Results 421 - 430 of 24337
This workshop brings together participants from the Special Year on Large Language Models and Transformers program, parts 1 and 2, held in the 2024-25 academic year, as well the Deep Learning Theory summer cluster. The workshop will provide an opportunity...
Abstract not available.
I argue that problems in chemistry and materials science that are of interest today mainly have a structure compatible with efficient classical simulation. Nonetheless, utilizing that structure when designing quantum algorithms can offer a new opportunity for quantum advantage.
Molecular dynamics is one of the most used primitives in computational chemistry for drug design. It requires accurate forces to propagate nuclei in time. A direct translation of this algorithm to quantum computers demonstrated that it is hard to scale and might not be the way forward. In this talk I will present a coherent algorithm that uses a Liouvillian evolution to propagate the nuclei while remaining in the Born-Oppenheimer picture that was introduced to address this problem. I will show the scaling and demonstrate that it is possible to use a simple Nosé thermostat with minor modifications. I will also present a second method that uses the Liouvillian as a subroutine in the calculation of relative binding free energies through thermal integration.
We discuss barriers to dequantization of near-term molecular electronic structure methods. In particular, we prove classical simulation hardness under the generalized P vs. NP conjecture, for quantum circuit families with applications in near-term molecular electronic ground state estimation. The proof exploits a connection to particle number conserving matchgate circuits with fermionic magic state inputs, which are shown to be universal for quantum computation under post-selection, and are therefore not classically simulable in the worst case, in either the strong (multiplicative) or weak (sampling) sense. We apply this result to quantum non-orthogonal multi-reference methods designed for near-term hardware, by ruling out certain dequantization strategies for computing the off-diagonal matrix elements between reference states. We demonstrate quantum speedups for two choices of ansatz incorporating both static and dynamic correlations to model the electronic eigenstates of molecular systems: linear combinations of orbital-rotated matrix product states, which can be prepared in linear depth, and linear combinations of states prepared by generalized UCCSD circuits of polynomial depth, for which computing the expectation values of local fermionic observables up to a constant additive error is BQP-complete. We discuss the implications for achieving practical quantum advantage in resolving the electronic structure of catalytic systems composed from multivalent transition metal atoms using near-term quantum hardware.