For a prime \( p \), define \( f(p) \) to be the minimum \( i \) such that all residues mod \( p \) can be expressed as sums of subsets of the values \( \{\tfrac{1}{1}, \tfrac{1}{2}, \tfrac{1}{3}, \dots \tfrac{1}{i} \} \bmod p \). We need only consider residues other than \( 0 \) and \( 1 \), since \( 0 \) is always the sum of the empty subset and \( 1 = \tfrac{1}{1} \). E.g.,

\( f(2) = 1 \), because \( 1 = \tfrac{1}{1} \).

\( f(3) = 2 \), because \( 2 = \tfrac{1}{2} \bmod 3 \).

\( f(5) = 3 \), because \( 2 = \tfrac{1}{3} \), \( 3 = \tfrac{1}{2} \), and \( 4 = \tfrac{1}{1} + \tfrac{1}{2} \bmod 5 \).

\( f(7) = 3 \), because \( 2 = \tfrac{1}{2} + \tfrac{1}{3} \), \( 3 = \tfrac{1}{1} + \tfrac{1}{2} + \tfrac{1}{3} \), \( 4 = \tfrac{1}{2} \), \( 5 = \tfrac{1}{3} \), and \( 6 = \tfrac{1}{1} + \tfrac{1}{3} \bmod 7 \).

\( f(11) = 4 \), because \( 2 = \tfrac{1}{2} + \tfrac{1}{3} + \tfrac{1}{4} \), \( 3 = \tfrac{1}{4} \), \( 4 = \tfrac{1}{3} \), \( 5 = \tfrac{1}{1} + \tfrac{1}{3} \), \( 6 = \tfrac{1}{2} \), \( 7 = \tfrac{1}{1} + \tfrac{1}{2} \), \( 8 = \tfrac{1}{1} + \tfrac{1}{3} + \tfrac{1}{4} \), \( 9 = \tfrac{1}{2} + \tfrac{1}{4} \), and \( 10 = \tfrac{1}{2} + \tfrac{1}{3} \bmod 11 \).

\( f(13) = 5 \). \( 2 = \tfrac{1}{2} + \tfrac{1}{5} \bmod 13 \) and \( 12 = \tfrac{1}{1} + \tfrac{1}{2} + \tfrac{1}{3} + \tfrac{1}{5} \bmod 13 \); all other residues can be expressed as sums of subsets of \( \{\tfrac{1}{1}, \tfrac{1}{2}, \tfrac{1}{3}, \tfrac{1}{4} \} \).

How quickly does \( f(p) \) grow with \( p \)? It must be at least \( \log_2 p \) else there would not be enough different subsets to cover all the residues. It seems plausible to me (e.g. using an argument in which we replace each unit fraction by a random variable) that \( f(p) = \Theta(\log p) \).

The relevance of this is for bounding the maximum denominator of an Egyptian fraction as a function of its denominator. For each \( p \), some fraction \( x/p \) requires a maximum denominator of at least \( f(p)p \) in any Egyptian fraction expansion, as can be seen by considering only the terms in the expansion in which the denominator is a multiple of \( p \). The simple lower bound for \( f(p) \) above thus leads to a lower bound of \( \Omega(y log y) \) on the maximum denominator in an expansion of \( x/y \), for infinitely many pairs \( x, y \). On the other hand, it seems plausible (although not completely obvious due to issues with repeated fractions and overlarge expansions) that something like the prime-multiples method of my Egyptian fraction code can expand any fraction \( x/y \) using max denominator \( f(p)y \), where \( p \) is the prime factor of \( y \) with the largest value of \( f(p) \). This approach might have some hope of improving the Tenenbaum–Yokota \( O(y \log^2 y / \log\log y) \) upper bound on the max denominator.