Quickstart

This page gets you from a fresh install to a written ENDF tape in about five minutes. You will run a committed example input file through the command-line interface, learn what each line of console output means, read an annotated minimal classic input file, and find the pointers to the GUI and the rest of the manual. It assumes IRMA is already installed (see Installation) and that you can run python -m irma.

The defaults deserve a word before you start: every value IRMA fills in on its own (the GUI's fresh form, the inputs irma mlip emit writes, any configuration field you leave out) is the production setting from the validation campaigns. Accepting them wholesale gives results of the quality shown in the validation record, not a rough first draft.

Run an input file in one command

IRMA's command line takes an input file and an output path:

python -m irma input.input out.endf

input.input is a LEAPR-style input file (the .input or .leapr extension is conventional, not required).
out.endf is the ENDF-6 File 7 tape IRMA writes. IRMA always writes the file you name on the command line; Card 1's nout field is kept only for LEAPR compatibility and is otherwise ignored.

A real, validated graphite input file ships with the repository, so you can try the full path right now (the path below exists in a clone of the repository; a pip-only install can fetch the file from GitHub first):

python -m irma \
  tests/native_LEAPR_NJOY_ENDF_validation/leapr_decks/tsl-crystalline-graphite.input \
  graphite.endf

This is a classic crystalline-graphite evaluation (a phonon density of states (DOS) supplied in the input file, isotropic Debye-Waller, the same physics level as iel=10 with inelastic_mode=0); it runs in seconds and reproduces the published ENDF/B-VIII.1 graphite tape to 3.4×10⁻⁵ (the graphite entry of the reference set).

If you installed IRMA with pip, the bare irma console script is equivalent to python -m irma, so irma input.input out.endf runs the same evaluation. The two-positional form also has explicit aliases that do exactly the same thing (irma evaluate input.input out.endf and irma run input.input out.endf), handy when you want the intent spelled out in a script.

Reading the console output

IRMA narrates each stage to stdout as it works. For the graphite run the output looks like this (the middle temperatures are omitted here):

IRMA: graphite
  ntempr=10, iprint=2, nphon=100
  mat=30, za=130, isabt=0, ilog=0, iint=0
  awr=11.898, spr=4.73918, npr=1, iel=1, ncold=0, nsk=0
  nalpha=150, nbeta=400, lat=1

  Principal scatterer...
  Temperature 1: 296.00 K
    DW lambda = 0.868056, T_eff = 706.228
  Temperature 2: 400.00 K
    (negative T: reusing previous temperature's scattering-law inputs)
    DW lambda = 1.490620, T_eff = 748.880
  ...
  Temperature 10: 2000.00 K
    (negative T: reusing previous temperature's scattering-law inputs)
    DW lambda = 33.902922, T_eff = 2089.988
  Found 345 Bragg edges below 5 eV
  Read 34 comment cards

  Writing ENDF output...
ENDF output written to graphite.endf
  IRMA complete.

Stage	What it means
`IRMA: <title>`	The banner. IRMA echoes Card 2's quoted title so you can confirm the right input file loaded.
Echoed control cards (`ntempr`, `mat`, `awr`, `nalpha`, ...)	IRMA reads and prints back the values it parsed from Cards 3-7. Use these to verify the parser saw what you intended: a wrong `iel` or `nbeta` shows up here.
`Temperature N: <T> K`	One line per temperature in the input file. The S(α,β) detail block for that temperature is computed next. A negative temperature in the input file prints `(negative T: reusing previous temperature's ...)`, the LEAPR shared-spectrum convention.
`DW lambda = ..., T_eff = ...`	The Debye-Waller lambda and effective temperature computed from that temperature's phonon spectrum. Both end up on the tape: the lambda sets the elastic Debye-Waller suppression, and the effective temperature drives the short-collision-time extension of THERMR (NJOY's thermal processing module) downstream.
`Found <N> Bragg edges below 5 eV`	The coherent-elastic phase. Printed when the principal scatterer has an elastic treatment (`iel` ≥ 1 or `iel=10`).
`Read <N> comment cards`	The trailing quoted comment cards become the ENDF MF1/MT451 description.
`Writing ENDF output...` / `ENDF output written to <path>`	The S(α,β) table and the elastic data are assembled into MF7 and written to the output path.
`IRMA complete.`	Success. The tape at your output path is ready for downstream processing (e.g. NJOY THERMR/ACER).

