Observability of wave equations with time- and space-dependent potentials: a counterexample under GCC and a positive criterion on the disk

arXiv:2610.25327 · math.OC, math.AP · Submitted 2026-10-18 · Read on arXiv

Zhonghua Liao, Qi Lü, Xu Zhang

Sichuan University

math.OC, math.AP

Submitted: 2026-10-18

Updated: 2026-10-27

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

The gist: The provided text details steps in a proof related to Proposition 7.1, involving repair and extension techniques, estimates for residual terms, verification of static completion criteria…

Terminology

Summary

The provided text details steps in a proof related to Proposition 7. p. 11, involving repair and extension techniques, estimates for residual terms, verification of static completion criteria, and the construction of a smooth periodic pair. p. 3 p. 3

Relevant extracted parts are:

(B3) Second-derivative ratio ϱ = c∗ϵ4. Equations (7. p. 0 p. 0 p. 013) and (7.14) both hold at this endpoint.

(B4) Second-derivative ratio ϱ = Kϵ4. p. 6 Equations (7.14) and (7.15) both hold at this endpoint.

(B5) First window endpoint r = 1. p. 2 Equations (7.26) and (7.27) both cover r = 1.

(B6) Lower growth cutoff r = clowϵ−1/5. p. 1 Since c0 < clow and ϵ−1/5 < ϵ−2/5, the interval [c0ϵ−1/5, clowϵ−1/5] is contained in both (7. p. 027) and (7.28). We use the weaker conclusion. p. 5

(B7) Lower interface endpoint r = c0ϵ−1/5. p. 1 This point belongs to (7. p. 2 p. 2 p. 227) and to (7.28).

(B8) Upper interface endpoint r = K1ϵ−2/5. p. 2 This point belongs to (7.28) and to (7.29), since K1 > 1 implies ϵ−2/5 ≤ K1ϵ−2/5. p. 0

(B9) Upper growth cutoff r = ϵ−2/5. p. 1 This point belongs to (7.28) and to (7.29).

(B10) Switching between the two source labels. p. 2 At a point of U1 ∩ U2, the two sets are disjoint by Lemma 7. p. 92, so no switch occurs on the source. At points where Ni vanishes, the choice of label is immaterial, and the repair coefficients vanish on an open neighborhood. p. 2

(B11) Band-index gap mh = ml + 2. p. 4 This is the endpoint of the index comparison in (7. p. 440). It is included in the estimate W ≤ 2β20Aβ20m, which is stated for d ≥ 2 and for d ≤ 1 separately; the two bounds coincide up to a fixed factor at d = 2. p. 3

(B12) The branch Ei = KE3−i. p. 3 The two bounds for Gi and Si are both stated with closed inequalities. p. 9 At the endpoint Ei = KE3−i, either representation may be used; the differentiated factors have the same exponential bound.

The proof proceeds through Steps 1 to 4 of Proposition 7. p. 61, constructing a smooth periodic pair by matching local solutions to an exterior static solution using techniques involving residual estimates and comparison theorems. p. 5 The final part involves verifying the static completion criterion (Lemma 7.7) and applying Theorem 6. p. 13 to obtain a smooth periodic pair on a neighborhood of the observation polygons, which is then extended globally using Lemma 7. p. 08 to form a complete solution satisfying the Dirichlet boundary condition, ultimately leading to the failure of observability on the GCC set for the fixed potential V. p. 0

Summary:

The text outlines steps for proving Proposition 7.1 by constructing a smooth periodic pair through repair and extension techniques, involving estimates derived from various bounds (B3-B12), residual analysis (Step 2), and matching local solutions to an exterior static solution (Step 4). p. 5 The construction relies on verifying the static completion criterion using Lemma 7. p. 07 and Theorem 6. p. 1 p. 13, resulting in a smooth periodic pair on a neighborhood of the observation polygons, which is then extended globally via Lemma 7. p. 6 p. 68 to satisfy Dirichlet boundary conditions, demonstrating that for the fixed potential V, observability fails on the GCC set. p. 0

Relevant Quotes:

The exponential comparison holds for every mi ≥ n0 after increasing C(r, N) over finitely many indices.

The residual bound and the index comparison then give maxα≤r ∂αS ≤ C(r, N) min[y1, y2]N.

"Finally, write S = S1 + S2, Si ≜ (1 − χ∗)Wif3−i. On supp Wi ∩ [Ei ≥ KE3−i], put hi ≜ f3−i/fi. Then hi ≤ 8C0/5K < 1/2, Si = -Rif3−i f2i / (1 + hi), and, for every spacetime multiindex α, ∂αSi = -X α1+α2+α3=α α! p. 4 α1!α2!α3! (∂α1Ri)(∂α2 f3−i)∂α3 f−2i (1 + hi) −1. The reciprocal derivative bound and the residual estimate, together with flatness of f3−i in its own normal variable, give maxα≤r ∂αb(i) + ∂α(χ∗Wi) ≤ C(r, N)β−Nmi. p. 4"

"The scalar cutoff defining the global X is the identity at every local level greater than 2. Hence X > 2 on U and Proposition 7. p. 16 gives (ploc, Vloc) = (1, 0) there. p. 1"

The static formula near the physical boundary gives the Dirichlet condition and smoothness up to ∂omega.

For the fixed potential V, however, the observation of p is identically zero on every finite interval and its initial state is nonzero. No finite-time observability inequality can therefore hold. p. 1

Parameter ledger for the counterexample in Theorem 2.1.

