The Journal of Chemical Physics,
Journal Year:
2023,
Volume and Issue:
158(6)
Published: Feb. 8, 2023
Kohn-Sham
(KS)
inversion,
in
which
the
effective
KS
mean-field
potential
is
found
for
a
given
density,
provides
insights
into
nature
of
exact
density
functional
theory
(DFT)
that
can
be
exploited
development
approximations.
Unfortunately,
despite
significant
and
sustained
progress
both
software
libraries,
inversion
remains
rather
difficult
practice,
especially
finite
basis
sets.
The
present
work
presents
method,
dubbed
"Lieb-response"
approach,
naturally
works
with
existing
Fock-matrix
DFT
infrastructure
sets,
numerically
efficient,
directly
meaningful
matrix
energy
quantities
pure-state
ensemble
systems.
Some
additional
yields
potential.
It
thus
enables
routine
even
systems,
as
illustrated
variety
problems
within
this
work,
outputs
used
embedding
schemes
or
machine
learning
effect
sets
on
also
analyzed
investigated.
Abstract
While
in
principle
exact,
Kohn–Sham
density
functional
theory—the
workhorse
of
computational
chemistry—must
rely
on
approximations
for
the
exchange–correlation
functional.
Despite
staggering
successes,
present‐day
still
struggle
when
effects
electron–electron
correlation
play
a
prominent
role.
The
limit
which
electronic
Coulomb
repulsion
completely
dominates
offers
well‐defined
mathematical
framework
that
provides
insight
new
able
to
deal
with
strong
correlation.
In
particular,
structure
this
limit,
is
now
well‐established
thanks
its
reformulation
as
an
optimal
transport
problem,
points
use
very
different
ingredients
(or
features)
respect
traditional
ones
used
present
approximations.
We
focus
strategies
these
build
chemistry
and
highlight
future
promising
directions.
This
article
categorized
under:
Electronic
Structure
Theory
>
Density
Functional
Physical Chemistry Chemical Physics,
Journal Year:
2023,
Volume and Issue:
25(31), P. 20817 - 20836
Published: Jan. 1, 2023
We
study
self-interaction
effects
in
solvated
and
strongly-correlated
cationic
molecular
clusters,
with
a
focus
on
the
hydroxyl
radical.
To
address
issue,
we
apply
DC-r2SCAN
method,
auxiliary
density
matrix
approach.
Validating
our
method
through
simulations
of
bulk
liquid
water,
demonstrate
that
maintains
structural
accuracy
r2SCAN
while
effectively
addressing
spin
localization
issues.
Extending
analysis
to
find
hemibonded
motif
[CH3S∴CH3SH]+
cluster
is
disrupted
simulation,
contrast
preserves
(three-electron-two-center)-bonded
motif.
Similarly,
for
[SH∴SH2]+
cluster,
restores
leakage,
predicts
weaker
hemibond
formation
influenced
by
solvent-solute
interactions.
Our
findings
potential
combined
improve
electronic
structure
calculations,
providing
insights
into
properties
clusters.
This
work
contributes
advancement
corrected
theory
offers
computational
framework
modeling
condensed
phase
systems
intricate
correlation
effects.
Journal of Chemical Theory and Computation,
Journal Year:
2022,
Volume and Issue:
19(14), P. 4451 - 4460
Published: Dec. 1, 2022
The
electron
density
of
a
molecule
or
material
has
recently
received
major
attention
as
target
quantity
machine-learning
models.
A
natural
choice
to
construct
model
that
yields
transferable
and
linear-scaling
predictions
is
represent
the
scalar
field
using
multicentered
atomic
basis
analogous
routinely
used
in
fitting
approximations.
However,
nonorthogonality
poses
challenges
for
learning
exercise,
it
requires
accounting
all
components
at
once.
We
devise
gradient-based
approach
directly
minimize
loss
function
regression
problem
an
optimized
highly
sparse
feature
space.
In
so
doing,
we
overcome
limitations
associated
with
adopting
atom-centered
learn
over
arbitrarily
complex
data
sets,
obtaining
very
accurate
comparatively
small
training
set.
enhanced
framework
tested
on
32-molecule
periodic
cells
liquid
water,
presenting
enough
complexity
require
optimal
balance
between
accuracy
computational
efficiency.
show
starting
from
predicted
single
Kohn-Sham
diagonalization
step
can
be
performed
access
total
energy
carry
error
just
0.1
meV/atom
respect
reference
functional
calculations.
Finally,
test
our
method
heterogeneous
QM9
benchmark
set,
showing
fraction
derive
ground-state
energies
within
chemical
accuracy.
The Journal of Physical Chemistry B,
Journal Year:
2022,
Volume and Issue:
126(45), P. 9349 - 9360
Published: Nov. 3, 2022
The
hydration
structure
of
Na+
and
K+
ions
in
solution
is
systematically
investigated
using
a
hierarchy
molecular
models
that
progressively
include
more
accurate
representations
many-body
interactions.
We
found
conventional
empirical
pairwise
additive
force
field
commonly
used
biomolecular
simulations
unable
to
reproduce
the
extended
X-ray
absorption
fine
(EXAFS)
spectra
for
both
ions.
In
contrast,
progressive
inclusion
effects
rigorously
derived
from
expansion
energy
allows
MB-nrg
potential
functions
(PEFs)
achieve
nearly
quantitative
agreement
with
experimental
EXAFS
spectra,
thus
enabling
development
molecular-level
picture
solution.
Since
PEFs
have
already
been
shown
accurately
describe
isomeric
equilibria
vibrational
small
ion–water
clusters
gas
phase,
present
study
demonstrates
effectively
represent
long-sought-after
able
correctly
predict
properties
ionic
aqueous
systems
liquid
which
has
so
far
remained
elusive.
The Journal of Chemical Physics,
Journal Year:
2023,
Volume and Issue:
158(6)
Published: Feb. 8, 2023
Kohn-Sham
(KS)
inversion,
in
which
the
effective
KS
mean-field
potential
is
found
for
a
given
density,
provides
insights
into
nature
of
exact
density
functional
theory
(DFT)
that
can
be
exploited
development
approximations.
Unfortunately,
despite
significant
and
sustained
progress
both
software
libraries,
inversion
remains
rather
difficult
practice,
especially
finite
basis
sets.
The
present
work
presents
method,
dubbed
"Lieb-response"
approach,
naturally
works
with
existing
Fock-matrix
DFT
infrastructure
sets,
numerically
efficient,
directly
meaningful
matrix
energy
quantities
pure-state
ensemble
systems.
Some
additional
yields
potential.
It
thus
enables
routine
even
systems,
as
illustrated
variety
problems
within
this
work,
outputs
used
embedding
schemes
or
machine
learning
effect
sets
on
also
analyzed
investigated.