How do I ...?
JAC ships with several hundred worked example branches. This page is not an index of them — it is a short, deliberately incomplete list of the questions users ask most, each pointing at one example that answers it. Start here; the Examples page is the full working reference when you need something not listed.
Every entry below points at an example branch that carries a Last successful date, so the test suite guards it.
Structure and levels
How do I get the level structure and energies of an ion? example-Ma.jl — computeLevelEnergies(...) from the ForPedestrians module, the shortest path from a configuration to a level scheme, and the one to start with. example-Ac.jl shows the fuller route through a configuration-interaction expansion for open-shell multiplets, which is what almost everything else on this page builds on.
How do I improve the orbitals with correlation? example-Ai.jl — restricted-active-space (RAS) expansions, adding correlation layer by layer. See also Self-consistent fields for which SCF route to use and what it costs.
How do I choose a radial grid — and know when it is wrong? example-Ab.jl — B-spline primitives and one-particle spectra in a local potential. Basics.recommendedGrid(configs, nuclearModel) chooses a sensible box for you, and is the recommended way to build one. A box that is far too large starves the basis exactly as badly as one that is too small, and the symptom looks like an angular-momentum bug rather than a grid problem.
How do I include the Breit interaction? example-Ad.jl — the frequency-independent Breit interaction in the CI step.
Radiative processes
How do I get transition rates and oscillator strengths (Einstein A)? example-Ca.jl for the Einstein module on a single multiplet, or example-Da.jl for PhotoEmission between separately generated initial and final multiplets.
How do I get a photoexcitation or photoionisation cross section? example-Db.jl for photoexcitation, example-Dc.jl for photoionisation. Both need a continuum orbital, so the radial grid matters more here than for bound-state work.
How do I get a radiative-recombination cross section? example-Dd.jl — PhotoRecombination, the inverse of photoionisation.
Electron-collision processes
How do I get an Auger rate and a level width? example-De.jl — AutoIonization. The rate is a width; JAC prints both.
How do I get dielectronic-recombination strengths and rate coefficients? example-Df.jl — resonant capture followed by radiative stabilisation.
How do I get an electron-impact excitation cross section? example-Dl.jl — ImpactExcitation.
Level properties
How do I get hyperfine A and B constants? example-Cb.jl — the Hfs module, hyperfine parameters and the hyperfine representation.
How do I get Landé g-factors and Zeeman splittings? example-Cd.jl — the LandeZeeman module.
How do I get an isotope shift? example-Cc.jl — mass and field shift from a single multiplet.
Larger computations
How do I get charge-state distributions after inner-shell ionisation? example-Fb.jl — a decay cascade followed through to the final charge states.
How do I get plasma rate coefficients? example-Jb.jl — collisional-radiative modelling on top of the atomic data.
And the question worth asking of every result
How do I know whether to trust what came back?
This is the one entry here that is about judgement rather than capability, and JAC answers a good part of it by itself:
- Gauge agreement. Any radiative quantity is computed in both Coulomb and Babushkin gauge. They agree when the wave functions are good and diverge when they are not.
PhotoEmissionprints the ratio and a verdict. - Cowan's cancellation factor, printed beside it, says whether the amplitude survived as a small difference of large terms — a value below 0.1 means the number carries little information, whatever the gauge agreement.
- The SCF route reports on itself — whether it converged, how far it got, and how far its answer can be trusted; see Self-consistent fields.
None of these proves a number right: two gauges can agree on a wrong answer when the same correlation is missing from both. They are reliable in the other direction — a bad indicator is good evidence of a bad number, and that is what makes them worth reading before you publish anything.