NOT JUST A CHILD’S PLAY CAKE CUTTING. How does one fairly divide goods among several people?

Slides:



Advertisements
Similar presentations
Stable Matching Problems with Constant Length Preference Lists Rob Irving, David Manlove, Gregg OMalley University Of Glasgow Department of Computing Science.
Advertisements

On allocations that maximize fairness Uriel Feige Microsoft Research and Weizmann Institute.
6.896: Topics in Algorithmic Game Theory Lecture 21 Yang Cai.
Chapter Thirty-One Welfare Social Choice u Different economic states will be preferred by different individuals. u How can individual preferences be.
Announcement Paper presentation schedule online Second stage open
Ariel D. Procaccia (Microsoft)  A cake must be divided between several children  The cake is heterogeneous  Each child has different value for same.
3.1 Fair Division: To divide S into shares (one for each player) in such a way that each player gets a fair share. Fair Division: To divide S into shares.
Cakes, Pies, and Fair Division Walter Stromquist Swarthmore College Rutgers Experimental Mathematics Seminar October 4, 2007.
Chapter 13: Fair Division Lesson Plan
Multi-item auctions with identical items limited supply: M items (M smaller than number of bidders, n). Three possible bidder types: –Unit-demand bidders.
Beyond Dominant Resource Fairness David Parkes (Harvard) Ariel Procaccia (CMU) Nisarg Shah (CMU)
Better Ways to Cut a Cake Steven Brams – NYU Mike Jones – Montclair State University Christian Klamler – Graz University Paris, October 2006.
CUTTING A BIRTHDAY CAKE Yonatan Aumann, Bar Ilan University.
No Agent Left Behind: Dynamic Fair Division of Multiple Resources Ian Kash 1 Ariel Procaccia 2 Nisarg Shah 2 (Speaker) 1 MSR Cambridge 2 Carnegie Mellon.
1 NP-Complete Problems. 2 We discuss some hard problems:  how hard? (computational complexity)  what makes them hard?  any solutions? Definitions 
Bounds on Code Length Theorem: Let l ∗ 1, l ∗ 2,..., l ∗ m be optimal codeword lengths for a source distribution p and a D-ary alphabet, and let L ∗ be.
Excursions in Modern Mathematics Sixth Edition
Cake Cutting is Not a Piece of Cake Malik Magdon-Ismail Costas Busch M. S. Krishnamoorthy Rensselaer Polytechnic Institute.
Online Cake Cutting Toby Walsh NICTA and UNSW Sydney, Australia.
Cutting a Pie is Not a Piece of Cake Walter Stromquist Swarthmore College Third World Congress of the Game Theory Society Evanston,
Not Everyone Likes Mushrooms: Fair Division and Degrees of Guaranteed Envy-Freeness* Second GASICS Meeting Computational Foundations of Social Choice Aachen,
Selfridge-Conway Fair Division Procedure An Envy-Free Cake Division Procedure.
Dividing a Cake Fairly among n players Thomas Yeo
Chapter 13: Fair Division Lesson Plan
Lau Ting Sum Samson Suen Wai.  Discuss what fairness is  Describe some methods for fair division: 1. Divide-and-choose 2. Last Diminisher 3. Selfridge-Conway.
TRUTH, JUSTICE, AND CAKE CUTTING Yiling Chen, John K. Lai, David C. Parkes, Ariel D. Procaccia (Harvard SEAS) 1.
Multiplication of Fractions
An Experimental Test of House Matching Algorithms Onur Kesten Carnegie Mellon University Pablo Guillen University of Sydney.
Ariel D. Procaccia (Microsoft)  Best advisor award goes to...  Thesis is about computational social choice Approximation Learning Manipulation BEST.
TRUTH, JUSTICE, AND CAKE CUTTING Ariel Procaccia (Harvard SEAS) 1.
An Introduction to Black-Box Complexity
1 On Approximately Fair Allocations of Indivisible Goods Elchanan Mossel Amin Saberi Richard Lipton Vangelis Markakis Georgia Tech AUEB U. C. Berkeley.
Competitive Analysis of Incentive Compatible On-Line Auctions Ron Lavi and Noam Nisan SISL/IST, Cal-Tech Hebrew University.
1 Fair Allocations of Indivisible Goods Part I: Minimizing Envy Elchanan Mossel Amin Saberi Richard Lipton Vangelis Markakis Georgia Tech CWI U. C. Berkeley.
Collusion and the use of false names Vincent Conitzer
Improved results for a memory allocation problem Rob van Stee University of Karlsruhe Germany Leah Epstein University of Haifa Israel WADS 2007 WAOA 2007.
Multi-Unit Auctions with Budget Limits Shahar Dobzinski, Ron Lavi, and Noam Nisan.
Some Key Facts About Optimal Solutions (Section 14.1) 14.2–14.16
Chapter 14: Fair Division Part 4 – Divide and Choose for more than two players.
Dominant Resource Fairness: Fair Allocation of Multiple Resource Types Ali Ghodsi, Matei Zaharia, Benjamin Hindman, Andy Konwinski, Scott Shenker, Ion.
Authors: David Robert Martin Thompson Kevin Leyton-Brown Presenters: Veselin Kulev John Lai Computational Analysis of Position Auctions.
Introduction to Matching Theory E. Maskin Jerusalem Summer School in Economic Theory June 2014.
Chapter 14: Fair Division Part 5 – Defining Fairness.
Excursions in Modern Mathematics, 7e: Copyright © 2010 Pearson Education, Inc. 3 The Mathematics of Sharing 3.1Fair-Division Games 3.2Two Players:
August 16, 2010 MPREF’10 Dynamic House Allocation Sujit Gujar 1, James Zou 2 and David C. Parkes 2 5 th Multidisciplinary Workshop on Advances in Preference.
Slide 1 of 16 Noam Nisan The Power and Limitations of Item Price Combinatorial Auctions Noam Nisan Hebrew University, Jerusalem.
Fair Division Ch. 13 Finite Math. Fair division There are about 1.2 million divorces every year in the U.S. alone. International disputes redefine borders.
Chapter 3 Fair Division.
1 The Theory of NP-Completeness 2 Cook ’ s Theorem (1971) Prof. Cook Toronto U. Receiving Turing Award (1982) Discussing difficult problems: worst case.
Automated Mechanism Design Tuomas Sandholm Presented by Dimitri Mostinski November 17, 2004.
McGraw-Hill/Irwin © The McGraw-Hill Companies, Inc., Table of Contents CD Chapter 14 (Solution Concepts for Linear Programming) Some Key Facts.
Unconventional Fixed-Radix Number Systems
By: Eric Zhang.  Indivisible items from multiple categories are allocated to agents without monetary transfer  Example – How paper presentations are.
Chapter 11 Resource Allocation by Mikhail Nesterenko “Distributed Algorithms” by Nancy A. Lynch.
Fair Shares.
Instructor: Shengyu Zhang 1. Resource allocation General goals:  Maximize social welfare.  Fairness.  Stability. 2.
Non-LP-Based Approximation Algorithms Fabrizio Grandoni IDSIA
Chapter 33 Welfare 2 Social Choice Different economic states will be preferred by different individuals. How can individual preferences be “aggregated”
Presented by Qifan Pu With many slides from Ali’s NSDI talk Ali Ghodsi, Matei Zaharia, Benjamin Hindman, Andy Konwinski, Scott Shenker, Ion Stoica.
WASTE MAKES HASTE: Erel Segal-Halevi, Avinatan Hassidim, Yonatan Aumann BOUNDED-TIME PROTOCOLS FOR ENVY-FREE CAKE CUTTING WITH FREE DISPOSAL.
A pie that can’t be cut fairly Walter Stromquist Swarthmore College Fair Division Seminar Dagstuhl, Deutschland June 26, 2007.
Fair Division Fair Division Problem: A problem that involves the dividing up of an object or set of objects among several individuals (players) so that.
Minimum Spanning Tree 8/7/2018 4:26 AM
Envy-Free Cake-Cutting in Bounded Time
Mathematical Foundations of AI
Fair division Lirong Xia Oct 7, 2013.
Intro to the Fair Allocation
Matching and Resource Allocation
Fair Division Fair Division Problem: A problem that involves the dividing up of an object or set of objects among several individuals (players) so that.
Chapter 34 Welfare Key Concept: Arrow’s impossibility theorem, social welfare functions Limited support of how market preserves fairness.
Presentation transcript:

