---
title: Tunneling Potential Method Overview
url: https://www.emergentmind.com/topics/tunneling-potential-method
type: topic
---

# Tunneling Potential Method Overview

The tunneling potential method denotes a family of tunneling formalisms in which a decay probability, decay exponent, or transmitted amplitude is computed through an auxiliary potential-like object rather than exclusively through a direct coordinate-space boundary-value problem. In one-dimensional quantum mechanics, the phrase can refer to direct integration of the time-independent Schrödinger equation from the transmission region backward through an arbitrary barrier, followed by matching to incident and reflected waves [1101.2620]. In relativistic vacuum decay, it refers more specifically to the introduction of a field-space function \(V_t\) that connects the metastable and stable phases and turns the bounce calculation into a variational problem for a one-dimensional action functional in field space [1805.03680]. Closely related constructions also appear in Q-ball theory, gauge-invariant false-vacuum decay, and several applied tunneling problems [2307.05667].

## 1. Terminological scope and defining constructions

A common misconception is that the tunneling potential method is a single universal algorithm. The literature instead uses the expression for several technically distinct constructions. In one usage, the relevant object is the transmitted-side boundary condition imposed on the wavefunction in a one-dimensional barrier problem. In another, the central object is an auxiliary function \(V_t(\phi)\) or \(\mathcal E(\varphi)\) defined in field space and optimized by an action principle. A plausible implication is that the phrase is best understood as a class of reformulations rather than a single formalism.

| Setting | Auxiliary object | Representative source |
|---|---|---|
| One-dimensional barrier scattering | Backward-integrated \(\Psi(x)\) with transmitted-side data | [1101.2620] |
| Vacuum decay of a scalar field | \(V_t(\phi)\) | [1805.03680] |
| Q-balls | \(V_t(\varphi)\) or \(\mathcal E(\varphi)\) | [2307.05667] |

For four-dimensional vacuum decay, the defining relation is
\[
V_t(\phi)\equiv V(\phi)-\tfrac12\,\bigl(d\phi_b/dr\bigr)^2,
\]
with \(\phi_b(r)\) the \(O(4)\)-symmetric bounce. The associated endpoint data are
\[
V_t(\phi_+)=V(\phi_+),\qquad V_t'(\phi_+)=0,\qquad
V_t(\phi_0)=V(\phi_0),\qquad V_t'(\phi_0)=\tfrac34\,V'(\phi_0),
\]
where \(\phi_+\) is the false vacuum and \(\phi_0\) the bounce exit point [1805.03680]. For Q-balls, the analogous construction starts from a stationary complex scalar \(\Phi(\vec x,t)=\varphi(r)e^{i\omega t}/\sqrt2\) and defines
\[
V_t(\varphi)\equiv V(\varphi)-\tfrac12\,\omega^2\varphi^2,
\]
so that the radial equation becomes structurally identical to a three-dimensional bounce equation [2307.05667].

## 2. One-dimensional quantum-mechanical barrier formulations

In the direct-integration or shooting formulation of one-dimensional tunneling, the system is partitioned into Region I \((x<0)\), Region II \((0\le x\le L)\), and Region III \((x>L)\). Region I contains incident plus reflected waves, Region II contains an arbitrary barrier \(V(x)\), and Region III has constant \(V_{\rm III}\) with only a transmitted right-travelling wave. The method begins from the time-independent Schrödinger equation,
\[
-\frac{\hbar^2}{2m}\frac{d^2\Psi}{dx^2}+V(x)\Psi=E\Psi,
\]
or equivalently
\[
\frac{d^2\Psi}{dx^2}=\frac{2m}{\hbar^2}\bigl[V(x)-E\bigr]\Psi.
\]
Rather than fixing the incident amplitude at the left, one sets \(F=1\) in Region III and imposes at \(x=L\)
\[
\Psi(L)=e^{ik_{\rm III}L},\qquad
\Psi'(L)=ik_{\rm III}e^{ik_{\rm III}L},
\]
then integrates backward to \(x=0\). Matching there to
\[
\Psi_{\rm I}(x)=A\,e^{ik_{\rm I}x}+B\,e^{-ik_{\rm I}x}
\]
yields
\[
A=\tfrac12\Bigl[\Psi(0)+\frac{i}{k_{\rm I}}\Psi'(0)\Bigr],
\]
and hence
\[
T(E)=\frac{k_{\rm III}}{k_{\rm I}}\,
\frac{4}{\left|\Psi(0)-\frac{i}{k_{\rm I}}\Psi'(0)\right|^2}.
\]
This construction automatically enforces a purely outgoing condition in Region III, typically needs fewer steps than a transfer-matrix discretization of comparable accuracy, yields the full complex \(\Psi(x)\), and extends easily to arbitrary, even discontinuous, \(V(x)\) [1101.2620].

