One slightly unintuitive fact about modern collider physics is that we simulate most things before we measure them. My project is no exception: a lot of the work happens in Monte Carlo (MC) generators long before anyone thinks about a real detector.
For the FCC studies, I mainly use two tools: MadGraph and Herwig. Both are Monte Carlo event generators, but they have different strengths and are designed to work together.
MadGraph is excellent at the “hard scattering” part of a collision – the short-distance interaction where, for example, two quarks annihilate into a Higgs and a Z boson. You specify the process you care about (something like “p p > z h” in its syntax), the model (Standard Model, or in my case the HEL EFT) and the energy. MadGraph then generates parton-level events: four-momenta for the outgoing particles plus a weight, based on the relevant matrix elements.
Herwig is more of a full event generator. It can also calculate hard processes, but its real strength is everything that happens after the hard scattering:
- parton showering, where outgoing quarks and gluons radiate further quarks and gluons,
- hadronisation, where those quarks and gluons are turned into colourless hadrons,
- decays of unstable particles down to the stable ones that actually reach a detector.
The workflow in my project is:
- Use MadGraph and Herwig to compute cross sections for specific HEL settings, at different energies (13 TeV and 85 TeV in the lit review).
- Cross-check that they agree when all Wilson coefficients are set to zero, i.e. the HEL setup should reduce to the Standard Model.
- Switch on a BSM point (for example, nonzero cH, cHW, cHB) and check again that both codes see similar enhancements in the ZH rate.
In the literature review I showed exactly this: for coefficient-zero HEL at 13 TeV, MadGraph and Herwig cross sections agree within tiny uncertainties; the same holds at 85 TeV. For a benchmark BSM point at 85 TeV, both codes predict a huge increase in the cross section and still agree within errors.
This cross-checking is more than just box-ticking. EFT setups are complicated, and you want to be absolutely sure that “HEL in generator A” means the same thing as “HEL in generator B”. If two completely different codes agree on both SM and BSM points, you can be much more confident that any effect you see later in the analysis is physics, not a bug.
Once a configuration is validated, I lean more heavily on Herwig because it produces fully hadronised, detector-level final states. Those are what I feed into Rivet for the boosted-Higgs analysis. MadGraph remains in the background as a reference and for sanity checks when I change Wilson coefficients or energy.
So when I talk about “simulated FCC collisions” on here, what I actually mean is: millions of pseudo-random events generated by MadGraph and Herwig, encoding our best theoretical knowledge plus a parameterised description of possible new physics through HEL. From that starting point, the Rivet analysis tries to answer the question: if the world looked like this particular HEL point, would we be able to tell?
Noot in Circle 