No label defined (Q4954191)

There are many conjectures as to whether or not a fraction \(m/n\), \(n,m\in \mathbb{N}\), can be represented by a sum of unit fractions with \(k\) summands \[ m/n= 1/t_1+\cdots+ 1/t_k \quad (\text{repetitions of \(t_i\) allowed}) \] [which is of course an old conjecture by \textit{P. Erdős}, Mat. Lapok. 1, 192-210 (1950)]. Let \(E_{m,k}(N)\) be the number of solutions for given \(m\) and \(k\) with \(n\leq \mathbb{N}\). \textit{R. C. Vaughan} [J. Number Theory 5, 64-79 (1973; Zbl 0266.10041)] derives a nice estimation of this term. Here the author improves this by lowering some constants [see \textit{C. Elsholtz}, Trans. Am. Math. Soc. 353, 3209-3227 (2001) and the review of this paper by his thesis advisor \textit{W. Schwarz} in Zbl 0982.11016]. The rest of the dissertation is concerning historical questions [e.g., \textit{P. Bartoš}, Čas. Pěst. Mat. 98, 261-264 (1973; Zbl 0267.10018), \textit{P. Bartoš} and \textit{K. Pehartzová-Bošanská}, ibid. 96, 294-299 (1971; Zbl 0218.10023), \textit{C. P. Popovici}, An. Univ. Bucureşti, Ser. Şti. Nat. Mat.-Fiz. 10, No. 29, 29-44 (1961; Zbl 0117.28101)]. Studying some of these references, I was delighted to find the article of \textit{R. C. Vaughan} [in: Tagungsbericht Zahlentheorie 1970, 203-207 (1971; Zbl 0219.10024)] where he announces that Prof. Diamond told him that \[ E_{4,3}(N)\ll N(\log N)^{\varepsilon-1} \] -- does somebody know a little bit more about this?