Skip to content

Simulation

Running the calculation

Every wizard ends on the generated Python. It is plain ASE, no wrapper, and the same script runs locally, in a queue, or on a cluster over SSH.

Common workflows

The four calculations most people open Calango Studio to run.

Band structure

A k-path taken from the Brillouin-zone viewer, a calculator, and the run returns E(k) with the projected density of states beside it. Orbital and site projections draw as fatbands. The irreducible-representation labels at each high-symmetry point are computed from the wavefunctions rather than looked up from the space group, so they stay correct when a defect or a strain lowers the symmetry. Supercell calculations unfold onto the primitive zone, which is what makes an alloy or defect band structure legible at all.

Phonons

Finite displacements in a supercell, reduced by spglib to the irreducible set, give the force constants and from them the dispersion, the phonon DOS and the thermodynamic functions. Born effective charges supply the LO–TO splitting for polar materials. Imaginary branches are reported rather than hidden. They are the signature of a structure that is not at a minimum, and the most useful thing the calculation can tell you.

Optical properties

The dielectric function ε(ω) from the independent-particle or RPA response, with the intraband Drude term added for metals, and from it the absorption coefficient, reflectivity and refractive index. Non-centrosymmetric structures also give second-harmonic generation and the shift current. Monolayers use a 2D formulation, because the bulk expression depends on a vacuum spacing that is a convergence parameter, not a property.

Wannierization

Maximally localized Wannier functions from the converged Bloch states, with the spread minimization reported per iteration so a failed disentanglement is visible instead of silent. The resulting tight-binding model interpolates bands onto a dense k-mesh for almost nothing, which is what puts Fermi surfaces, Berry curvature and topological invariants within reach.

Advanced workflows

Built on the same ground state, and driven from the same wizard.

Electron–phonon coupling

The Eliashberg function α²F(ω), λ, transport λtr and ρ(T), and Allen–Dynes Tc over a range of μ*.

Absolute free energies

Thermodynamic integration from a reference whose free energy is known in closed form, with propagated error bars.

Alloy thermodynamics

Cluster expansion fitted by LASSO, ridge or ARD, feeding a Cluster Variation Method solver for order–disorder transitions.

Quasiparticle corrections

G₀W₀ on top of the DFT ground state, through GPAW or Yambo.

Topology and Fermi surfaces

Berry curvature, topological invariants and Fermi-surface sheets from the Wannier-interpolated model.

Molecular dynamics and NEB

NVE/NVT/NPT with the full ASE thermostat set, simulated annealing, Monte Carlo, and minimum-energy paths.

Core-level spectroscopy

X-ray absorption spectra from the same converged ground state.

Hubbard U by linear response

U determined self-consistently rather than fitted to reproduce an answer you already wanted.

Calculator-agnostic

Every wizard shares the same staged flow, moving through calculator, settings, task and script review. Each engine exposes the parameters it actually has, not a lowest-common-denominator form.

Density functional theory
  • GPAW
  • Quantum ESPRESSO
  • VASP
  • SIESTA
Quantum chemistry
  • ORCA
Semi-empirical tight binding
  • xTB / GFN2
  • DFTB+
Classical potentials
  • LAMMPS
  • GROMACS
  • EMT
  • Lennard-Jones
  • ASAP
Machine-learning potentials
  • MACE
  • DeepMD-kit
  • NequIP
  • Allegro
  • CHGNet
  • MatterSim
  • FAIRChem

Getting a run to a machine

Chaining, watching and submitting are the parts between pressing run and having a result.

Orchestration

The Orchestration dock chains calculations into a DAG on a node-graph canvas, such as relax, then converge, then compute a spectrum. It passes geometries and ground states between nodes, with per-node status and failure propagation.

Live monitoring

Energy, temperature, force and pressure stream into plots as the job runs; MD trajectories stream frame-by-frame into the viewport while they are being computed. Jobs queue instead of refusing, and a process manager keeps every run's logs, metrics and artifacts one click away.

Remote HPC execution

Connect over SSH and Calango Studio stages the same script and structure, generates a SLURM/PBS/SGE wrapper, uploads, submits, polls the queue, streams remote logs and downloads results automatically when the job finishes.

How it is put together

core/
Qt-free data model, geometry, analysis algorithms and the ASE script generators.
python_bridge/
Converts core::Structure ↔ ase.Atoms via pybind11. Python types never escape it.
render/
Instanced OpenGL renderer. Reads the model, never mutates it.
jobs/ + remote/
Subprocess and SSH job execution with a stdout marker protocol for live updates.
gui/
Main window, viewport, docks, wizards and viewers.
ui/
SVG icon theming.

Simulations never run inside the GUI process: local jobs are isolated QProcess subprocesses and remote I/O lives in a paramiko helper process, so a crashed calculation never takes the application down.

Keep reading

Representations, spatial references, volumetric data, rendering.

Symmetry labels, RDF, structure factor, coordination numbers.

Try it on your own structure

Build from source, or read the manual first.