Bolker Condition in Inverse Problems

• Bolker condition is a geometric–microlocal criterion that defines when a canonical relation enables stable, artifact-free inversion in generalized Radon transforms.
• It ensures injectivity by requiring the left projection to be an injective immersion, thereby mapping singularities uniquely between domains.
• This condition is pivotal in inverse problems and tomography, as it prevents artifacts like folds, cusps, and mirror points in reconstructed images.

The Bolker condition is a pivotal geometric–microlocal criterion arising in the analysis of generalized Radon transforms, Fourier Integral Operators (FIOs), and integral geometry. It delineates when the canonical relation associated with an FIO enables artifact-free, stable inversion, and injectivity—properties critical in tomography, inverse problems, and the propagation of singularities. Although originally formulated in the context of double-fibration transforms and Radon transforms, it now appears in a wide range of geometric analysis, microlocal analysis, and mathematical imaging frameworks (Webber et al., 2022, Mazzucchelli et al., 2023, Homan et al., 2015, Webber et al., 2022, Webber et al., 2023, Webber et al., 2020, Webber, 2023, Busch, 13 May 2025, Busch et al., 27 Oct 2025). Distinct from conditions concerning the analytic or smooth category, the Bolker condition bridges geometry, linear analysis, and the recoverability of singularities in data.

1. Formal Definitions and Canonical Relations

The Bolker condition is defined within the setting of FIOs associated to a canonical relation

$C \subset T^*(Y) \times T^*(X)$

where $X$ is the object manifold and $Y$ is the data or parameter manifold. The FIO is typically represented as an oscillatory integral with a nondegenerate phase function $\Phi$, leading to a canonical relation

$$C = \{ (y, \eta; x, \xi) : \exists \theta,\, d_\theta\Phi = 0,\, \eta = d_y\Phi,\, \xi = -d_x\Phi \}.$$

Two fundamental projections are associated to $C$:

• Left projection: $\Pi_L: C \rightarrow T^*Y$, mapping $(y, \eta; x, \xi) \mapsto (y, \eta)$
• Right projection: $\Pi_R: C \rightarrow T^*X$, mapping $(y, \eta; x, \xi) \mapsto (x, \xi)$

The Bolker condition is satisfied if:

• $X$0 is an injective immersion (i.e., an embedding): each $X$1 arises from at most one $X$2, and the differential $X$3 is everywhere of full rank.
• Equivalently, $X$4 is a local canonical graph over $X$5; the mapping from singularities of $X$6 to those in $X$7 is locally diffeomorphic and non-folding [(Webber et al., 2022), Definition 2.11; (Webber et al., 2022); (Webber et al., 2023); (Homan et al., 2015); (Mazzucchelli et al., 2023)].

For transforms arising from a double-fibration $X$8, with $X$9 and canonical relation $Y$0, the Bolker condition is equivalent to requiring that the projection of $Y$1 to $Y$2 is an injective immersion and that its projection to $Y$3 is an immersion transverse to the zero section (Busch et al., 27 Oct 2025).

2. Equivalent Geometric Characterizations

The Bolker condition admits several geometric formulations:

• Transversality: The conormal bundle $Y$4 projects cleanly to $Y$5; projections do not have fold or cusp singularities; there are no caustics (Homan et al., 2015, Mazzucchelli et al., 2023).
• Injectivity along fibers: For each parameter $Y$6, there is at most one $Y$7 in $Y$8. No conjugate (multiple) points map to the same data covector; this is particularly relevant in integral geometry and X-ray transforms (Mazzucchelli et al., 2023).
• No tangency to support: In geometric Radon transforms (e.g., with ellipsoidal or lemon surfaces), Bolker is equivalent to the support of $Y$9 not intersecting any tangent plane to the family of center surfaces $\Phi$0 [(Webber et al., 2022), Theorem 2.1; (Webber et al., 2023)].
• Variation/Jacobi-field condition: For ray transforms, the variation field along rays must be non-vanishing in directions conjugate to measurement covectors (Mazzucchelli et al., 2023).
• Moment map surjectivity: In $\Phi$1-plane or higher-codimension transforms, a specific linearization (combining differentials of the defining functions in parameters and object directions) must be surjective (Mazzucchelli et al., 2023, Homan et al., 2015).

3. Key Examples and Verification

A variety of tomographic and geometric transforms either satisfy or fail the Bolker condition, with direct implications for inversion:

Transform Type	Bolker Condition	Implications / Notes
Classical X-ray, Euclidean	Satisfied	Artifact-free inversion, uniqueness
Geodesic X-ray (simple manifold)	Satisfied	No conjugate points, no fold artifacts
Ellipsoidal/Hyperboloidal Radon (Webber et al., 2022)	Satisfied if support avoids tangent planes	No mirror-point artifacts, recoverability of singularities
Compton / Bragg Scattering (Webber et al., 2020, Webber et al., 2022)	Satisfied for standard lemon/Bragg geometry	Stable edge recovery, only boundary artifacts
Sinusoidal integration (CST) (Webber et al., 2020)	Fails (non-injective $\Phi$2)	Predictable fold artifacts along degenerate loci
Restricted apple/lemons (partial data) (Webber et al., 2022)	Fails for apple with fixed axis; holds for lemon with support restriction	Ghost singularities if Bolker fails
Minimal surface transform (double fibration) (Busch et al., 27 Oct 2025)	Satisfied under foliation and analytic assumptions	Invertibility, analytic wavefront recovery

