs in Dynamic Fair Division David Zeng, Alex Psomas items arrive online, agents Agent has value [0,1] for item that we learn when the item arrives Item must be allocated


slide-1
SLIDE 1

Fairness-Efficiency Tradeoffs in Dynamic Fair Division

  • π‘ˆ items arrive online, π‘œ agents
  • Agent 𝑗 has value 𝑀𝑗𝑒 ∈ [0,1] for item 𝑒 that we learn when the item

arrives

  • Item must be allocated immediately and irrevocably
  • After π‘ˆ rounds, we will have some allocation 𝐡 = (𝐡1, … , π΅π‘œ)
  • Additive valuations: 𝑀𝑗 π΅π‘˜ = Οƒπ‘•π‘’βˆˆπ΅π‘˜ 𝑀𝑗𝑒
  • Ideally, allocation is both fair and efficient

David Zeng, Alex Psomas

slide-2
SLIDE 2

Adversary Model and Results

𝑀𝑗𝑒~𝐸 𝑀𝑗𝑒~𝐸𝑗 Τ¦ 𝑀𝑒~𝐸 Non-adaptive adversary Adaptive adversary Fairness and efficiency compatible Fairness and efficiency incompatible ex-post Pareto efficient and pairwise (EF1 or EF w.h.p.) allocation algorithm No

1 π‘œ + 𝜁 -Pareto efficient and sublinear

envy allocation algorithm

slide-3
SLIDE 3

Algorithm for Correlated Agents ( Τ¦ 𝑀𝑒~𝐸)

  • Reduce finding a
  • online ex-post Pareto efficient and (EF1 or EF w.h.p.) allocation algorithm to

finding a

  • offline Pareto efficient and CISEF fractional allocation
  • Algorithm sketch
  • Given online problem and distribution 𝐸 with support 𝛿1, … , 𝛿𝑛 , use the

support of 𝐸 as the items for offline problem, scaling by the probabilities.

  • Use the fractional allocation π‘Œ to guide our allocation in the online problem.
  • If π‘Œπ‘—π‘™ = 0.4, if the item arriving at time 𝑒 has type 𝛿𝑙, allocate the item to agent 𝑗 with

probability 0.4

  • Treat cliques as one combined agent when doing randomized allocation
  • When item is allocated to the clique, give to unhappiest agent in clique
slide-4
SLIDE 4

CISEF

Strongly envy-free Identical allocations and valuations

1 2 3

Clique Identical Strongly Envy-Free

  • CISEF
  • Either agent 𝑗 strictly prefers her own

bundle to the bundle of agent π‘˜

  • Or 𝑗 and π‘˜ have identical allocations

and the same value (up to a scaling factor) for all the items that are allocated to either of them

  • How to find CISEF and Pareto

efficient allocation?

  • Start with solution to Eisenberg-Gale

convex program

  • If agent 𝑗 is indifferent to agent π‘˜,

(carefully) move items from π‘˜ to 𝑗 to create strong envy-free edges