NOT JUST A CHILD’S PLAY CAKE CUTTING

How does one fairly divide goods among several people?

What is “fairness” ? Envy-freeness Each participant prefers keeping his own allocation to swapping with any other participant. Proportionality Each of the n participants receives at least 1/n of his value for getting everything.

Divisible goods and Indivisible goods Divisible goods Such as land, time, or memory on a computer. Indivisible goods Such as a house or the computer itself.

What’s cake-cutting ?

Cake Cutting = Allocating a heterogeneous divisible good among multiple players with different preferences

Outline 1. Cake cutting mechanisms 2. Complexity of cake cutting 3. A game-theoretic viewpoint 4. Optimizing welfare

Cake cutting mechanisms Chapter 1

Cut and Choose The famous cut and choose algorithm: 1st step : the first player divides the cake into two pieces that he values equally. 2rd step : the second player then chooses the piece that he prefers. 3rd step : the first player receives the remaining piece.

Is the cut and choose algorithm fair? Proportional? Yes! Note that -- The first player values both pieces at exactly 1/2, While the second player receives his preferred piece, which must be worth at least 1/2. Envy-free? Yes! In fact – For the case of two players, the concepts of envy-freeness and proportionality coincide!

What about n -player setting? As we move from the two player setting to the n - player setting, fairness becomes harder to achieve. Nevertheless, several elegant algorithms guarantee proportional allocations.

Dubins-Spanier 1961 In each stage, a referee slowly moves a knife over the cake from left to right. When the knife reaches a point such that the piece of cake to the left of that point is worth 1/n to one of the players, this player shouts “stop”. The referee makes a cut, and the piece of cake to the left of the cut is given to the player. The satisfied player and allocated piece are then removed. The process is repeated with the remaining players and leftover cake, until there is only one player left. The last player receives the unclaimed piece.

