Self-Refinement of Auxiliary-Field Quantum Monte Carlo via Non-Orthogonal Configuration Interaction
Journal of Chemical Theory and Computation,
Journal Year:
2025,
Volume and Issue:
unknown
Published: April 28, 2025
For
optimal
accuracy,
auxiliary-field
quantum
Monte
Carlo
(AFQMC)
requires
trial
states
consisting
of
multiple
Slater
determinants.
We
develop
an
efficient
algorithm
to
select
the
determinants
from
AFQMC
random
walk
eliminating
need
for
other
methods.
When
contribute
significantly
nonorthogonal
configuration
interaction
energy,
we
include
them
in
state.
These
refined
wave
functions
reduce
phaseless
bias
and
sampling
variance
local
energy
estimator.
With
100
200
determinants,
lower
error
by
up
a
factor
10
second-row
elements
that
are
not
accurately
described
with
Hartree-Fock
function.
HEAT
set,
improve
average
within
chemical
accuracy.
benzene,
largest
studied
system,
80%
214
find
10-fold
increase
time
solution.
show
errors
prevail
systems
static
correlation
or
strong
spin
contamination.
such
systems,
improved
enable
stable
free-projection
calculations,
achieving
accuracy
even
strongly
correlated
regime.
Language: Английский
Walking through Hilbert Space with Quantum Computers
Chemical Reviews,
Journal Year:
2025,
Volume and Issue:
unknown
Published: May 2, 2025
Computations
of
chemical
systems'
equilibrium
properties
and
nonequilibrium
dynamics
have
been
suspected
being
a
"killer
app"
for
quantum
computers.
This
review
highlights
the
recent
advancements
algorithms
tackling
complex
sampling
tasks
in
key
areas
computational
chemistry:
ground
state,
thermal
state
properties,
calculations.
We
broad
range
algorithms,
from
hybrid
quantum-classical
to
fully
quantum,
focusing
on
traditional
Monte
Carlo
family,
including
Markov
chain
Carlo,
variational
projector
path
integral
etc.
also
cover
other
relevant
techniques
involving
tasks,
such
as
quantum-selected
configuration
interaction,
minimally
entangled
typical
states,
entanglement
forging,
Carlo-flavored
Lindbladian
dynamics.
provide
comprehensive
overview
these
algorithms'
classical
counterparts,
detailing
their
theoretical
frameworks
discussing
potentials
challenges
achieving
advantages.
Language: Английский