het-agi/collideragent/calchep-generator

CalcHEP Generator

Overview

This skill generates CalcHEP model directories from FeynRules .fr files using remote Mathematica execution via the Magnus cloud platform. CalcHEP models are the standard interchange format between model-building tools (FeynRules) and dark matter calculation packages (micrOmegas, CalcHEP).

Workflow

Step 1: Prepare the .fr File

Ensure the .fr file is validated (via feynrules-model-validator skill) and saved to disk. Know the Lagrangian symbol name (e.g., LSnew, LBSM).

Step 2: Generate the CalcHEP Model

Execute the generate-calchep blueprint using magnus run (see magnus skill):

magnus run generate-calchep -- --model path/to/model.fr --lagrangian LSnew --output path/to/MyModel_CH

Parameters:

The .fr file (and .rst if provided) is automatically uploaded via the FileSecret mechanism. On success, the CalcHEP directory is automatically downloaded to --output.

WARNING: If --output points to an existing directory, it will be deleted and replaced by the download.

Model type is auto-detected:

• With M$GaugeGroups: loaded directly as standalone model
• Without M$GaugeGroups: BSM extension — SM.fr and SM restrictions (Massless.rst, DiagonalCKM.rst) are loaded automatically

Step 3: Verify CalcHEP Output

Check the result JSON (magnus job result <job-id>):

• success (bool): Whether generation succeeded
• calchep_path (str): Path to the downloaded CalcHEP directory

Step 4: Read CalcHEP Output Files

After successful generation, the CalcHEP directory contains .mdl files used by CalcHEP and micrOmegas:

• vars1.mdl — independent variables and parameters
• func1.mdl — dependent parameters (functions of independent variables)
• prtcls1.mdl — particle definitions (name, spin, mass, width, color, charge)
• lgrng1.mdl — interaction vertices (Lagrangian terms)
• extlib1.mdl — external library references (if any)

These files are plain text in CalcHEP's proprietary format — they are not Python files like UFO.

Z₂-Odd Convention (for micrOmegas consumption)

If this model will be passed to micrOmegas for dark-matter calculations, every Z₂-odd particle (the DM candidate plus any coannihilation partners) must appear with a leading `~` in the particle-name (`P`) column of `prtcls1.mdl`. micrOmegas identifies the DM sector purely by this naming convention — it does not read a separate Z₂ quantum number.

Examples from shipped micrOmegas models:

After `generate-calchep` succeeds, verify prtcls1.mdl lists ~-prefixed names for every particle that should be Z₂-odd. Hand-editing prtcls1.mdl after the fact is not a fix: the edit is lost the next time generate-calchep runs.

Current repo limitation: this blueprint calls plain WriteCHOutput[...] and does not inject a separate odd-particle list. So the only reliable contract is the generated CalcHEP output itself. A FeynRules-side recipe that is verified to work in the current pipeline is to place the tilde directly in ParticleName (and AntiParticleName for non-self-conjugate fields). Still, treat the generated prtcls1.mdl as the final authority, because export behavior is model-dependent. Likewise, do not assume QuantumNumbers -> {Z2 -> -1} alone is sufficient: custom quantum numbers require explicit model-level declarations, and even then the CalcHEP exporter may not map them to micrOmegas odd-particle names.

Practical rule: iterate on the .fr, rerun generate-calchep, and re-check prtcls1.mdl until the exported names are right. If you need a starting point, copy the naming strategy from a known-good dark-matter FeynRules model rather than inventing a new convention.

Without this convention, sortOddParticles in the downstream micrOmegas run will fail to locate a DM candidate and every dark-matter observable will silently be zero or NaN.