CSP-S 2020 Travels
DAY 0
I hit the board in the morning before departure.
The rest of the time is decadent.
DAY 1
I was quite nervous before the exam, after all, this is the first major exam.
There is no test machine for this year’s exam, and a virtual machine is used.
At the beginning of the exam, I was not familiar with the virtual machine. When the virtual machine was turned on, there was an option to move or copy.
I asked the invigilator. The invigilator said it was a problem with the machine, and then changed to another machine.
After changing the machine, there is still such an option, and then I just choose one.
The virtual machine is slow to boot, so go to the topic first.
After reading the question of \(T1\), the virtual machine will boot.
However, the window will not be enlarged after booting. It has been almost half an hour since the test started, so I can only continue typing in the small window.
Because there was a lot of time wasted at the beginning, the idea of \(T1\) was not very clear at the beginning, basically it was thinking while playing.
So I didn’t notice that it was the Julian calendar before \(1582\), and the second example would not survive.
Then start outputting intermediate variable debugging.
After discovering this problem, almost an hour later, I found that I had written all false.
After the change, the second point of the second example is still unclear.
Comparison of output intermediate variables and violence found that the loop section was found wrong.
Quickly change the \(100\) cycle section to \(400\).
And then passed the second example.
But the third big example is still nothing.
Basically in a state of always finding errors and correcting them, but still can’t get past the big sample.
There is one and a half hours left in the exam, so I can only throw away \(T1\) first, and go see \(T2\).
\(T2\) seems to be much simpler than \(T1\), as long as you consider it according to the position, we have gone through the big example again.
In the rest of the time, \(T3\) hit a violent line segment tree, and it can actually run through the large sample.
\(T4\) played a very fake greedy, but couldn’t even survive the second example.
It should be \(n=3\) or it can be passed.
I adjusted \(T1\) in the last twenty minutes and found that the boundary was judged wrong.
Should be greater than to greater than or equal.
After the change, the sample can be oversized.
I checked the file name in the last ten minutes and the exam was over.
Because I didn't make any data to check my codes,so I was a little nervous.
DAY 2
When it was tested in Los Angeles, \(T1\) became \(40\).
Did not change the boundary of \(<2000\) to be greater than or equal to.
In this way, when encountering \(1\) month \(1\) day, it will output \(13\) month \(1\) day.
\(T2\) For everyone who forgot to mark \(vis\), the code loops from \(1\) to \(m\), which directly degenerates into violence \(65\) points.
\(T3\) was stuck as \(30\) violent.
\(T4\) ended up writing, and it ended up with \(5\) points.
The total score is \(40+65+30+5=140\), it’s about to retire.
DAY 10
\(CCF\) official data is out.
Score \(60+80+70+0=210\).
Save \(rk17\), hovering on the edge of retirement.
Reflection
\(1\). Be sure to read the questions first when you start the exam to get a general impression of the overall difficulty of the question.
\(2\), Don’t believe in big examples, be sure to make a few sets of extreme data to examine my code.
\(3\), Don’t waste too much time on one question.
\(4\), Pay attention to observe the data range, whether you need to open \(long\ long\), \(unsigned\ long\ long\) or even high precision, one or two points can be judged specially.