P7848 "JZOI-2" Filling Out the Questionnaire

All the members are thinking about organizing the anniversary celebration, but Xiao Xi just wants to slack off. So Xiao Xi opened his favorite game.

After playing this game for so long, today Xiao Xi wants to join an event and fill out a questionnaire to see how well he understands the game. There are a total of $n$ true/false questions on the questionnaire. Poor Xiao Xi found that he cannot do a single question, so he decided to ask his teammate for help. His teammate is a veteran user of the game and can do all the questions, so Xiao Xi decided to ask the teammate for assistance. Specifically, each time Xiao Xi will ask in the following format: `int ask(int m,vector w,vector ans)` Here, $m$ indicates the total number of questions in this query, $w_i$ indicates the question number of the $i$-th queried question (you need to ensure that in a single query the $w_i$ are all distinct, since the teammate is very impatient), and $ans_i$ indicates the answer that Xiao Xi believes is correct for the $i$-th question ($0\le i

The sample data belongs to the first part and is given in the following format: $n\\a_{1...n}$ where $a_i$ indicates the correct answer to question $i$.

The following shows the error messages $\text{Wrong[x]}$ from `grader.cpp`: 1. Some $w_i$ in the query is greater than $n$ or less than $1$. 2. Duplicate $w_i$ appears in the query. 3. $m$ does not match the length of $w$ or $ans$. 4. The answers are wrong or the returned array format is incorrect. After that, `grader.cpp` will return your number of queries.

### Sample Explanation One feasible plan with $2$ queries is shown below. `ask(1,[2],0)`, the return value is $1$, indicating that the answer to question 2 is $0$. `ask(2,[1,3],[0,1])`, the return value is $2$, ~~maybe because the luck is too good and you guessed everything right~~, so we can conclude that the answers to questions 1 and 3 are $0,1$. So you can return $\{0,0,1\}$, and the number of queries is $2$. Of course, if the result of the second query is $1$, then you cannot directly determine the answers. So, guessing has risks. ### Constraints For all testdata, it is guaranteed that $n\le2^{17}$. To ensure the quality of users' answers and to prevent users from guessing all true/false questions correctly, the questionnaire creator will carefully construct the answer to each question. Of course, it is guaranteed here that the answers already exist after the questionnaire is made, and they will not be constructed based on your queries. Assume that for each testdata you made $Q$ queries. The scoring standard is given below. | Range of $Q$ | Score | | :----------: | :----------: | | $(2^{17},+\infty)$ | $0$ | | $2^{17}$ | $10$ | | $(64000,2^{17})$ | $\min(90,\lfloor\frac{2^{17}-Q}{2^{16}}\times100\rfloor)$ | | $(63000,64000]$ | $95$ | | $[0,63000]$ | $100$ | Note that this problem has multiple test points, and your final score is the minimum score among all testdata. Implementation details: be sure to add `extern "C"`. You may choose to fill in the functions in `problem.cpp`, or implement them yourself. When compiling, directly compile and run `grader.cpp`. Friendly reminder: if your random approach performs too poorly, even if your number of queries is less than $2^{17}$, your score may still be less than $10$. Translated by ChatGPT 5