BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20220812T074334Z
LOCATION:Foyer 2nd Floor
DTSTART;TZID=Europe/Stockholm:20220628T090000
DTEND;TZID=Europe/Stockholm:20220628T110000
UID:submissions.pasc-conference.org_PASC22_sess181_pos131@linklings.com
SUMMARY:P18 - Physics-Inspired Representations for Atomistic Machine Learn
ing
DESCRIPTION:Poster\n\nP18 - Physics-Inspired Representations for Atomistic
Machine Learning\n\nFraux, Pozdnyakov, Ceriotti\n\nIn the last decade, ma
chine learning (ML) methods have been used to predict properties of molecu
les and materials with great success, reducing the cost of these predictio
ns while keeping the accuracy high. A crucial step in atomistic ML methods
is the mapping of atomic configurations to a set of features. By encoding
physical symmetries, sum rules, asymptotic tails, and other physical conc
epts directly in the representation of the atomic systems, we can dramatic
ally improve the accuracy of the resulting ML model and the data efficienc
y of the training exercise.
Our group is incorporating more phy
sical symmetries in these representations, we recently proposed a way to d
eal with different variances with respect to atomic permutations, which al
lowed us to learn properties defined on multiple centers. This has proven
a very useful tool to learn hamiltonian matrix elements directly, the inhe
rent symmetry of the model making it very robust to noise and errors.
Such fundamental developments go hand to hand with software develop
ment, as we are writing software libraries to compute atomic representatio
ns efficiently on HPC hardware. We are integrating these libraries within
well-known simulation tools such as LAMMPS, making atomistic machine learn
ing available to the whole community.
END:VEVENT
END:VCALENDAR