(Table 1 details parameters like Z, n0, β = 21/16, γ = 24 log 2 — fixed local solution Section 4; domain width, starting index, frequency ratio, second edge type; S geometry d0, α0 32/S < d0 sin α0/104; makes aρg/y∗ large. R = 2Sρg, ζi = Sei · (v0 − c0i) S S, ρg, reference midpoints local data of Proposition 7. p. 11.)

(Mmerge = M0(α0, R, ζ1, ζ2) R, ζi Proposition 7.1 vertex-merging threshold; independent of M. p. 3 ε2, c∗, q∗, θQ, η, c(η), K local data Lemma 7. p. 93 small-envelope thresholds and exponents; independent of M. p. 5)

(εQ, ε1, ε′2, clow, c0, K1, Cmax local data and ε2 Lemma 7.4 comparable-envelope thresholds; independent of M. p. 9)

(δg S geometry 32/S < δg < ρg sin α0/100; sets δp = aδg once M is fixed.)

(Mact all above Proposition 7.6 activation threshold for the first-derivative and geometric estimates; independent of M. p. 2)

(M Mmerge, Mact, all above (7.55) M ≥ max[Mmerge, Mact], M ≥ n0 + C(S) + 10, a diam omega0 < min[1/100, Y∗], and the two inequalities 2l < min[8cα − 2/cα, 2/cα, 6/cα]y∗ and 2γMl < 8y∗ with l = y2∗ = β−2M.)

The summary is: The provided text details steps in a proof related to Proposition 7.1, involving repair and extension techniques, estimates for residual terms, verification of static completion criteria, and the construction of a smooth periodic pair. This construction relies on estimates derived from various bounds (B3-B12), residual analysis (Step 2), and matching local solutions to an exterior static solution (Step 4). p. 9 The final part involves verifying the static completion criterion using Lemma 7.7 and Theorem 6.3, resulting in a smooth periodic pair on a neighborhood of the observation polygons, which is then extended globally via Lemma 7.8 to form a complete solution satisfying Dirichlet boundary conditions, ultimately leading to the failure of observability on the GCC set for the fixed potential V. p. 0"

Quote:

The exponential comparison holds for every mi ≥ n0 after increasing C(r, N) over finitely many indices.

The residual bound and the index comparison then give maxα≤r ∂αS ≤ C(r, N) min[y1, y2]N.

"Finally, write S = S1 + S2, Si ≜ (1 − χ∗)Wif3−i. On supp Wi ∩ [Ei ≥ KE3−i], put hi ≜ f3-i/fi. Then hi ≤ 8C0/5K < 1/2, Si = -Rif3-i f2i / (1 + hi), and, for every spacetime multiindex α, ∂αSi = -X α1+α2+α3=α α! p. 4 α1!α2!α3! (∂α1Ri)(∂α2 f3−i)∂α3 f−2i (1 + hi) −1.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved system can rigorously distinguish between geometric control condition (GCC) satisfaction and actual internal observability for wave equations with time-dependent potentials on bounded domains with analytic boundaries. This allows AI to identify scenarios where GCC is a necessary but not sufficient condition, or conversely, where observability fails even when GCC holds, specifically when a smooth 2pi-periodic potential exists.

  2. Improved system can determine the optimal observation geometry for achieving exact controllability of the wave equation under arbitrary bounded measurable time-dependent potentials on domains with analytic boundaries. This means the AI can suggest specific observation regions, such as those containing a central disk and a neighborhood of an embedded cross-cut, that guarantee observability for all weak solutions up to a sufficiently large time T.

  3. Improved system can construct explicit counterexamples demonstrating the failure of observability based solely on geometric control condition (GCC) when time-dependent potentials are introduced. This capability allows the AI to generate complex physical setups where a smooth 2pi-periodic potential exists alongside an observation set satisfying GCC, yet no finite observation horizon yields an observability inequality for a nonzero solution.

  4. Improved system can perform local unique continuation analysis for wave equations with smooth real-valued scalar potentials across noncharacteristic hyperplanes, even when the underlying complex-valued construction fails to produce a real-valued pair. This is crucial for understanding the role of unique continuation in observability proofs and allows the AI to analyze local nonuniqueness phenomena using real-valued, globally periodic solutions.

  5. Improved system can verify if an observation geometry satisfies criteria that supply both propagation and continuation for arbitrary bounded time-dependent potentials on a disk, as described in Theorem 2.3. This enables the AI to confirm whether a proposed observation set, defined by an embedded cross-cut and neighborhood of its image, is sufficient to guarantee observability for all weak solutions with bounded measurable potentials up to a sufficiently large time T.

  6. Improved system can perform finite-chain repair operations on oscillatory wave solutions near intersecting straight interfaces to produce smooth pairs that preserve the zero set of the solution while satisfying the wave equation. This function allows AI to construct physically realistic, smooth solutions that vanish on observed polygons for all time by merging local nonunique constructions along edges and vertices.

  7. Improved system can perform global assembly of localized solution pairs into a complete global counterexample demonstrating failure of exact controllability for arbitrary bounded measurable potentials on domains with analytic boundaries. This capability allows the AI to generate a single, complex construction combining local solutions along edges, vertex repairs, and gluing to a positive static pair that vanishes on observed polygons for all time.

  8. Improved system can analyze the phase and normal derivative scales of oscillatory wave solutions to rigorously classify when observability holds or fails under various amplitude envelope conditions. This allows the AI to predict whether a solution is governed by a pure scale, a mixed-derivative term, or a combination thereof, which is critical for proving nondegeneracy conditions like Lemma 7.4 in the context of merging two solutions.

Sources

Related papers