The numerical implementation described in the same source rewrites the complex second-order ODE as four coupled first-order equations for \(u_1=\Re\Psi\), \(u_2=\Im\Psi\), \(u_3=\Re\Psi'\), and \(u_4=\Im\Psi'\), and advances them with a standard fourth-order Runge–Kutta scheme. The step size \(\Delta x\) must resolve the shortest local wavelength in oscillatory regions or the exponential decay length in forbidden regions; a rule of thumb is 20–50 points per local half-wavelength. Because the equation has no first-derivative term it is not stiff in the classical sense, although strong forbidden regions can generate overflow or underflow unless \(\Delta x\) is sufficiently small or periodic renormalization is used [1101.2620].

For continuous barriers, a distinct but nearby methodology uses a uniform asymptotic expansion based on Airy functions. There the forbidden-region solution is written in terms of \(\Ai\) and \(\Bi\), remains uniformly valid up to the turning points, and gives
\[
T\approx \frac14\,e^{-2S}\Bigl[1+\frac{5}{48S}+\cdots\Bigr],
\qquad
S=\int_{x_1}^{x_2}\kappa(x)\,dx.
\]
This approach captures the \(1/4\) prefactor absent in naive WKB and the first beyond-WKB correction, while retaining the assumptions of a smooth barrier, linear turning points, and \(S\gg1\) [1104.1748].

## 3. Field-space variational reformulation of vacuum decay

In the vacuum-decay formalism introduced for metastable scalar potentials, the bounce action is rewritten as a field-space functional of \(V_t\). In flat space,
\[
S[V_t]=54\pi^2\int_{\phi_+}^{\phi_0}
\frac{\bigl[V(\phi)-V_t(\phi)\bigr]^2}{\bigl[-V_t'(\phi)\bigr]^3}\,d\phi,
\]
and extremizing \(S[V_t]\) reproduces the ordinary Euclidean bounce action. The corresponding Euler–Lagrange equation is
\[
(4V_t'-3V')\,V_t'+6\,(V-V_t)\,V_t''=0.
\]
The Euclidean radius can then be reconstructed from
\[
r(\phi)=3\,\sqrt{\frac{2\,[V(\phi)-V_t(\phi)]}{[V_t'(\phi)]^2}},
\]
so the usual profile \(\phi_b(r)\) is not discarded but encoded indirectly in \(V_t\) [1805.03680].

The same formalism admits an alternative derivation via canonical transformations. In that derivation one trades the canonical pair \((r,p)\) for \((V_t,P)\), finds a generating function \(G(r,V_t)=-\tfrac12\pi^2V_t r^4\), and then Legendre-transforms to recover exactly the field-space action above. The same canonical logic extends to multi-field systems and to gravity, where the Hamiltonian constraint plays the role of the Euclidean Friedmann equation [2209.03293].

With gravity included, the functional becomes
\[
S[V_t]=\frac{6\pi^2}{\kappa^2}\int_{\phi_+}^{\phi_0}
\frac{\bigl[D(\phi)+V_t'(\phi)\bigr]^2}{D(\phi)\,V_t(\phi)^2}\,d\phi,
\]
with
\[
D(\phi)=\sqrt{V_t'(\phi)^2+6\kappa\,[V(\phi)-V_t(\phi)]\,V_t(\phi)}.
\]
The gravitational Euler–Lagrange equation is
\[
(4V_t'-3V')\,V_t'+6\,(V-V_t)\,[V_t''+\kappa(3V-2V_t)]=0.
\]
In the limit \(\kappa\to0\), the gravitational functional reduces to the flat-space one [2404.19657].

