larryzhong: 你回顾你打的不好的比赛，你会发现很多都是时间分配出了比较大的问题
\(2p+48+20=268\). Not good at all. The main problem seems to be missing subtasks with extra limitations in C and D. It's actually kinda nice to come up with the in/deduction approach which passes \(k\le 8\) with complexity that's not even close, but with the disadvantage of lim A and B, it has only grant me 8 extra points over common line. And of course, missing limitations in D was a big blunder. It costs totally 44 points and became the final barrier of getting 300+.
For a long time I believed that it was a problem of strategy or tactics, and I changed my mindset to focus on subtasks in my preparation. It worked pretty well, and surely my grades became significantly more stable after this tweak. But this time, after listening to larryzhong's opinion on tactics, another thing important has came into my mind: Time control.
In this NOIP contest, something unexpected did happen. As usual, I smashed the first 2 problems within the first 40 minutes, and after reading the C and D, C looks incredibly juicy to me and I successfully deducted it down to a inclusion-exclusion principle model after 50 minutes of thinking. This outcome worth 48 points. But now, things started to go wrong. Going all out for a full solve, I tried a different path with similar thought, firmly assumed some completely wrong conclusions, with wrong but hell convincing oral proof, spent a whole 2 hour-ish time debugging my re-rooting dp code with several attempts to cover the flaws. From time to time I feared that my assumption is wrong, but my dopamine and eagerness to get a full solve made me to completely ignore the flaws and continue to find imperfect fixes for my code with the hope to pass the next sample. At 11:52, I was staring at the failed samples approachlessly, and forced to rethink the solution. 2 minutes later, I came up with a \(n=7\) hack that completely destroyed my hope. My "different path" has came out to be a stupid placebo candy that made me ignoring some corners cases I previously came up with. All what left now is 1 hour of time and a mess with not even 1 single significant line of code for C or D. In the last hour I rushed through C and D in a hurry, with only 68 points totally implemented at last.
By the way, the lack of time seems to be the most significant reason regarding the loss of extra limitation subtasks.
That fake re-rooting dp on C was truly a nightmare. The whole 2 hours thinking about it is a complete waste.
So what do I learn?
The obvious one is "donot rush". Shouldn't have rushed so much when thinking about C. A better way is to write things down on a paper and give it a confirmation. Problems in OI usually doesn't require florid formal mathematical proof, but in my opinion simple oral proofs on key points of solutions are still not acceptable. It's much better to write it out and see how it goes on a paper before implementing next time.
And, to all-in problem C (in this contest) doesn't seem clever. Mainly because I was convinced that NOIP C is supposed to be a simple problem with around 2400ish difficulty. For once again, there aren't 1000 testers for every round, and expectations like this are dangerous. Obviously, it doesn't mean that crushing any problem is wrong, but at least do not hold difficulty expectations for every problem when trying to solve them. "This is a *2200 (for whatever reason) thus I must solve it with some dummy stuff" is not optimal.
Understand how OI style contests work, and get the hang of them. This should be one of the training goals. Time control is a major topic.
标签：Rando,my,but,Note,points,time,problem,was From： https://www.cnblogs.com/mindeveloped/p/18586676/rando-4