Nuclear Forces and the Chiral Expansion

Chiral nuclear EFT orders irreducible pion-exchange and contact kernels by $Q/\Lambda_b$, then iterates the promoted kernel through a few-body equation because near-on-shell nucleon propagation is infrared enhanced. Two-, three-, and higher-body forces follow a declared topology count, but their potentials are regulator- and scheme-dependent intermediaries; only renormalized scattering amplitudes, bound-state properties, and current matrix elements are observables.

Required background. Chiral EFT for Pions and Nucleons supplies the pion–nucleon vertices and recoil choices; Nuclear and Few-Body EFT Architecture supplies the general promoted-iteration and matching logic.

Irreducible kernels and enhanced iteration

Take nucleons and pions as explicit degrees of freedom, with

$$Q\in\{|\mathbf p|,|\mathbf p'|,m_\pi\}\ll\Lambda_b.$$

The exact contents of $\Lambda_b$ depend on which resonances are explicit and on the regulator strategy. Electromagnetic and isospin-breaking effects are separate expansions unless included explicitly.

A two-nucleon intermediate state near threshold has kinetic energy $O(Q^2/M_N)$. Its propagator scales as $M_N/Q^2$, rather than $1/Q$ for an irreducible pion–nucleon state. Combined with $d^3q/(2\pi)^3$, a reducible loop contributes the characteristic factor $M_NQ/(4\pi)$. If

$$V_{\rm LO}\sim\frac{4\pi}{M_NQ},$$

then every insertion $V_{\rm LO}G_0$ is $O(1)$ and the geometric series must be summed:

$$T_\Lambda(E)=V_\Lambda+V_\Lambda G_0(E)T_\Lambda(E), \qquad G_0(E;\mathbf q)=\frac{1}{E-\mathbf q^2/M_N+i0}.$$

The Lippmann–Schwinger equation is not itself a power counting. The counting must state which pieces of $V$ are iterated, which are perturbative insertions, and which counterterms accompany that choice. Weinberg’s original separation of irreducible kernels from enhanced iteration is developed in Weinberg 1990, pp. 288–292 and Weinberg 1991, pp. 3–12.

Chiral index and the force hierarchy

For an irreducible diagram with $A$ nucleons, $C$ separately connected pieces, $L$ loops, and vertices $v$, use the declared convention

$$\nu=-2+2A-2C+2L+\sum_v\Delta_v, \qquad \Delta_v=d_v+\frac{n_v}{2}-2.$$

$d_v$ counts derivatives and pion-mass powers, with one light-quark-mass insertion counting as two; $n_v$ counts nucleon fields. Chiral symmetry gives $\Delta_v\ge0$ for the leading strong vertices in the standard basis.

For a connected two-nucleon kernel, $A=2$ and $C=1$, so

$$\nu=2L+\sum_v\Delta_v.$$

The hierarchy then follows directly:

Relative index	Representative two-nucleon content	Counting check
$\nu=0$	One-pion exchange; two nonderivative $NN$ contacts	$L=0$ and every $\Delta_v=0$
$\nu=2$	Leading two-pion exchange; two-derivative contacts	Either $L=1$ with leading vertices or $\sum\Delta_v=2$
$\nu\ge3$	Subleading pion vertices, loops, relativistic corrections, and additional contacts as allowed	Every insertion and loop raises the declared index

At leading order,

$$V_{2N}^{(0)}=C_S+C_T\boldsymbol\sigma_1\cdot\boldsymbol\sigma_2 -\frac{g_A^2}{4F_\pi^2} \frac{(\boldsymbol\tau_1\cdot\boldsymbol\tau_2) (\boldsymbol\sigma_1\cdot\mathbf q) (\boldsymbol\sigma_2\cdot\mathbf q)} {\mathbf q^2+m_\pi^2},$$

projected into the allowed spin, isospin, and partial-wave channels. The constants $C_S,C_T$ are fitted in a stated regulator and normalization; they are not observable contact “forces” independent of that setup.

In the standard delta-less Weinberg organization, the first nonvanishing three-nucleon force occurs at $\nu=3$ and has three topologies,

$$V_{3N}^{(3)}=V_{2\pi}(c_i)+V_{1\pi\text{-ct}}(c_D)+V_{\rm ct}(c_E).$$

The $c_i$ also enter subleading pion–nucleon and two-pion-exchange physics; $c_D$ participates, convention dependently, in a short-range axial current. These shared constants create correlations across forces and currents. The statement “three-body forces are higher order” does not mean that a three-body counterterm can never be promoted: pionless systems with resonant two-body interactions provide the explicit counterexample on Three-Body Renormalization and Universality.

Regulator, renormalization, and promotion

A typical regulated kernel is

$$V_\Lambda(\mathbf p',\mathbf p) =f_\Lambda(p')V(\mathbf p',\mathbf p)f_\Lambda(p),$$

with $f_\Lambda(0)=1$ and suppression for momenta near $\Lambda$. Regulator shape and cutoff are part of the calculation, not additional physics. After all low-energy constants at the claimed order are refitted, an observable $X^{(k)}$ should obey

$$\frac{dX^{(k)}}{d\Lambda} =O\!\left[X_{\rm ref}\left(\frac{Q}{\Lambda_b}\right)^{k+1}\frac1\Lambda\right]$$

over a declared window where $Q\ll\Lambda$ and the regulator does not resolve physics beyond the EFT. This schematic equation is a scaling test, not a probability distribution.

Iteration can change ultraviolet scaling. In attractive channels the tensor part of one-pion exchange is singular, and an iterated kernel may need a contact earlier than naive dimensional analysis predicts. If cutoff dependence survives at the same size as retained terms, the response is to revise the counting or regulator claim—not to call the spread a truncation band. The channel-by-channel promotion test was demonstrated in a sharp-cutoff analysis by Nogga, Timmermans, and van Kolck 2005, pp. 054006-1–054006-9.

Two logically different regulator strategies must not be mixed:

• In a finite-cutoff EFT, choose $\Lambda$ below or around the breakdown region, include the counterterms assigned by the counting, and demand residual variations of omitted-order size.
• In a renormalization-group limit, ask whether observables approach a limit as $\Lambda\to\infty$ after a specified set of counterterms runs. A finite-cutoff fit does not establish this stronger statement.

The broad chiral hierarchy and its regulator qualifications are reviewed by Epelbaum, Hammer, and Meißner 2009, §§III–IV.

Potentials, unitary transformations, and observables

An energy-independent potential depends on field variables, off-shell continuation, unitary transformations, and regulator. If $U$ is a short-range unitary transformation,

$$H'=UHU^\dagger, \qquad |\Psi'\rangle=U|\Psi\rangle, \qquad J'=UJU^\dagger,$$

then spectra and consistently transformed matrix elements are unchanged even though the displayed two- and three-body potentials differ. Dropping the induced many-body force or current after transforming $H$ breaks that equivalence.

The observable workflow is therefore:

1. declare degrees of freedom, counting, regulator, cutoff window, and subtraction scheme;
2. construct the irreducible two- and many-body kernels through order $k$;
3. fit the required low-energy constants to a named calibration set with covariance;
4. solve the regulated few-body equation to numerical accuracy below the EFT remainder;
5. calculate held-out scattering, binding, or response observables with consistently transformed currents;
6. vary cutoff, regulator shape, order, and numerical resolution independently.

A reproducible calculation checks the exact index convention above: leading one-pion exchange and a zero-derivative contact have $\nu=0$, whereas a leading one-loop two-pion graph and a two-derivative contact have $\nu=2$.

Independent checks

Check	Passing result	Diagnostic failure
Dimensions	$V$ has mass dimension $-2$ and $VG_0d^3q$ has the same dimension as $V$	State normalization or loop measure mismatch
Elastic unitarity	A Hermitian kernel and exact LS solution give the correct on-shell discontinuity	Inconsistent boundary condition, omitted channel, or perturbative insertion used nonperturbatively
Cutoff stability	Refitted observables vary only at first-omitted order	Missing or incorrectly counted counterterm
Order pattern	Successive increments scale with the declared $Q/\Lambda_b$ in held-out observables	Breakdown scale, counting, or regulator contamination is wrong
Many-body rank	Induced and explicit $3N$ terms accompany transformations of $2N$ interactions	Scheme dependence leaks into observables
Force–current consistency	Current and Hamiltonian satisfy the relevant continuity or chiral Ward identity	Regulator or operator-order mismatch

Common pitfalls

Calling a potential observable. Off-shell matrix elements and separate topology contributions change under field redefinitions and unitary transformations. Compare phase shifts, pole positions, energies, or consistently transformed response functions.

Using cutoff spread as a confidence interval. Variation is an essential renormalization diagnostic, but a probability statement requires an explicit statistical model with calibrated coverage.

Iterating a truncated subleading potential without checking induced orders. Iteration generates arbitrarily high powers and can expose new ultraviolet divergences. State whether subleading pieces are perturbative or fully iterated and verify the corresponding counterterm set.

References

• Epelbaum, Evgeny, Hans-Werner Hammer, and Ulf-G. Meißner. “Modern Theory of Nuclear Forces.” Reviews of Modern Physics 81 (2009): 1773–1825. DOI.
• Nogga, Andreas, Robert G. E. Timmermans, and U. van Kolck. “Renormalization of One-Pion Exchange and Power Counting.” Physical Review C 72 (2005): 054006. DOI.
• Weinberg, Steven. “Nuclear Forces from Chiral Lagrangians.” Physics Letters B 251 (1990): 288–292. DOI.
• Weinberg, Steven. “Effective Chiral Lagrangians for Nucleon-Pion Interactions and Nuclear Forces.” Nuclear Physics B 363 (1991): 3–18. DOI.