Rationally conditioned two-rational approximation conjecture

Let ϵ>0\epsilon>0 be any small positive real number and let N1N\geq 1. Suppose that α\alpha has a rational approximation a/qa/q with integers a,qa,q, 1qN21\leq q\leq N^2, (a,q)=1(a,q)=1, and

αaq1qN2.\left|\alpha-\frac aq\right|\leq\frac{1}{qN^2}.

The case qNq\leq N is immediate by taking a1/q1=a/qa_1/q_1=a/q and a2/q2=0/1a_2/q_2=0/1; thus one may restrict attention to q>Nq>N. Rationally conditioned two-rational approximation conjecture. There exist integers a1,a2a_1,a_2 and 1q1,q2N1\leq q_1,q_2\leq N such that

αa1q1a2q2ϵ1qN2ϵ.\left|\alpha-\frac{a_1}{q_1}-\frac{a_2}{q_2}\right|\ll_{\epsilon}\frac{1}{qN^{2-\epsilon}}.

This says that a sufficiently good rational approximation should yield an almost equally good approximation by a sum of two rationals; the paper notes that the bound is best possible apart from the factor involving ϵ\epsilon.

Sources & referencesView supporting material

Primary source

Tsz Ho Chan, “Approximating reals by sums of two rationals”, arXiv:math/0609322 (2007).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.