| № | Пользователь | Рейтинг |
|---|---|---|
| 1 | tourist | 3985 |
| 2 | jiangly | 3814 |
| 3 | jqdai0815 | 3682 |
| 4 | Benq | 3529 |
| 5 | orzdevinwang | 3526 |
| 6 | ksun48 | 3517 |
| 7 | Radewoosh | 3410 |
| 8 | hos.lyric | 3399 |
| 9 | ecnerwala | 3392 |
| 9 | Um_nik | 3392 |
| Страны | Города | Организации | Всё → |
| № | Пользователь | Вклад |
|---|---|---|
| 1 | cry | 169 |
| 2 | maomao90 | 162 |
| 2 | Um_nik | 162 |
| 4 | atcoder_official | 161 |
| 5 | djm03178 | 158 |
| 6 | -is-this-fft- | 157 |
| 7 | adamant | 155 |
| 8 | awoo | 154 |
| 8 | Dominater069 | 154 |
| 10 | luogu_official | 150 |
By long break I mean 3 months or more.
I recently experienced a great positive effect on my job-related performance after having a long break. It seemed weird, I expected an opposite effect.
So my question is, have you ever:
Did your performance increase or decrease?
В этом блоге я сделал тест, который поможет вам определить, оптимален ли ваш сетап или нет. Ниже я напишу список критериев, по которым я буду его оценивать и разбаловку. И в конце вы сможете узнать, к какой "группе людей" вы относитесь, прямо как в тестиках на персонажа из аниме в Вконтакте.
Я сюда включил практически всё, что мне пришло в голову, если я что-то забыл, предложите в комментариях. Но добавлять вещи как У меня есть интернет и На моей клавиатуре есть большая часть клавиш я не буду :)
Это несерьёзный блог, но если вы новичок, вы, может быть, найдёте для себя что-то новое. Но это всё равно никак не поможет вам поднять рейтинг, для этого нужно решать задачи.
Я видел много школьников, которые отправляют посылки никогда не запуская. Даже в институте такие встречались. На последнем див 4 соревновании [73 страницы Compilation Error] в задаче А (https://codeforces.net/contest/2009/status/page/1?order=BY_JUDGED_DESC) Кажется, таких людей даже больше, чем я ожидал)
Если вы не умеете это делать, вам нужно этому научиться. Даже турист проверяет свои решения на сэмплах, иногда даже в простых задачах
Когда я в школьные годы участвовал в олимпиадах, я постоянно мучался с предложенным компьютером. Он зависал, крашился, несколько раз мне даже давали поменять компьютер. Но я тогда ничего не знал и зачем-то пользовался вижлой на слабых машинах. Это очень неудобно и тратит много времени
Существуют такие онлайн компиляторы, которыми я совсем не рекомендую пользоваться. Зависеть от посторонних сайтов или от интернет провайдера — странная проблема, её лучше избегать
Если вы дальтоник, или вы, как мой батя, просто привыкли писать синими буквами на синем фоне, то подсветка кода разными цветами должна вам помочь, это облегчает его чтение.
Полагаться на то, что если на вашем компьютере работает решение < 1 секунды, то на проверяющих серверах произойдёт то же самое — нельзя. Но иметь возможность сравнить, быстрее ли работает ваше решение или нет после какого-нибудь фикса — бывает очень полезно.
Иногда сложно точно сказать, сколько памяти потребляет ваша программа на макс тесте, и бывает удобно запустить её самому и проверить это, а не тратить лишнюю одну посылку
Если у вас не включено подсвечивание ошибок на ходу, то тогда вам нужно перезапускать код несколько раз, чтобы исправить такие ошибки как vctor<int> A;. Это тратит лишнее время
У некоторых задач слишком жёсткий тайм лимит, чтобы заливать её на медленных языках. Иногда это всё-таки можно сделать, но нужно приложить дополнительные усилия. Я не буду называть какие же языки нужно относить к "проблемным" — у меня нет такой компетенции, но если вы когда-нибудь сталкивались с похожими проблемами, вероятно, вы неоптимальны.
Писать дерево отрезков руками 3 раза за контест — точно плохая идея. Скопировать его даже 1 раз всегда быстрее. Особенно если вы можете это сделать, не выходя из своего редактора. Часто приходится в алгоритмах что-то поменять, поэтому лучше пользоваться своей библиотекой, чтобы на 100% понимать, как всё работает, и если что быстро менять.