Three echoed values deserve a note. S(α,β) is the thermal scattering law, tabulated in the dimensionless momentum transfer α and energy transfer β (β = E/kT; the theory page defines the conventions). za=130 is the historical TSL convention for graphite carried by this classic input file (thermal libraries often assign such codes to compound materials), while the generalized iel=10 treatment requires the physical 1000·Z+A form. And iint=0 is Card 4's optional interpolation flag, defaulted here; the input file reference describes it.

MAT, on Card 4, is the ENDF material number: the integer that labels the evaluation in a library, yours to assign.

If the input file is malformed, IRMA stops at the failing card: the console shows the lines above up to that point, then an Input deck error: message that names the offending card, the expected-vs-found values, and the input line number, then exits non-zero. A clean engine-level failure (for example a missing phonopy file) prints IRMA failed: <reason>. No ENDF tape is written in either case.

One hazard lives downstream of a successful run, and you should know about it before taking a coherent tape to NJOY. Stock NJOY2016 THERMR has a cliq bug that turns a correct coherent inelastic_mode = 2 tape (graphite and similar) into ~1e91-barn garbage above ~0.27 eV, or stalls for hours. The bug fires on the shape of the tabulated S(α,β), not on its grid, so no beta-grid choice avoids it; apply the one-line thermr.f90 patch described in NJOY interoperability before processing. inelastic_mode = 0/1 tapes and most materials are unaffected.

A minimal classic input file, annotated

The input file format is NJOY free-format: values are separated by spaces and every card is terminated by /. The simplest useful input file is a classic evaluation, the same physics level as inelastic_mode=0: an isotropic Debye-Waller treatment plus a cubic phonon expansion built from a scalar phonon DOS you supply in the input file. No phonopy is needed.

Here is the card skeleton (graphite-shaped, one temperature). The phonon DOS and grid arrays are shown as ... / placeholders; in a real input file they are the full lists of values.

24 /                          Card 1: nout (LEAPR-compat; output goes to the CLI path)
'graphite' /                  Card 2: quoted title (echoed as the banner)
1 2 100 /                     Card 3: ntempr iprint nphon
30 6012.0 0 0 /               Card 4: mat za isabt ilog (smin and iint defaulted; mat is the evaluation's MAT number)
11.898 4.73918 1 1 0 0 /      Card 5: awr spr npr iel ncold nsk
0 /                           Card 6: nss (0 = no secondary scatterer)
150 400 1 /                   Card 7: nalpha nbeta lat
3.32e-03 ... 6.64e+01 /       Card 8: alpha grid (nalpha values, increasing)
0.0 ... /                     Card 9: beta grid (nbeta values, increasing)
296.0 /                       Card 10: temperature [K]
7.8e-04 351 /                 Card 11: delta ni (DOS energy spacing, point count)
0.0 ... /                     Card 12: rho(1..ni) phonon DOS on the equidistant grid
0.0 0.0 1.0 /                 Card 13: twt c tbeta (translational/continuous weights)
0 /                           Card 14: nd (number of discrete oscillators)
'crystalline graphite' /      Optional MF1/MT451 comment card(s)
/                             A bare slash ends the comment section

What each control card carries, at a glance:

Card	Fields	Role
1	`nout`	Output unit (LEAPR compatibility only).
2	`'title'`	Quoted title; becomes the console banner.
3	`ntempr iprint nphon`	Temperature count, print level, phonon-expansion order.
4	`mat za isabt ilog smin [iint]`	ENDF MAT/ZA, storage flags, and the optional interpolation flag echoed in the log.
5	`awr spr npr iel ncold nsk`	Principal scatterer: mass ratio, free-atom σ, atom count, elastic option, cold-H and pair-correlation options.
6	`nss b7 aws sps mss`	Secondary scatterer block (`nss=0` for none).
7	`nalpha nbeta lat`	Grid sizes; `lat=1` means grids are at the 0.0253 eV reference temperature.
8 / 9	α grid / β grid	Strictly increasing value lists, one `/` ends each card.
10	`T`	Temperature [K]; a negative value reuses the previous block.
11-16	`delta ni`, `rho`, `twt c tbeta`, `nd`, ...	The S(α,β) detail block (DOS, translational, oscillators).

