Unformatted text preview: d to solve
hard local optimization problems to compute their costs to perform di erent bundles of
jobs; each bundle may require that the agent computes an optimal schedule for its eet of
vehicles.
In these types of combinatorial problems a mechanism must not require an agent to
report its complete valuation function. In addition, an agent must be able to compute its
optimal strategy without computing its complete valuation function. It is not helpful to
require less information if the agents must still compute values for all bundles to provide
that information. In Chapter 8 I introduce a boundedrational compatible characterization
of auctions. The theory of boundedrational compatibility precisely captures this idea that
an agent can participate in an auction without performing unnecessary valuation work.
In a boundedrational compatible auction an agent can compute its equilibrium strategy
with an approximate valuation function, at least in some problem instances.
Two interesting approaches to reduce information revelation in mechanisms for combinatorial allocation problems are:
1 Retain the directrevelation structure, but provide a highlevel bidding language or
bidding program" to allow an agent to represent and de ne" its local problem without
explicitly solving its local problem in all possible scenarios.
2 Implement a dynamic mechanism, that requests information incrementally from
agents and computes the optimal allocation and Vickrey payments without complete information revelation.
The rst method may be helpful when speci cation is easier than valuation, i.e. it
is easier for an agent to de ne how it determines its value for a bundle of items than it
is to explicitly compute its value for all possible bundles. The second method may be
73 helpful when the iterative procedure terminates without complete information revelation
by agents, and when an agent can provide incremental information without computing its
complete valuation function. Let us consider each in turn. Bidding Programs and High Level Bidding Languages
In choosing a bidding language for a mechanism there is a tradeo between the ease with
which an agent can represent its local preferences, and the ease with which the mechanism
can compute the outcome. Nisan Nis00 describes the expressiveness of a language, which
is a measure of the size of a message for a particular family of valuation functions, and the
simplicity of a language, which is a measure of the complexity involved in interpreting a
language and computing values for di erent outcomes.
A natural starting point in combinatorial auctions is the XOR bidding language,
S1 ; p1 xor S2 ; p2 , which essentially allows an agent to enumerate its value for all possible sets of items. This bidding language is simple to interpret, in fact given a bid b in the
XOR language, the auctioneer can compute the value bS for any bundle in polynomial
time Nis00 . However, this bidding language is not very expressive. An obvious example is
P
provided with a linear valuation function, vS = x2S vx. XOR bids for this valuation
function are exponential in size explicitly enumera...
View
Full Document
 Fall '13
 DavidParkes

Click to edit the document details