Search

We demonstrate how to extend the pair natural orbital (PNO) methodology for excited states, presented in a previous work for the perturbative doubles correction to configuration interaction singles (CIS(D)), to iterative coupled cluster methods such as the approximate singles and doubles model CC2. The original O(N5) scaling of the PNO construction is reduced by using orbital-specific virtuals (OSVs) as an intermediate step without spoiling the initial accuracy of the PNO method. Furthermore, a slower error convergence for charge-transfer states is analyzed and resolved by a numerical Laplace transformation during the PNO construction, so that an equally accurate treatment of local and charge-transfer excitations is achieved. With state-specific truncated PNO expansions, the eigenvalue problem is solved by combining the Davidson algorithm with deflation to project out roots that have already been determined and an automated refresh with a generation of new PNOs to achieve self-consistency of the PNO space. For a large test set, we found that truncation errors for PNO-CC2 excitation energies are only slightly larger than for PNO-CIS(D). The computational efficiency of PNO-CC2 is demonstrated for a large organic dye, where a reduction of the doubles space by a factor of more than 1000 is obtained compared to the canonical calculation. A compression of the doubles space by a factor 30 is achieved by a unified OSV space only. Moreover, calculations with the still preliminary PNO-CC2 implementation on a series of glycine oligomers revealed an early break even point with a canonical RI-CC2 implementation between 100 and 300 basis functions. [ABSTRACT FROM AUTHOR]

We present an algorithm for computing explicitly correlated second- and third-order Mo\ller-Plesset energies near the basis set limit for large molecules with a cost that scales formally as N4 with system size N. This is achieved through a hybrid approach where locality is exploited first through orbital specific virtuals (OSVs) and subsequently through pair natural orbitals (PNOs) and integrals are approximated using density fitting. Our method combines the low orbital transformation costs of the OSVs with the compactness of the PNO representation of the doubles amplitude vector. The N4 scaling does not rely upon the a priori definition of domains, enforced truncation of pair lists, or even screening and the energies converge smoothly to the canonical values with decreasing occupation number thresholds, used in the selection of the PNO basis. For MP2.5 intermolecular interaction energies, we find that 99% of benchmark basis set limit correlation energy contributions are recovered using an aug-cc-pVTZ basis and that on average only 50 PNOs are required to correlate the significant orbital pairs. [ABSTRACT FROM AUTHOR]

We explore how in response calculations for excitation energies with wavefunction based (e.g., coupled cluster) methods the number of double excitation amplitudes can be reduced by means of truncated pair natural orbital (PNO) expansions and localized occupied orbitals. Using the CIS(D) approximation as a test model, we find that the number of double excitation amplitudes can be reduced dramatically with minor impact on the accuracy if the excited state wavefunction is expanded in state-specific PNOs generated from an approximate first-order guess wavefunction. As for ground states, the PNO truncation error can also for excitation energies be controlled by a single threshold related to generalized natural occupation numbers. The best performance is found with occupied orbitals which are localized by the Pipek-Mezey localization. For a large test set of excited states we find with this localization that already a PNO threshold of 10-8-10-7, corresponding to an average of only 40-80 PNOs per pair, is sufficient to keep the PNO truncation error for vertical excitation energies below 0.01 eV. This is a significantly more rapid convergence with the number doubles amplitudes than in domain-based local response approaches. We demonstrate that the number of significant excited state PNOs scales asymptotically linearly with the system size in the worst case of completely delocalized excitations and sub-linearly whenever the chromophore does not increase with the system size. Moreover, we observe that the flexibility of state-specific PNOs to adapt to the character of an excitation allows for an almost unbiased treatment of local, delocalized and charge transfer excited states. [ABSTRACT FROM AUTHOR]

We explore using a pair natural orbital analysis of approximate first-order pair functions as means to truncate the space of both virtual and complementary auxiliary orbitals in the context of explicitly correlated F12 methods using localised occupied orbitals. We demonstrate that this offers an attractive procedure and that only 10-40 virtual orbitals per significant pair are required to obtain second-order valence correlation energies to within 1-2% of the basis set limit. Moreover, for this level of virtual truncation, only 10-40 complementary auxiliary orbitals per pair are required for an accurate resolution of the identity in the computation of the three- and four-electron integrals that arise in explicitly correlated methods. [ABSTRACT FROM AUTHOR]