Even-Paz 1984 Assume for ease of exposition that the number of players is a power of 2. Similarly to the discretized version of the latter algorithm, each time the procedure is executed, the players make marks where the cake to the left of the mark is valued at 1/2 (rather than 1/n, as before).

Even-Paz 1984 Rather than removing a single player ---- we separate the players into two subsets of equal size, such that all the marks made by the players of the first subset lie to the left of the marks made by the players of the second subset. The players in the first subset then receive the piece of cake that lies to the left of their rightmost mark, while the players in the second subset receive the remaining cake.

Even-Paz 1984

How to guarantee envy-freeness? While proportionality is well understood, envy- freeness is a far more elusive property. Proportionality is always implied by envy-freeness, in the case of divide the whole cake.

How to guarantee envy-freeness? For three players, Selfridge-Conway For an arbitrary number of players, Brams-Taylor

Selfridge-Conway 1960 Stage 0 Player 1 divides the cake into three equal pieces according to his valuation. Player 2 trims the largest piece (that is, cuts off a slice) such that there is a tie between the two largest pieces in his eyes. We call the original cake without the trimmings Cake 1, and we call the trimmings Cake 2.

Selfridge-Conway 1960 Stage 1 (Division of cake 1) Order: – 1 Player 3 chooses one of the three pieces of Cake 1. Either player 2 or player 3 receives the trimmed piece; denote that player by U, and the other player by Ū. Player1 is allocated the remaining(untrimmed) piece. Stage 2 (Division of cake 2) Ū divides cake 2 into three equal pieces according to his valuation. Players U, 1, and Ū choose the pieces of Cake 2, in that order.

Brams-Taylor 1992 The first envy-free cake cutting algorithm for an arbitrary number of players! However— Through the computational lens, the algorithm’s running time is unbounded.

Brams-Taylor 1992 A bounded envy-free algorithm for the five-player case was recently proposed by Saberi and Wang However, the n-player case remains open ! Is envy-free cake cutting inherently complex?

Complexity of cake cutting Chapter 2

Robertson-Webb’s model 1998 Evaluation queryCut query

Robertson-Webb’s model Even-Paz Dubins-Spanier

The complexity of proportional cake cutting

The complexity of envy-free cake cutting We would like to be able to establish that bounded envy-free cake cutting algorithms do not exist. Stromquist 2008 established such a nonexistence result under the assumption that the algorithm must allocate connected pieces (even for the 3-person case).

The complexity of envy-free cake cutting

Ways to circumvent the problem Approximately approach Lipton et al Special valuation structure Chen et al

A game-theoretical viewpoint Chapter 3

Strategyproof Strategyproof : Players must not be able to gain from manipulating the algorithm, regardless of the actions of others.

Cut and choose algorithm is not strategyproof

Strategyproof cake cutting algorithm

How to compute a perfect partition Chen et al showed that perfect partitions can be computed efficiently when valuations have a piecewise constant structure. piecewise constant : each player only desires certain pieces of cake, and values each of these pieces uniformly. Additionally, if each player has a single desired piece of cake that he values uniformly, fairness and truthfulness can be guaranteed without resorting to randomization.

Strategyproof cake cutting algorithm The design of strategyproof cake cutting algorithms is still largely an open problem. First, because the above algorithms (especially the deterministic one) can only handle restricted valuations. Second, because these algorithms cannot be simulated in the Robertson-Webb model.

Optimizing welfare Chapter 4

Social welfare Utilitarian social welfare and egalitarian social welfare We assume that all players have the same value for the whole cake, say $1. Price of fairness : the worst-case ratio between the social welfare of the optimal allocation, and the social welfare of the best fair allocation.

Price of fairness—an example

Any proportional allocation The social welfare is smaller than 2. The welfare-max allocation

Price of fairness—an example

Dumping paradox Aumann –Dombb 2010 studied the price of fairness under the assumption that connected pieces must be allocated. A interesting insight in this context is dumping paradox : by throwing away pieces of cake, one can increase the social welfare of the best proportional (or envy- free) allocation!

Dumping paradox

Optimal fair cake divisions For piecewise constant valuations, welfare-maximizing proportional or envy-free allocations can be computed in polynomial time Cohler et al Computing optimal fair cake divisions with connected pieces is significantly harder X. Bei et al Even if we abandon fairness completely and just focus on optimizing welfare Aumann et al

How about Pareto-efficient? Pareto-efficient : in the sense that no other allocation is valued at least as highly by all players, and is strictly better for at least one player. there are examples where no welfare-maximizing envy-free allocation is Pareto-efficient, even when there are only three players with piecewise constant valuations Brams et al

How about Pareto-efficient? Should we sacrifice social welfare to obtain Pareto- efficiency? How much must be sacrificed? More generally, what would constitute an ideal cake division ? ??………. These conceptual questions may lead to significant technical insights on the role of optimization in cake cutting!

ZHENG BO THANK YOU! This field is fun, potentially very significant, and gives rise to great intellectual challenges!