Rare flavor decays and effective Hamiltonians

At energies well below the W mass, a process can be expanded in local operators built from the remaining fields. Their Wilson coefficients encode heavy-particle effects, while matrix elements describe the chosen external hadrons. For an illustrative bottom-to-strange transition choose $\mathcal H_{\mathrm{eff}}=-(4G_F/\sqrt2)V_{tb}V_{ts}^*\sum_i C_iO_i+\mathrm{h.c.}$ as an operator-basis normalization. The displayed factorization is conventional; CKM ratios for additional contributions may remain inside coefficients. A number quoted without its operator definitions and extracted prefactors is not a portable observable.

For example, define the electromagnetic dipole operator using the running bottom mass, a chirality projector, and the photon field strength. Each quark field has mass dimension three halves, the field strength dimension two, and the inserted mass dimension one, giving operator dimension six. The Fermi constant has dimension minus two, so the Hamiltonian density has dimension four with dimensionless coefficients. Including the mass or loop factor in a different place changes the numerical coefficient convention without changing the underlying decay amplitude.

Standard Model flavor-changing neutral currents have no tree-level neutral gauge coupling between distinct same-charge mass eigenstates. A loop transition can instead contain $\sum_{i=u,c,t}V_{ib}V_{is}^*f(m_i^2/m_W^2)$. For different external flavors, column orthogonality makes the CKM sum zero. Any part of f independent of the internal flavor therefore cancels. If all internal masses were equal, the entire displayed contribution would vanish. This GIM mechanism explains a structural suppression, not a universal statement that every rare process is numerically dominated by the same internal quark.

Subtracting an arbitrary common value of the loop function exposes the cancellation explicitly and often clarifies which mass differences matter. Heavy-particle behavior then depends on the specific operator and symmetry structure; it cannot be inferred solely from dimensional counting. A full prediction also requires matching at a high scale and renormalization-group evolution to a hadronic scale. Operators generally mix under this evolution. Keeping one coefficient while neglecting its coupled matrix elements can spoil the cancellation of artificial scale dependence and create a misleading apparent effect.

The rate is a squared coherent amplitude, so changing the sign or phase of one contribution may leave a rate unchanged. Angular observables, CP asymmetries, other decay modes, and spectral shapes can help resolve these degeneracies. Long-distance hadronic effects may imitate some short-distance coefficient shifts, particularly when resonances or nonlocal weak-current insertions contribute. Quantifying that ambiguity is part of the prediction. A clean mathematical coefficient fit becomes a new-interaction claim only after experimental and theoretical uncertainties are propagated consistently.

A heavy scale can sometimes be bounded through a coefficient proportional to coupling squared divided by mass squared, but an unknown coupling prevents a unique mass inference. Correlated operators, interference, and matching assumptions further affect that translation. Our hand example deliberately holds all matrix elements and phase space fixed and retains one real coefficient. It illustrates sensitivity and a sign degeneracy, not a numerical bound on a particular particle model. This distinction keeps an effective description useful without assigning it information that the chosen observable never supplied.