A central structural property of the method is that the physical \(V_t\) is a minimum of \(S[V_t]\), not a saddle point. The second variation is positive definite, or equivalently the Hessian operator governing \(V_t\) fluctuations has a strictly positive spectrum. This is the feature on which the frequent description of the method as a downhill minimization rests [2404.19657].

## 4. Boundary conditions, limiting cases, and gauge-invariant implementation

The boundary data depend on the asymptotics of the false vacuum. For Minkowski or AdS false vacua,
\[
V_t(\phi_+)=V(\phi_+),\qquad V_t'(\phi_+)=0,
\]
while at the endpoint on the true side of the barrier,
\[
V_t(\phi_0)=V(\phi_0),\qquad V_t'(\phi_0)=\tfrac34\,V'(\phi_0).
\]
For a de Sitter false vacuum, there is always an initial segment \(\phi_+\to\phi_{0+}\) on which \(V_t(\phi)=V(\phi)\), and the actual bounce range is \(\phi_{0+}\to\phi_0\) with the same endpoint condition \(V_t'=\tfrac34 V'\) at both ends of that range. When the range collapses to the barrier top \(\phi_T\), the minimizing configuration is simply \(V_t=V\) from \(\phi_+\) to \(\phi_T\), and the action reduces to the Hawking–Moss result
\[
S_{HM}=\frac{24\pi^2}{\kappa^2}\left(\frac1{V(\phi_+)}-\frac1{V(\phi_T)}\right).
\]
The same framework also accommodates pseudo-bounces and bubbles of nothing by relaxing the endpoint condition at \(\phi_0\); in these cases the Euclidean approach often requires boundary terms, while the \(V_t\)-action gives the correct \(\Delta S_E\) without adding any extra boundary term [2404.19657].

Near-degenerate vacua recover the thin-wall limit. In the original field-space reformulation, the compact action density reduces to Coleman’s thin-wall expression,
\[
S_{E,\rm tw}=\frac{27\pi^2}{2}\,\frac{\sigma^4}{(\Delta V)^3},
\]
so the field-space formalism functions as a generalization of the thin-wall action to arbitrary potentials [1805.03680]. A concrete analytic realization appears in a piecewise linear/quadratic landscape potential, where the bubble radius \(R_T\) obeys
\[
2\alpha+\sqrt{\Delta}\,mR_T
=
mR_T\,\frac{I_2(mR_T)}{I_1(mR_T)},
\]
and the tunneling exponent is
\[
B
=
2\pi^2\phi_-^2R_T^2
\left[
\alpha^2
+\tfrac12\Bigl(\tfrac43\alpha\sqrt\Delta+\frac{I_2(mR_T)}{I_1(mR_T)}\Bigr)mR_T
-\tfrac{(1-\Delta)}{8}m^2R_T^2
\right].
\]
In the large-radius regime \(mR_T\gg1\), the result reproduces the Coleman–de Luccia thin-wall expression; in the small-radius regime \(mR_T\ll1\), it reproduces the Duncan–Jensen linear-potential limit. Across all \(\alpha,\Delta\in(0,1)\), one of the two closed-form approximations is within 10% of the exact \(mR_T\) and within 50% in \(B\) [1102.4742].