4. Role in Microlocal Analysis and Inversion

The Bolker condition is essential for:

• Microlocal invertibility: If Bolker holds and the FIO is elliptic, the normal operator $\Phi$3 is an elliptic pseudodifferential operator. Thus, all visible singularities of $\Phi$4 are mapped one-to-one to data singularities, enabling inversion and visible singularity recovery (Webber et al., 2022, Busch, 13 May 2025, Homan et al., 2015, Webber et al., 2023).
• Artifact suppression: When Bolker fails, artifacts—such as mirror-point, fold, or cusp singularities—arise in inversion due to multiple preimages or tangential collapse (Webber et al., 2020, Webber et al., 2023).
• Exact support theorems: The Bolker property is both necessary and sufficient for support and uniqueness theorems in analytic and smooth categories (as in microlocal Holmgren, Helgason support theorems) (Mazzucchelli et al., 2023, Homan et al., 2015).
• Fredholm property: Under Bolker and ellipticity, the forward FIO is Fredholm between appropriate Sobolev spaces and produces no exotic singularities not predicted by the canonical relation (Webber et al., 2022).

5. Analytical and Stability Results

The Bolker condition ensures robust analytic properties:

• Injectivity and stability: Under Bolker, generic (open and dense) classes of analytic and smooth Radon-type transforms are injective and satisfy two-sided Sobolev stability estimates. Small perturbations of the geometry or weight (in sufficiently high $\Phi$5 topology) preserve these properties (Homan et al., 2015).
• Wavefront mapping: Analytic FIOs with Bolker canonical relations admit direct analytic wavefront set recovery: absence of data singularities at a covector implies absence in the preimage (Busch, 13 May 2025, Mazzucchelli et al., 2023, Homan et al., 2015).
• Explicit inversion: In various settings (e.g., surfaces of revolution, minimal surface transforms), the Volterra equation framework, together with the absence of artifacts under Bolker, yields constructive inversion strategies (Webber et al., 2023, Busch et al., 27 Oct 2025).

6. Implications in Specific Tomographic and Geometric Inverse Problems

• Ultrasound Reflection Tomography (URT): The Bolker condition is satisfied for spheroidal measurement geometries where the support of $\Phi$6 does not intersect tangent planes to the cylindrical surface $\Phi$7; stable and unique recovery is thereby enabled (Webber et al., 2022).
• Compton Scattering Tomography (CST): For lemon-type integration surfaces, Bolker is satisfied except on boundary support, leading to predictable (boundary-only) artifacts; for more degenerate or partial data, failure of Bolker directly predicts streak and cusp artifacts observed in numerical reconstructions (Webber et al., 2020, Webber, 2023, Webber et al., 2022, Webber et al., 2023).
• Generalized Boundary Rigidity/Minimal Surface Problems: The minimal surface transform fits within the double fibration paradigm; under Bolker (Guillemin) and analytic assumptions, analytic wavefront and metric recovery results are established (Busch et al., 27 Oct 2025).
• Genericity and stability: The local and global forms of the Bolker condition are generically satisfied in open, dense sets of geometric data for a large class of transforms (Homan et al., 2015). Counterexamples exist in low regularity or for special geometric degeneracies.

7. Related Frameworks and Broader Connections

• Double fibration transforms: The Bolker condition provides the precise geometry under which double fibration transforms and associated FIOs become elliptic and invertible; it underpins the theoretical development of support theorems, microlocal stability, and analytic inversion in integral geometry (Mazzucchelli et al., 2023, Busch et al., 27 Oct 2025).
• Analytic microlocal analysis: In the analytic category, Bolker ensures that parametrix constructions via FBI transforms and analytic stationary phase are valid, facilitating analytic wavefront propagation and recovery (Busch, 13 May 2025).
• Stochastic population models (Bolker-Pacala): In a different context, the "Bolker condition" captures a functional inequality ensuring that death and competition dominate birth, leading to sub-Poissonian statistics and self-regulation (Kondratiev et al., 2017). This condition prevents clustering and guarantees global well-posedness.

---

In summary, the Bolker condition is the definitive microlocal hypothesis ensuring that a generalized Radon or fibration-based transform acts as an elliptic FIO with graph-type canonical relation, enabling the stable, unique, and artifact-free recovery of singularities from data. Its failure is directly associated with the appearance of image artifacts, loss of injectivity, and instability in inverse and imaging problems across both analytic and smooth frameworks. It is thus indispensable in the rigorous analysis and implementation of geometric inverse problems (Webber et al., 2022, Webber et al., 2023, Webber et al., 2022, Homan et al., 2015, Mazzucchelli et al., 2023, Busch et al., 27 Oct 2025, Busch, 13 May 2025).