BEGIN:VCALENDAR
PRODID:-//Ben Fortuna//iCal4j 1.0//EN
VERSION:2.0
CALSCALE:GREGORIAN
X-WR-CALNAME:Calendar
BEGIN:VEVENT
DTSTAMP:20221129T070731Z
SUMMARY:Jordy van Velthoven: Density theorems for discrete series restric
ted to lattices
DESCRIPTION:Seminarium\, GOAT (Grupp och operator-algebror träff)
LOCATION:Kräftriket\, House 5\, Room 16
DTSTART:20220310T100000Z
DTEND:20220310T110000Z
UID:2187728f-7124-4463-96b0-91032ed21640
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20221129T070731Z
SUMMARY:Matteo Tanzi: Self-sustaining measures for high-dimensional weakl
y coupled maps
LOCATION:Room 3721\, Lindstedtsvägen 25
DTSTART:20220310T120000Z
DTEND:20220310T120000Z
UID:26cd2f4f-0699-4479-8f52-67476a553a19
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20221129T070731Z
SUMMARY:Erik Burman: Spacetime finite element methods for control problem
s subject to the wave equation
LOCATION:Zoom
DTSTART:20220310T130000Z
DTEND:20220310T140000Z
UID:6fef539e-7a1d-4205-9262-13b96ae3a205
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20221129T070731Z
SUMMARY:Michael Weiss: Configuration space methods applied to automorphis
m groups of spheres
LOCATION:Institut Mittag-Leffler\, Seminar Hall Kuskvillan and Zoom
DTSTART:20220310T131500Z
DTEND:20220310T150000Z
UID:6bc9e6f8-6c0c-4d86-a9fb-d3ada8fa70a5
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20221129T070731Z
SUMMARY:Recursive McCormick Linearizations of Multilinear Programs: Minim
um Size Formulations
DESCRIPTION:Multilinear Programs (MLPs) are a particular class of MINLPs
which have multilinear functions in both the objective and the constrain
ts. MLPs are typically solved by using Branch-and-Bound algorithms which
rely on Linear Programming (LP) relaxations to obtain lower bounds. The
se LP relaxations are derived using Recursive McCormick Linearizations i
.e.\, by recursively introducing additional variables that represent bil
inear products and by relaxing using McCormick envelopes. The size of th
e LP relaxation depends on the heuristic employed to identify the collec
tion of variables to add. In this talk\, we introduce the first approach
for identifying the smallest size LP relaxation for a given MLP\, by in
vestigating an Integer Programming (IP) model that solves a specialized
network flow representation where linearizations are encoded as in-trees
. Our results on a collection of benchmarks indicate that the IP formula
tion can find smaller linearizations (up to 30% reduction in number of v
ariables) and tighter relaxations (30% reduction in the root-node optima
lity gaps). We also provide insights into computing the best bound Recur
sive McCormick Linearization for a given LP size.
LOCATION:3721\, note that it is online\, but we setup with projector here
DTSTART:20220310T140000Z
DTEND:20220310T150000Z
UID:232f295c-706a-4386-98d7-0f5abe589a90
END:VEVENT
END:VCALENDAR