Skip to content

Read and download the papers

Two manuscripts are drafted from the same evidence base. The LaTeX sources in manuscript/ and manuscript-quan/ are authoritative; these PDFs are regenerated by make paper and make quan and copied into the site only after a successful build.

A. Two devices, one operator

The method paper: the outgoing complex-frequency mode expansion, its geometry gradient, both the Quan–Lončar and the multilayer Design23 campaigns, and the finding that ascent directions transfer to an independent solver while absolute linewidths do not.

Your browser cannot embed the PDF. Download it here.

B. One device, in full

A separate, narrower draft: the simplest complete account of a single optimization. One cavity, 45 variables, 56 accepted steps, the field profile before and after, quality factor against iteration, and an independent solver holding a veto outside the loop. It uses the indefinite campaign run rather than the three-step Phase-58 endpoint that paper A quotes, so its numbers are different by construction.

Your browser cannot embed the PDF. Download it here.

Draft boundary

Neither paper's quality factors are publication-final. Paper A states the matched convergence panel still required. Paper B bounds its own numbers explicitly: the largest improvement it measured over a full photon lifetime is 2.12×, and every larger figure it quotes rests on ringdown fits that did not observe a full decay.