Answers are grounded in this paper and cite their sources.
Unrefereed research candidate. In one sentence: five fixed lists of numbers force a running total to reach 5/2 in every even dimension from four up, so in dimension four the best possible balancing constant for this problem is exactly 5/2; deciding on the fly, 5/2 cannot be beaten in any dimension from four up, and from dimension eight it cannot even be reached. The question. Lists (vectors) of n numbers between −1 and 1 arrive one after another and are added to a running total. Before adding each list you may flip all its signs and shuffle its entries into any order, but you may not reorder the stream. The aim is to keep every entry of every running total as close to zero as possible; the score is the largest absolute entry ever seen. This is a member of the vector-balancing family (Spencer’s “six standard deviations”, Steinitz-type prefix problems, online discrepancy). With signs alone the best guarantee grows like the square root of n; allowing the shuffles makes it a constant. What was known. The problem, the upper bound and the lower bound 2 are due to Niles-Weed, Sadovsky and Shkrob (arXiv:2610.02127, October 2026). They proved that 3 − 2/n always suffices in dimension n, even choosing on the fly, recorded the example showing that 2 cannot be beaten, and asked for the sharp constant. Results. (1) For every even n ≥ 4, five fixed vectors force some running total to have an entry of size at least 5/2 (Theorem 3.1). In dimension four they are (−1,−1,−1,1), (1/2,1/2,1/2,1/2), (−1,−1,1,1), (1,1,1,1), (−1,−1,1,1). (2) Since 3 − 2/4 = 5/2, the sharp constant in dimension four is exactly 5/2, both when the whole stream is known in advance and when choices are made on the fly (Corollary 4.3). (3) For deterministic on-the-fly (online) strategies: the constant is at least 5/2 in every dimension n ≥ 4, odd ones included; strictly above 5/2 in every dimension n ≥ 8 (for example at least 5/2 + 1/4096 when n = 8); and at least 161/64 = 2.515625 in every dimension n ≥ 52 (Section 5). (4) In dimension three, online, at least 13/6, from a computer-checked strategy certificate with 782 states (Proposition 6.1). How. All lower bounds are counting arguments: a list with exactly h entries −1 can rescue at most h coordinates that are already in danger. For the online bounds above 5/2 the new tool is a plateau amplification lemma: once two or more entries of the running total are tied at the maximum, a short sequence of lists, each with a few entries 1 and a tiny negative number everywhere else (doubling each time), forces the maximum a little higher. The upper bound 3 − 2/n is re-derived in a short self-contained form; it is not new. Limits. Whether 5/2 can be beaten when the whole stream is known in advance is open, and so is whether 3 − 2/n is the exact answer in every dimension (it is for n = 2 and n = 4). The 5/2 obstruction concerns running (prefix) totals: for the five lists above the final total alone can be kept to 3/2. The margins above 5/2 are tiny except for the 161/64 bound. Odd dimensions have only the online bound. Contents. The paper (LaTeX source and PDF); verify.py, a standard-library Python script of exact checks (about one minute; it exits with status 1 on any failure, also under python -O) and its recorded output verification.json; explore.py and explore_log.txt, the bounded exploratory searches reported as negative results (not proofs); the bibliography from registry records; a machine-readable index; a research-gate record; a rating-free review summary; a replay receipt and SHA-256 sums. Assurance. Written proofs for every result except Proposition 6.1, which is computer-assisted. Internal AI-assisted review only (five reviewer roles, an adversarial review and audit by a second AI model, clean replays, a re-review and a proof audit): not peer review, independent reproduction or formal verification. Creator: Anonymous. Text and data CC0-1.0; code MIT.
No takes yet. Share an insight, caveat, or question.
A 2026 study studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: