Pionless EFT and Shallow Two-Body Systems

Pionless EFT describes particles with momenta well below the pion mass by keeping only nonrelativistic nucleons and external probes. When an S-wave scattering length is much larger than the interaction range, the leading contact interaction must be iterated: its bubble sum gives the unitary amplitude $T(k)=4\pi/[M(-1/a-ik)]$. Derivative operators then correct this result in powers of $Q/\Lambda_b$, provided their coefficients are renormalized in the same regulator and no pole outside the EFT domain is mistaken for a prediction.

Required background. Nuclear and Few-Body EFT Architecture supplies the separation of low scales from the breakdown scale and the logic of promoting an interaction for iteration.

Helpful background. Partial-Wave Unitarity supplies the elastic unitarity and sheet conventions used to classify poles.

Large scattering length and the leading amplitude

Take a single two-nucleon S-wave channel $\alpha$, with nucleon mass $M$, center-of-mass momentum $k$, and $Q\sim k\sim 1/|a|\ll\Lambda_b$. The breakdown scale is the smallest omitted inverse range, particle-production scale, or resolved-exchange mass; in nuclear applications it is commonly bounded by $m_\pi$, but the physical channel can introduce a lower scale. A minimal Lagrangian is

$$\begin{aligned} \mathcal L={}&N^\dagger\left(i\partial_0+\frac{\boldsymbol\nabla^2}{2M}\right)N -\frac{C_{0,\alpha}}{2}(N^T P_\alpha N)^\dagger(N^T P_\alpha N)\\ &+\frac{C_{2,\alpha}}{16} \left[(N^T P_\alpha N)^\dagger (N^T P_\alpha\overleftrightarrow{\boldsymbol\nabla}^{,2}N) +\text{h.c.}\right]+\cdots . \end{aligned}$$

The spin–isospin projector $P_\alpha$ selects, for example, the ${}^1S_0$ or ${}^3S_1$ channel. Its normalization is absorbed into the displayed couplings. Pions, $Δ$ resonances, and other short-distance modes have been integrated out; their effects reside in $C_{2n,\alpha}$ and in operators involving external currents.

For a natural channel, $a\sim\Lambda_b^{-1}$ and every contact insertion can be perturbative. Here $1/a\sim Q$, so a two-nucleon loop scales as $M Q/(4\pi)$ and renormalization requires

$$C_0\sim\frac{4\pi}{M Q}.$$

Thus $C_0$ contributes at every order and its geometric series must be summed. With the amplitude convention

$$T(k)=\frac{4\pi}{M}\frac{1}{k\cot\delta(k)-ik},$$

matching $k\cot\delta=-1/a$ at leading order gives

$$T_{\mathrm{LO}}(k)=\frac{4\pi}{M}\frac{1}{-1/a-ik}.$$

This is not an expansion in $ka$: it retains $ka$ exactly while omitting range corrections. The effective-range expansion

$$k\cot\delta(k)=-\frac1a+\frac{r_e}{2}k^2+v_2k^4+\cdots$$

shows the remaining hierarchy. If $r_e\sim\Lambda_b^{-1}$, one insertion of $C_2$ produces a relative correction $r_e k^2/Q\sim Q/\Lambda_b$; equivalently $C_2\sim4\pi/(M Q^2\Lambda_b)$. Iterating $C_2$ without a separate counting argument can create regulator-sensitive deep poles, so the default pionless expansion treats natural range terms perturbatively. The bubble resummation and this promoted counting are developed systematically in Hammer, König, and van Kolck 2020, § II.B, pp. 7–15.

Renormalization with a sharp cutoff

A regulator is part of the calculation, not part of the observable. For a momentum-independent interaction with sharp cutoff $\Lambda$, define the principal-value loop for $0\le k<\Lambda$ by

$$I_{\mathrm{PV}}(k,\Lambda) =\frac{M}{2\pi^2} \left[-\Lambda+\frac{k}{2}\ln\!\left(\frac{\Lambda+k}{\Lambda-k}\right)\right].$$

The bubble sum can be written in terms of the real K matrix,

$$K^{-1}=C_0^{-1}(\Lambda)-I_{\mathrm{PV}}, \qquad T^{-1}=K^{-1}-i\frac{Mk}{4\pi}.$$

Matching the scattering length fixes the running coupling,

$$C_0^{-1}(\Lambda) =-\frac{M}{4\pi a}-\frac{M\Lambda}{2\pi^2}.$$

The linear divergence therefore cancels exactly. At finite cutoff,

$$T^{-1}(k,\Lambda) =-\frac{M}{4\pi a}-i\frac{Mk}{4\pi} -\frac{Mk}{4\pi^2}\ln\!\left(\frac{\Lambda+k}{\Lambda-k}\right),$$

and the last term is $-Mk^2/(2\pi^2\Lambda)+O(k^4/\Lambda^3)$. This residual is a regulator artifact of the omitted derivative operators, not a probability distribution for their size. Other schemes—dimensional regularization with power-divergence subtraction, for example—assign different finite pieces to $C_0$ but reproduce the same matched on-shell amplitude order by order Kaplan, Savage, and Wise 1998, pp. 390–396.

A reproducible calculation makes this cancellation numerical. For the dimensionless fixture $M=k=1$, $a=10$, the residual magnitudes at $\Lambda=10$ and $20$ are

$$\frac{\ln(11/9)}{4\pi^2}, \qquad \frac{\ln(21/19)}{4\pi^2},$$

whose ratio is $2.00503358\ldots$. Doubling the cutoff nearly halves the artifact, consistently with $Mk^2/(2\pi^2\Lambda)$; it does not halve the physical amplitude.

Unitarity, poles, and their domain

The inverse amplitude obeys exact elastic unitarity,

$$\operatorname{Im}T^{-1}(k+i0)=-\frac{Mk}{4\pi},$$

independently of $a$ and of the ultraviolet regulator. Equivalently, $S=(k\cot\delta+ik)/(k\cot\delta-ik)$ has $|S|=1$ for real $k$ below the first inelastic threshold. This is a stronger check than fitting a phase shift: an incorrect sign in the $ik$ term violates probability conservation immediately.

Analytically continue $k$ through its cut while keeping $E=k^2/M$. At leading order:

• If $a>0$, the zero at $k=i/a$ lies on the physical sheet and represents a shallow bound state with binding energy $B=1/(Ma^2)$.
• If $a<0$, the zero at $k=-i/|a|$ lies on the adjacent sheet and represents a virtual state, not a normalizable bound state.
• An off-axis resonance pole belongs on an unphysical sheet. A momentum-independent S-wave contact has only the imaginary-axis pole above; producing a shallow resonance generally requires additional tuning, range terms, a barrier, or an explicit degree of freedom.

Including $r_e$ makes the pole equation quadratic. One solution can track the shallow state, while the other is typically at $|k|\sim2/|r_e|$ and is outside pionless EFT when $r_e$ is natural. Treating that second root as a prediction extends a truncated effective-range expansion beyond its radius of validity. The pole and effective-range conventions trace to Bethe 1949, pp. 38–50.

A calculation contract

A reproducible two-body prediction should state the following in this order:

1. Degrees of freedom and channel. Name the particles, spin–isospin projector, Coulomb treatment if present, and breakdown scale.
2. Counting. Declare which low scales are $O(Q)$, which contact is promoted, and whether effective-range terms are inserted perturbatively or iterated.
3. Regulator and renormalization. Give the cutoff or subtraction prescription and the data used to determine each coupling. Vary the regulator only within a window that remains below omitted hard physics and above resolved momenta.
4. Observable. Solve for the on-shell amplitude, pole, or matrix element; a fitted potential or bare $C_0(\Lambda)$ is not itself observable.
5. Checks. Verify dimensions $[T]=M^{-2}$ in natural units, the unitarity identity, cutoff scaling after refitting, and stability under the next allowed operator.

Two-body scattering fixes strong contact interactions but does not, in general, determine short-range two-body couplings to external probes. Those enter Electroweak Currents in Few-Body Systems. Adding a third particle also changes the renormalization problem: Three-Body Renormalization and Universality shows why two-body data alone can then be insufficient.

Common pitfalls

Expanding the shallow denominator. A series in $C_0 I$ assumes $C_0 M Q/(4\pi)\ll1$, but promoted counting makes this combination order unity. Sum $C_0$ first and expand only range and shape corrections around the resulting pole.

Calling cutoff variation an EFT error bar. Residual cutoff dependence diagnoses missing counterterms and implementation errors. A probabilistic truncation statement additionally needs an explicit coefficient model, correlation assumptions, and validation.

Reading every fitted pole as physical. Poles at the cutoff or inverse effective range are not controlled when those momenta approach $\Lambda_b$. Quote the sheet and verify $|k_{\rm pole}|\ll\Lambda_b$ before assigning a state.

References

• Bethe, Hans A. “Theory of the Effective Range in Nuclear Scattering.” Physical Review 76 (1949): 38–50. DOI.
• Hammer, Hans-Werner, Sebastian König, and U. van Kolck. “Nuclear Effective Field Theory: Status and Perspectives.” Reviews of Modern Physics 92 (2020): 025004. DOI.
• Kaplan, David B., Martin J. Savage, and Mark B. Wise. “A New Expansion for Nucleon–Nucleon Interactions.” Physics Letters B 424 (1998): 390–396. DOI.