Gauge invariance requires additional care. In the Abelian Higgs model, a naive implementation of the tunneling-potential approach inherits the gauge dependence of the bounce formalism. A gauge-invariant procedure uses the counting \(\lambda\sim e^4\), \(m^2\sim e^4v^2\), expands \(V_{\rm eff}=V_{e^4}+V_{e^6}+\cdots\) and \(Z(\phi)=1+Z_{e^2}+\cdots\), computes the leading action from \(V_{e^4}\), and adds the subleading corrections from \(Z_{e^2}\) and \(V_{e^6}\) evaluated on the leading \(V_t\). With a quartic ansatz for \(V_t\), the method is typically 10–100 times faster than shooting the bounce ODE, and the residual gauge dependence is attributed solely to the polynomial approximation; with the refinements described in the source, it is typically at the \(10^{-4}\)–\(10^{-5}\) level for \(\xi\in[0,100]\) [2105.07588].

## 5. Exact solutions, multi-field embeddings, and Q-balls

The tunneling-potential formalism is unusually effective at generating exact solutions. One result is that any single-field pair \(\{V(\varphi),V_t(\varphi)\}\) can be embedded in an \(N\)-field potential by choosing a path \(\boldsymbol\phi(\varphi)\) in field space and extending the potential transversely so as to satisfy the multi-field equations. Along the path, the ODE for \(V_t\) is identical to the single-field equation, while the transverse force is fixed by
\[
2\,[V(\varphi)-V_t(\varphi)]\,\boldsymbol\phi''(\varphi)=\nabla_TV.
\]
This construction yields explicit analytic two-field and three-field examples, including a quarter-circle trajectory, a catenary, and a helix, and the bounce action matches the single-field value. In one thin-wall example, the exact action is \(S_{\rm exact}=80.59\) and FindBounce returns \(80.70\) with default settings, making these solutions useful as benchmarks for multi-field numerical codes [2312.12360].

The same paper gives a gravitating extension with
\[
S[V_t]
=
\frac{6\pi^2}{\kappa^2}
\int_{\varphi_+}^{\varphi_0}
\frac{(D+V_t')^2}{D\,V_t^2}\,d\varphi,
\qquad
D=\sqrt{V_t'^2+6\kappa(V-V_t)V_t},
\]
and an analytic example based on
\[
V_t(\varphi)=V_+-\sin^2\varphi,\qquad V_+\le0.
\]
This permits closed-form expressions for \(E(\varphi)\), \(F(\varphi)\), and hence \(V(\varphi)\), with the Minkowski limit reducing to an elementary combination of \(\sin\), \(\cos\), and \(\mathrm{arctanh}\) [2312.12360].

For Q-balls, the structural identity with three-dimensional bounce solutions allows a direct translation of the field-space strategy. Besides the definition
\[
V_t(\varphi)=V(\varphi)-\tfrac12\,\omega^2\varphi^2,
\]
the problem can be reformulated in terms of a monotonic “mechanical energy” function
\[
\mathcal E(\varphi)\equiv \tfrac12\bigl(d\varphi/dr\bigr)^2+V(\varphi),
\]
which satisfies
\[
(3\mathcal E'-2V')\,\mathcal E' = 4(\mathcal E-V)\,\mathcal E''.
\]
The corresponding boundary conditions are
\[
\mathcal E(\varphi_+)=V(\varphi_+),\quad
\mathcal E(\varphi_0)=V(\varphi_0),\quad
\mathcal E'(\varphi_+)=0,\quad
\mathcal E'(\varphi_0)=\tfrac23\,V'(\varphi_0).
\]
Low-order polynomial ansätze \(\mathcal E_1,\mathcal E_2,\mathcal E_3,\mathcal E_4\) then approximate small-\(Q\) and intermediate-\(Q\) Q-balls with percent-level accuracy, while exactly solvable examples follow by choosing \(\mathcal E(\varphi)\) directly and reconstructing \(V(\varphi)\) from it [2307.05667].

