[Prev][Next][Index]
No XTAG meeting, talk which will probably be of interest at that time
-
To: srini@linc.cis.upenn.edu, beth@linc.cis.upenn.edu, cdoran@linc.cis.upenn.edu, mpalmer@linc.cis.upenn.edu, joshi@linc.cis.upenn.edu, anoop@linc.cis.upenn.edu, skulick@linc.cis.upenn.edu, mickeyc@linc.cis.upenn.edu, bhatt@linc.cis.upenn.edu, josephr@linc.cis.upenn.edu, siegel@BABEL.ling.upenn.edu, kipper@gradient.cis.upenn.edu, vshanker@linc.cis.upenn.edu, schuler@gradient.cis.upenn.edu, izvorski@BABEL.ling.upenn.edu, spc@gradient.cis.upenn.edu, park@linc.cis.upenn.edu, prolo@gradient.cis.upenn.edu, fxia@gradient.cis.upenn.edu, kinyon@linc.cis.upenn.edu, merlo@linc.cis.upenn.edu, jmacdoug@central.cis.upenn.edu, tsmorton@gradient.cis.upenn.edu, phopely@linc.cis.upenn.edu (Philip D Hopely), cparkes@linc.cis.upenn.edu (Cornelia Parkes), tbleam@linc.cis.upenn.edu (Tonia Bleam), karttunen@linc.cis.upenn.edu, jason2@linc.cis.upenn.edu, nari@linc.cis.upenn.edu
-
Subject: No XTAG meeting, talk which will probably be of interest at that time
-
From: Christy Doran <cdoran@linc.cis.upenn.edu>
-
Date: Mon, 13 Apr 1998 16:51:46 -0400 (EDT)
Title: The Mathematical Programming of Logic
Speaker: Vijay Chandru
Department of Computer Science and Automation
Indian Institute of Science, Bangalore
Time & Date Thursday, April 16, 1998
10:30 - 12:00
Location: To be determined (Anoop will forward)
Abstract:
Leibniz, in the 17th century, brought about the first
important synthesis of logical deduction and mathematical
computation. His contribution was to point out that they are
fundamentally the same. Solving logical inference problems
with mathematical programming methods may seem a bit like
eating sauerkraut with chopsticks, because the two come from
vastly different worlds. Logical inference comes from a "left
brain" world of formal languages and symbolic manipulation.
vastly different worlds. Logical inference comes from a "left
brain" world of formal languages and symbolic manipulation.
Mathematical programming derives from a "right brain" world of
spatial models and numerical calculation. The thesis of this
talk is that many deductive inference problems do in fact have
the sort of mathematical structure that geometric methods
(linear programming) can exploit.
This perspective has yielded lovely structural results
relating Modus Ponens / Tollens with linear programming
relaxations, Resolution with cutting planes and Proof
Signatures with mathematical programming duality. New
embeddings of inference in modal, temporal, predicate and
partially interpreted logics in finite and infinite
mathematical programmes will be presented. And finally, the
use of these results in devising decision procedures for
hybrid systems will be discussed.
[In collaboration with V.S.Borkar (IISc), J.N.Hooker (CMU),
D. Micciancio (MIT) and S.K.Mitter (MIT)