BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//RLASKEY//CALENDEROUS//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
DTSTAMP:20260910T083458Z
LAST-MODIFIED:20170908T135842Z
DTSTART:20170927T150000Z
DTEND:20170927T160000Z
UID:event1830@bu.edu
URL:http://physics.bu.edu/internal/events/show/1830
SUMMARY:"Iterative Compression-Decimation Scheme for Tensor Network Optimiz
	ation"
DESCRIPTION:Featuring Zhicheng Yang\, Boston University\, Physics Departmen
	t\n\nPart of the Condensed Matter Theory Seminar Series.\n\nMotivated by st
	atistical physics models connected to computation problems\, we devise a te
	nsor network technique that is suited to problems with or without translati
	on invariance and with arbitrary boundary conditions. We introduce a compre
	ssion-decimation algorithm as an efficient iterative scheme to optimize ten
	sor networks that encode generalized vertex models on regular lattices. The
	 algorithm first propagates local constraints to longer ranges via repeated
	 contraction-decomposition sweeps over all lattice bonds\, thus achieving c
	ompression on a given length scale. It then decimates the lattice via coars
	e-graining tensor contractions. Repeated iterations of these two steps allo
	w us to gradually collapse the tensor network while keeping the tensor dime
	nsions under control\, such that ultimately the full tensor trace can be ta
	ken for relatively large systems. As a benchmark\, we demonstrate the effic
	iency of the algorithm by computing the ground state entropy density of the
	 planar ice model and the eight-vertex model. We then apply it to reversibl
	e classical computational problems based on a recently proposed vertex mode
	l representation of classical computations [Nat. Commun. {\bf8}\, 15303 (20
	17)]. Our protocol allows us to obtain the exact number of solutions for co
	mputations where a naive enumeration would take astronomically long times\,
	 suggesting that the algorithm is a promising practical tool for the soluti
	on of a plethora of problems in physics and computer science.
LOCATION:SCI 328\, 590 Commonwealth Avenue\, 02215
STATUS:CONFIRMED
CLASS:PUBLIC
END:VEVENT
END:VCALENDAR