
This is an interactive problem.
Vim, Emacs, and Nano are playing a guessing game. Vim secretly told Nano a random binary sequence $\{a_i\}$ of length $n$. Emacs can query Nano with a set of indices $I \subseteq \{1, 2, \ldots, n\}$. Nano will reply with $\sum_{i \in I} a_i$. Could you please help Emacs find $\{a_i\}$ in less than $n/2$ queries? Additionally, the total size of the sets in all queries must not be greater than $3n$.
The size $n = 10^5$ in all tests. The example with $n = 4$ shows the format but will not be tested.
There are at most $50$ tests in this problem. The tests were generated randomly, but are fixed in advance. In each test, every binary sequence of length $n$ had the probability of $1/2^n$ to be generated.