Affine Frequency Division Multiplexing shapes signals in a modulation-symbol domain that appears to require direct joint optimization — designing the pulse shape as a single object in this combined space. The joint domain looks irreducible.
It factors.
The pulse shape in the modulation-symbol domain is mathematically the convolution of the time-domain and frequency-domain shaping pulses. This means Nyquist pulse shaping — the condition for zero inter-symbol interference — can be achieved by designing the two domains independently and letting convolution do the work. There is no need to optimize in the joint domain directly.
The factorization also reduces pilot-to-data interference, because each domain's shaping can be separately tuned for sparsity without compromising the other. The time-domain pulse controls temporal localization; the frequency-domain pulse controls spectral containment. Their convolution inherits both properties.
The mathematical structure was always present in the AFDM framework. The modulation-symbol domain was defined as a joint transform, and joint transforms of products are convolutions of individual transforms — this is a standard result. But the standard result was not applied, because the modulation-symbol domain appeared to be a genuinely new object requiring genuinely new design methods. The appearance of novelty masked the applicability of classical factorization.
This pattern — a joint optimization problem that looks irreducible but secretly factors into independent problems — recurs across engineering. The value of recognizing it is not just computational savings (though those are real: independent 1D optimizations are far cheaper than joint 2D optimization). The deeper value is conceptual clarity. When a problem factors, the interaction you thought you had to manage doesn't exist. The coupling was in the formulation, not in the physics. Separability was hidden by the domain's apparent unity.