Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
The imported dyadic theorem says that for with irrational, every sufficiently large integer can be written as
where are finite subsets of the nonnegative integers. An index is used once in each ray; equal values at different indices are permitted by that theorem. Its recorded acceptance is on the dyadic claim page.
Fix as in L33 and put . A rational th root of an integer is an integer, and no integer has th power for . Thus is irrational and .
Choose an integer so large that
These inequalities are possible because and . The floor values with indices at least are positive, strictly increasing, and outside .
Set and . Both are positive and is irrational. Since ,
The classes and are disjoint for . Consequently each representation given by the dyadic theorem uses distinct indices of the original tail. Strict increase on that tail makes their integer values distinct as well. All those values are positive and avoid , which proves L33.
For Problem 354(ii), choose and . Completeness of this single ray implies completeness after adjoining the ray for every , whether or not is irrational. The argument works with first index or and with either indexed or distinct-value sums. The threshold is allowed to depend on every fixed parameter and the finite deletion. No assertion for arbitrary follows.