Periodic multi-pulses and spectral stability in Hamiltonian PDEs with symmetry

The paper’s Theorem 5.10 rigorously derives a quadratic normal form for the local Evans-function equation near the first collision between an interaction eigenvalue and the first essential-spectrum eigenvalue, yielding the expansion λ = λ*(r) − s i ± √(T1(r) − s^2) + O(r^{5/4}|log r|^{1/2}) and the ensuing Krein bubble with endpoints s_± = ±√T1(1 + O(1/|log r|)) at X = X*(r) ± ΔX(r) (equations (5.15)–(5.20)) . The proof (Section 11.7) implements Lin’s method, reduces the 4×4 block matrix determinant det S to a scalar quadratic equation in rescaled variables G(ĥ, k̃, r) = ĥ^2 + 2 k̃ i ĥ − T + O(1/|log r|) = 0, and then applies a contraction/implicit-function argument to obtain the circle and its perturbations . In contrast, the candidate solution repeats the same broad strategy but makes two substantive errors: (1) it asserts a(r) < 0 for m0 = 1 and hence claims λ*(r) > 0 (real), whereas the paper shows ã(0; m0, s1) > 0 for m0 = 1 (so the leading interaction pair is purely imaginary), i.e., a > 0 for the m0 = 1 branch that produces the Krein bubble ; (2) it states the bubble is “centered at λ*(r) on the real axis,” while the theorem and its proof place the center on the imaginary axis at the (purely imaginary) collision value λ = λ*(r) (in the λ-variable used) . The candidate also incorrectly calls the two roots a “complex-conjugate pair” for |s| < √T1, whereas the theorem stresses symmetry across the imaginary axis (not complex conjugacy unless s = 0) . Aside from these issues, the candidate’s reduction to a quadratic normal form with the stated O(r^{5/4}|log r|^{1/2}) accuracy and the description of endpoints and parameterization X(s,r) match the paper’s structure and estimates , and the use of a block-matrix/Lin-method reduction is consistent with the paper’s 4×4 S(λ) construction and determinant expansion . The paper’s argument is complete and correct; the model’s solution contains sign/axis misstatements that conflict with the main theorem.