This table is an orientation, not the full specification. Field meanings, defaults, valid ranges, the generalized-elastic cards (iel=10, Cards 6b-6g), and the noncubic inelastic_mode=1/2 workflows are documented card by card in the Input file reference; reach for it whenever a field above is unclear.

Before you write your own input file, know Card 5's iel. It selects the coherent-elastic treatment: 0 = none, 1-6 = the legacy built-in materials (graphite, Be, BeO, Al, Pb, Fe), and 10 = generalized Bragg edges from your own crystal structure. For a new material on a noncubic crystal, iel=10 is recommended; it also makes the anisotropic Debye-Waller treatment and the inelastic_mode=1/2 workflows, computed from a phonopy calculation, available. The example above uses the built-in iel=1 graphite tables, which keeps the input file self-contained.

Skipping the grid: let IRMA build it

Cards 8 and 9 are explicit α and β grids. The input file format has no auto-grid flag. Automatic grids are built in the GUI (the Grids tab) or from Python via the irma.core.grids functions; either way the result is an explicit grid that lands on Cards 7–9 of the input file (or in the saved JSON configuration). The Automatic grids page explains the controls (N lower, N phonon, N upper, and the linear-in-Q alpha parameters). The default alpha grid is linear in momentum transfer Q (dQ = 0.05 1/Å up to Q = 12 1/Å, then a logarithmic tail), which resolves the thermal upscatter windows that a beta-mirrored (recoil-relation) alpha grid under-samples.

First GUI launch

To point and click, or to inspect an unfamiliar input file field by field, launch the graphical interface:

python -m irma --gui

The irma-gui console script does the same thing. From the GUI you can define the crystal structure, set scattering parameters and grids, point at phonopy data, run the calculation with a progress display, and save or load the whole configuration as JSON. Existing .input / .leapr input files can be imported via File > Import Input File to populate every field. The GUI is covered in detail on the Graphical interface page.

Predict an instrument spectrum

The same install (plus pip install -e ".[spectra]") also predicts what an inelastic neutron scattering (INS) instrument measures. One command produces a VISION-resolved spectrum from a phonopy calculation (in the command, --scatterer decodes as symbol, bound cross section [b], atomic weight ratio, b_coh [fm], sigma_inc [b]; IRMA's built-in table supplies these values for any element):

python -m irma spectra vision --phonopy-yaml phonopy.yaml --inelastic-mode 2 \
    --scatterer "C,5.551,11.898,6.646,0.001" --temperature 300 -o vision.csv

or, with no eigenvectors at all, from a phonon DOS via a config file. The Neutron scattering spectra page has the end-to-end example and the full CLI/config reference; DOS-based spectra (mode 0) covers the DOS-only workflow; ready-to-run configurations live under examples/spectra/.

Where to go next

Input file reference: the authoritative card-by-card spec, including iel=10 and the noncubic modes.
Automatic grids: let IRMA build α/β grids for you.
Scattering modes: the elastic iel options (including the generalized iel=10 Bragg edges for any crystal), SEF (single-channel elastic format) vs MEF, and when to use inelastic_mode 0, 1, or 2.
Neutron scattering spectra: instrument-resolved INS spectra and S(Q,E) maps with irma spectra; mode 0's DOS-only workflow has its own page.
NJOY interoperability: preparing tapes for THERMR/ACER.
Validation: how IRMA's results compare against NJOY, Euphonic, OCLIMAX, and measured graphite cross sections.