## 6. Applied and adjacent usages

Outside false-vacuum decay, the term also appears in more application-specific settings. In nitride-based resonance-tunneling structures, an analytical method combines the effective-mass Schrödinger equation with the dielectric-continuum Poisson equation and iterates a self-consistent potential profile
\[
V(z)=\Delta E_C(z)+V_E(z)+V_H(z)+V_{HL}(z).
\]
At the zeroth iteration each layer has a linear slope \(F_p\) and the Schrödinger equation is solved with Airy functions; after Poisson integration and inclusion of the Hedin–Lundquist exchange-correlation term, the potential is linearized on a fine mesh and the cycle is repeated until
\[
\delta=\max_z\bigl|n_{\nu+1}(z)-n_\nu(z)\bigr|/n_\nu(z)<10^{-6}.
\]
For the experimentally realized three-well QCD cascade cited in the source, the iteration converges in approximately 25 steps, predicts \(\Omega_{13}^I=782.5\) meV, and agrees with the experimental \(\Omega_{13}^{\rm exp}=800.0\) meV to within 2.5% [1812.08544].

For explicitly time-dependent barriers, a different construction introduces a localized negative imaginary absorber \(V_{\rm im}(x,t)\) into the Schrödinger equation,
\[
i\hbar\partial_t\psi=
\Bigl[-\frac{\hbar^2}{2m}\partial_x^2+V_{\rm real}(x,t)\Bigr]\psi
+i\,V_{\rm im}(x,t)\psi,
\]
with \(V_{\rm im}\) built from Gamow-like outgoing states. In the current-biased Josephson-junction application, this leads to a TDIP of the form
\[
V_{\rm im}(\phi,t)=
\begin{cases}
\displaystyle
\frac{\beta}{\sqrt{2(\phi-\phi_{\rm turn}(t))}\,\Phi_0\,\omega_p(t)},
& \phi>\phi_{\rm turn}(t),\\[1ex]
0,&\phi<\phi_{\rm turn}(t),
\end{cases}
\]
and a survival probability \(S(t)=\|\psi(t)\|^2\) with instantaneous decay rate
\[
\Gamma(t)=-\frac1{S(t)}\frac{dS(t)}{dt}.
\]
The resulting switching-current distributions reproduce the zero-field switching peak, microwave-induced resonance side peaks, multi-peak structures under chirp, dynamical bifurcation, and real-time oscillatory detection of a weak field [1303.5644].

In field emission from curved nanotips, the tunneling potential is the local normal-direction barrier including curvature corrections. For sharply curved emitters the corrected form is
\[
V(\Delta s)=
-\,e\,\gamma E_0\,\Delta s
\left[
1-\frac{\Delta s}{R_2}
+\frac43\left(\frac{\Delta s}{R_2}\right)^2+\cdots
\right],
\]
with the same coefficients \(a_1=-1\) and \(a_2=4/3\) for hemiellipsoid, hyperboloid, and sphere. Numerical validation on a rounded conical tip shows the corrected expansion reproduces the exact barrier to better than 1–2% out to \(\Delta s\sim1\) nm, and the empirical regime of validity quoted in the source is \(R_a\gtrapprox2\) nm with local field strengths of \(5\) V/nm or higher [1710.03992].

A further adjacent usage appears in singular-interaction models, where tunneling is encoded not by a spatial barrier profile but by exponentially small off-diagonal matrix elements of a finite-dimensional principal matrix \(\Phi(E)\) in Krein’s formula. In the one-dimensional double-\(\delta\) problem,
\[
\Phi_{12}(E_B)= -\frac{\lambda}{4}\,e^{-\lambda a},
\]
and degenerate perturbation theory gives
\[
\Delta E=2|\Phi_{12}(E_B)|=\frac{\lambda}{2}e^{-\lambda a}\sim \lambda^2 e^{-\lambda a}.
\]
This is not the same formalism as the field-space \(V_t\) construction, but it shares the same basic strategy of trading a tunneling boundary-value problem for a more compact auxiliary object [1904.12031].

Across these usages, the common thread is methodological rather than notational: the tunneling problem is reformulated so that the decisive quantity is obtained from a potential-like auxiliary structure with simpler boundary conditions, a lower-dimensional variational principle, or a more stable numerical scheme. In vacuum decay, that reformulation is particularly sharp, since the tunneling potential \(V_t\) replaces the Euclidean bounce by a true minimization problem in field space [2404.19657].

Source: https://www.emergentmind.com/topics/tunneling-potential-method