{"id":"arxiv/2609.38425","url":"https://arxiv.org/abs/2609.38425","tags":["math.PR"],"score":0,"title":"Majorizing-measure bounds in the realized square-function metric","domain":"math.PR","source":"arxiv","authors":["Witold M. Bednorz","Rafał M. Łochowski"],"reviews":0,"subject":"math.PR","summary":"The predictable square function measures the size of a martingale. Applied to parameter differences, it also defines a random pseudometric. We explain how to perform a majorizing-measure argument directly in this realized geometry, without conditioning the terminal field to be Gaussian and without first replacing the metric by a random multiple of a deterministic one. For a finite family of predictable Gaussian sums in a $(2,D)$-smooth Banach space, and a fixed probability measure $μ$ on the parameter set, we prove \\[ \\norm{\\max_j\\osc_T f_j}_{L^p} \\le CD\\norm{\\G_μ(d)+\\sqrt…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-09-29","reactions":{}}