Results and Molecule Data¶

After runner.run(), the Runner keeps the active calculation object at runner.mol. This Molecule object stores the parsed configuration, native OpenQP data handle, scalar results, and NumPy-accessible arrays.

runner.run()

mol = runner.mol
print(mol.config["input"]["runtype"])
print(mol.get_scf_energy())
print(mol.get_system())

Runner Summary¶

runner.results() returns a compact dictionary for Python-to-Python communication:

Key	Meaning
`atoms`	Atomic numbers.
`system`	Cartesian coordinates in Bohr.
`energy`	Energy array or scalar data accumulated by the workflow.
`grad`	Gradient data when available.
`dcm`	Derivative-coupling matrix data when available.
`nac`	Nonadiabatic coupling data when available.
`soc`	Spin-orbit coupling data when available.
`data`	Raw OpenQP data tags serialized into Python lists.

For file-style result export, use runner.mol.get_results(). It returns a JSON friendly dictionary with atoms, coordinates, total energy, symmetry metadata, TDDFT energies, gradients, NAC, SOC, Hessian data, MP2 correlated totals, and MRSF-EKT records when present.

Wavefunction calculations additionally populate get_results()["energies"] with ordered total energies in Hartree. For FCI, CASCI, and CASSCF this is the per-root total-energy list; a state-averaged CASSCF calculation still reports the solved roots even though its scalar objective is a weighted average. For CASPT2, NEVPT2, and QDPT variants the values are the corrected total energies, not the perturbative corrections alone, ordered by the requested target roots (or by ascending eigenvalue for a diagonalized multistate model space). Corresponding native tags include OQP::FCI_ENERGIES, the CASCI/CASSCF energy tags, and OQP::CASPT2_ENERGIES; mol.energies is the source used for this portable result key.

For dedicated SA-CASSCF gradients, gradient_state=averaged stores the weighted-objective derivative in gradient row zero, but the energy array still contains the individual roots; there is no matching weighted-objective energy row. An integer gradient_state=J stores the analytic Z-vector derivative in physical CI-root row J, and the corresponding reported energy is that same root. Input validation requires the direct-gradient or optimizer selector to address the same row.

For method=mp2, the exported energy is the correlated MP2 total after the SCF reference energy and MP2 correlation are combined. The run log also prints the reference SCF energy plus same-spin and opposite-spin MP2 components.

For method=ccsd and method=ccsd(t), the exported energy follows the same rule: it is the correlated total, not the reference. ccsd exports E(SCF) + E(CCSD correlation), and ccsd(t) exports E(SCF) + E(CCSD correlation) + E((T) correction) — so switching between them changes the value under the same key, and the key alone does not say which method produced it. The individual pieces are not exported separately; the run log prints them as a labelled table:

Log line	Meaning
`E(reference, SCF)`	The converged HF reference energy.
`E(CCSD, correlation)`	The CCSD correlation energy alone.
`E(CCSD, total)`	Reference plus CCSD correlation.
`E((T), correction)`	The perturbative triples correction (`ccsd(t)` only).
`E(CCSD(T), correlation)`	CCSD correlation plus triples (`ccsd(t)` only).
`E(CCSD(T), total)`	The exported total for `ccsd(t)`.

Consumers that need the decomposition rather than the total should record which method they requested and read these lines. The exported keys may gain a dedicated coupled-cluster breakdown in a later release.

When the corresponding property is requested via [properties] scf_prop, get_results() also includes the following (identically for file-based and input_dict/scripting-API runs):

Key	Requested by	Meaning
`dipole`	`el_mom`	Electric dipole vector (a.u.).
`mulliken_charges`	`mulliken`	Mulliken atomic partial charges (e).
`lowdin_charges`	`lowdin`	Löwdin atomic partial charges (e).
`resp_charges`	`resp`	RESP/ESP-fitted atomic charges (e).
`nmr_shielding`	`nmr`	Isotropic NMR shielding (ppm), shape `(natom, 5)`: columns are `dia`, `para_uncoupled`, `para_coupled`, `total_uncoupled`, `total_coupled`.

nac is populated with the NACME derivative-coupling matrix for runtype=nacme and is empty otherwise.

Common Molecule Methods¶

Method	Returns
`get_atoms()`	Atomic numbers as a copied NumPy array.
`get_mass()`	Atomic masses as a copied NumPy array.
`get_system()`	Cartesian coordinates in Bohr as a copied NumPy array.
`get_scf_energy()`	Total SCF energy.
`get_scf_energy("all")`	Dictionary of available SCF energy components.
`get_grad()`	Gradient in Hartree/Bohr.
`get_hess()`	Cartesian Hessian matrix when available.
`get_soc()`	SOC eigenvalues in cm-1 when available.
`get_data()`	Raw registered OpenQP data tags converted to lists.
`get_results()`	JSON-friendly calculation summary.
`write_molden(filename)`	Writes Molden orbital data.

Dynamic Tag Accessors¶

OpenQP registers several native data tags and creates matching Python getter and setter methods. The method names are lower-case versions of the tag names.

Tag	Getter	Typical content
`OQP::DM_A`	`get_dm_a()`	Alpha density matrix.
`OQP::DM_B`	`get_dm_b()`	Beta density matrix.
`OQP::FOCK_A`	`get_fock_a()`	Alpha Fock matrix.
`OQP::FOCK_B`	`get_fock_b()`	Beta Fock matrix.
`OQP::E_MO_A`	`get_e_mo_a()`	Alpha orbital energies.
`OQP::E_MO_B`	`get_e_mo_b()`	Beta orbital energies.
`OQP::VEC_MO_A`	`get_vec_mo_a()`	Alpha MO coefficients.
`OQP::VEC_MO_B`	`get_vec_mo_b()`	Beta MO coefficients.
`OQP::Hcore`	`get_hcore()`	Core Hamiltonian.
`OQP::SM`	`get_sm()`	AO overlap matrix.
`OQP::TM`	`get_tm()`	Kinetic-energy matrix.
`OQP::td_energies`	`get_td_energies()`	TDHF/TDDFT excitation energies.
`OQP::td_mrsf_density`	`get_td_mrsf_density()`	MRSF density data.
`OQP::td_states_overlap`	`get_td_states_overlap()`	State-overlap matrix.
`OQP::soc_eval`	`get_soc_eval()`	SOC eigenvalues.

Each dynamic tag also has a setter named set_<tag>(), for example set_dm_a(array). Setters mutate the native data store and are intended for advanced workflows such as custom SCF experiments or internal coupling between OpenQP modules.

Units and Shapes¶

Coordinates returned by get_system() are in Bohr.
Gradients returned by get_grad() are in Hartree/Bohr.
get_soc() reports SOC eigenvalues in cm-1.
Many raw tag arrays are stored in flattened native order. Convert or reshape them using the current basis size, atom count, or state count before analysis.

When possible, prefer runner.results() or mol.get_results() for automated post-processing because those summaries are less tied to native storage layout.