Excited-State DMRG Made Simple with FEAST

Baiardi, Alberto; Kelemen, A.; Reiher, Markus

doi:10.1021/acs.jctc.1c00984

Cited by 21 publications

(15 citation statements)

References 82 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These MR diagnostics oen disagree with each other, 24,43 with the diagnostics derived from DFT being less predictive than those derived from wavefunction theory (WFT). 44 Data-driven methods have augmented conventional approaches [45][46][47][48][49] for making system-specic decisions associated with carrying out quantum chemical calculations. For example, Jeong et al 50 demonstrated an ML protocol that performs automated selection of active spaces for bond dissociation of main group diatomic molecules, alleviating the computational cost.…”

Section: Introductionmentioning

confidence: 99%

Detection of multi-reference character imbalances enables a transfer learning approach for virtual high throughput screening with coupled cluster accuracy at DFT cost

et al. 2022

View full text Add to dashboard Cite

show abstract

Section: Introductionmentioning

confidence: 99%

Detection of multi-reference character imbalances enables a transfer learning approach for virtual high throughput screening with coupled cluster accuracy at DFT cost

et al. 2022

View full text Add to dashboard Cite

show abstract

“…However, WFT calculations scale at least N 4 with the system size and thus are impractical in VHTS of transition metal chemistry. Moreover, certain MR diagnostics have been observed to be more transferable over different chemical spaces (i.e., organic molecules vs TMCs) than others. − Therefore, data-driven approaches − have been developed to bridge the gap between DFT- and WFT-based diagnostics, making system-specific MR/SR decisions for method selection and automating active space selection for MR WFT calculations. ,, …”

Section: Introductionmentioning

confidence: 99%

Exploiting Ligand Additivity for Transferable Machine Learning of Multireference Character across Known Transition Metal Complex Ligands

Duan

Ladera

Liu

et al. 2022

J. Chem. Theory Comput.

View full text Add to dashboard Cite

Accurate virtual high-throughput screening (VHTS) of transition metal complexes (TMCs) remains challenging due to the possibility of high multireference (MR) character that complicates property evaluation. We compute MR diagnostics for over 5,000 ligands present in previously synthesized octahedral mononuclear transition metal complexes in the Cambridge Structural Database (CSD). To accomplish this task, we introduce an iterative approach for consistent ligand charge assignment for ligands in the CSD. Across this set, we observe that the MR character correlates linearly with the inverse value of the averaged bond order over all bonds in the molecule. We then demonstrate that ligand additivity of the MR character holds in TMCs, which suggests that the TMC MR character can be inferred from the sum of the MR character of the ligands. Encouraged by this observation, we leverage ligand additivity and develop a ligand-derived machine learning representation to train neural networks to predict the MR character of TMCs from properties of the constituent ligands. This approach yields models with excellent performance and superior transferability to unseen ligand chemistry and compositions.

show abstract

“…Because each additional root needs to fully solve the DMRG problem, memory and time requirements are significantly increased. Improved DMRG excited-state treatments for strongly correlated systems with intermediate m values are needed …”

Section: Resultsmentioning

confidence: 99%

“…Optimizations of excited states in quantum-chemical DMRG are not as straightforward as for traditional CI approaches. − In this work, we make use of a state-specific treatment (not to be confused with the treatment of orbitals in the state-average CASSCF optimization), which enforces electronic-state orthogonalization before and after each (two-site) tensor optimization for a given electronic state. For more information on the DMRG algorithms used in this work, see refs and − .…”

Section: Methodsmentioning

confidence: 99%

Relativistic Kramers-Unrestricted Exact-Two-Component Density Matrix Renormalization Group

Hoyer

et al. 2022

J. Phys. Chem. A

View full text Add to dashboard Cite

In this work we develop a variational relativistic density matrix renormalization group (DMRG) approach within the exact-two-component (X2C) framework (X2C-DMRG), using spinor orbitals optimized with the two-component relativistic complete active space self-consistent field. We investigate fine-structure splittings of p- (Ga, In, Tl) and d-block (Sc, Y, La) atoms and excitation energies of monohydride molecules (GeH, SnH, and TlH) with X2C-DMRG calculations using an all-electron relativistic Hamiltonian in a Kramers-unrestricted basis. We find that X2C-DMRG yields accurate 2P and 2D splittings compared to multireference configuration interaction with singles and doubles (MRCISD). We also investigated the degree of symmetry breaking in the atomic multiplets and convergence of electron correlation in the total energies. Symmetry breaking can be large in some cases (∼30 meV); however, increasing the number of renormalized block states m for the DMRG optimization recovers the symmetry breaking by several orders of magnitude. Encouragingly, we find the convergence of electron correlation to be close to MRCISDTQ5 quality. Relativistic X2C-DMRG approaches are important for cases where spin–orbit coupling is significant and the underlying reference wave function requires a large determinantal space. We are able to obtain quantitatively correct fine-structure splittings for systems up to 1019 number of determinants with traditional CI approaches, which are currently unfeasible to converge for the field.

show abstract

Excited-State DMRG Made Simple with FEAST

Cited by 21 publications

References 82 publications

Detection of multi-reference character imbalances enables a transfer learning approach for virtual high throughput screening with coupled cluster accuracy at DFT cost

Detection of multi-reference character imbalances enables a transfer learning approach for virtual high throughput screening with coupled cluster accuracy at DFT cost

Exploiting Ligand Additivity for Transferable Machine Learning of Multireference Character across Known Transition Metal Complex Ligands

Relativistic Kramers-Unrestricted Exact-Two-Component Density Matrix Renormalization Group

Contact Info

Product

Resources

About