The equation ax - xb = c posed not for matrices but for an unknown x in a Banach module M over a pair of Banach algebras A1 and A2, with a in A1, b in A2 and c in M. The classical theory solves it only when the spectra of a and b are disjoint, where a contour or Bochner integral formula produces the unique solution; this concept covers the singular case, where the spectra overlap and solvability becomes a genuine question. The technique is to embed the problem in the 2x2 operator matrix algebra built from A1, A2 and M, where consistency of the equation becomes a similarity statement about a block matrix, and to characterise solvability through nonprimary square roots of an element similar to diag(a,-b). Along the way you see why Roth's removal rule, which settles the matrix case, does not transfer, why the homogeneous equation ax = xb - the intertwining problem - controls the whole solution set, and why an equation as simple-looking as ax - xa = 1 has no solution in any unital Banach algebra.
Let A1 and A2 be unital complex Banach algebras and let M be a Banach (A1,A2)-module: a Banach space carrying compatible left multiplication by A1 and right multiplication by A2, with a submultiplica…