Копировать тест каждый раз при запуске программы — тратит очень много времени. Не стоит выполнять лишнюю работу. Также не стоит забывать, что иногда удобно иметь возможность вводить тест по частям, например в интерактивных задачах.
Я использую C++ и после того, как я настроил флаги компиляции, стало намного удобней. Раньше сообщения мне совсем не помогали, а теперь я всегда смотрю на окошко ошибок. Я не смог настроить по-нормальному санитайзеры, так что знаю, что можно сделать ещё удобней. Если вам интересны такие вещи, советую почитать этот блог.
Есть удобные тулы которые позволяют копировать тесты из контестов, очень удобно.
Из-за того, что на кодфосес всё ещё есть взломы, лучше делать код понятнее. Так как всегда лучше, если вас взломают на контесте и вы исправите ошибку, чем если вы упадёте на системных тестах. (Ну, кроме анти-хэш тестов и прочего)
Даже если вы не используете дебаггер, иметь возможность запустить его — точно плюс. Принтить в консольку не всегда быстрее.
cout или print для дебага.
Если вы не слышали про стресс тестирование, то вы получаете 0 очков за этот пункт, но также вам стоит посмотреть это крутое видео, которое хорошо про это рассказывает.
Выделять тексты мышкой — точно не оптимально. Если вы убираете постоянно руки с клавиатуры, или вы перемещаетесь по тексту стрелочками, знайте, что есть способ лучше, vim motions очень вас ускорит.
Если вы не знаете свою скорость, рекомендую затестить её тут. Я понимаю, что скорость печатанья это не совсем "сетап", но, мне кажется, это достаточно относится к делу, чтобы включить в блог.
Вообще я думаю, что скорость печатанья почти не влияет на результат, за исключением крайних случаев. Если вы печатаете уж слишком медленно, это может вам мешать. Также если вы очень очень быстрые, то тогда вы можете заработать много рейтинга в speedforces контестах.
Часто бывает, когда не дописав текущую задачу, вы начнёте писать другую. Если для этого вам нужно копировать куда-то ваш код, то тогда вы тратите лишнее время.
d = ((a+b)%mod - c + mod)%mod * e % mod
d = (a+b-c)*e
Когда в самплах большие тесты, сравнивать их вручную с правильными ответами похоже на игру "найди 5 отличий на картинке". Лучше автоматизировать этот процесс.
Теперь сложите все очки, которые вы получили, и найдите себя в табличке ниже, чтобы узнать какой же вы спортивный программист:
Кстати я веду занятия по программированию, $$${1900}$$$ руб в час за индивидуальные, $$${600}$$$ рублей за групповые занятия (группы из 3х человек), бесплатное первое занятие. Напишите мне на Codeforces, если заинтересованы.
I noticed that sometimes I do some horrible things while solving easy problems. If you struggled with the first $$$2$$$ problems from the previous round, you might also have made some huge mistakes that should not happen.
The point of this blog: I will show you my terrible ways of approaching each of these problems and you will be able to see all of the stupid things that I did. Then I will show you the proper way to approach the tasks, and then you can think which approach is more similar to yours, and maybe it will help you to improve.
It is clear as day that everything I did here was horrifying.
a) I did not read the statement properly and was solving a different problemb) I did not believe in the "proof" I came up with, it was just hopeful guessing with self-deception.c) The whole thing from start to finish is very messy. I did not think clearlyI think there are $$$2$$$ main reasons why all of it happened:
1) I was trying to be as fast as possible and was cutting corners2) I was treating the problem as an "easy one", and was not properly solving it. I was hoping that my experience would be enough to solve the problem without putting any effort into it.=> no matter how we divide the string into $$$2$$$ substrings, it will always be good.I think the paragraph above is so crazy that it is pretty difficult to point out specifically what was wrong(everything). Here is my attempt:
1) I have made a lot of incorrect assumptions, it seems like I did not bother to prove anything.2) I was not thinking about the problem at all, I tried to guess my way through the whole process.I think I can summarize the text above in the following way:
You should always strive to solve problems the second way, but it is hard to be disciplined enough to do so. During the contests, there is always an incentive to rush and guess, because sometimes it leads to good performance. For me, the best way to deal with it is to record myself solving problems. It is quite boring to watch through it, but I make it interesting this way: I open a screencast of a highly rated participant, I watch him solve a problem in several minutes and then I ask myself "how on earth have I spent 20 minutes on the problem", and then I watch myself doing stupid things. And in the next contest I try to stop myself from repeating the same mistakes.
If I somehow have not lost all of your respect by showing you my insane solutions to these problems and you still want me to teach you programming, send me a message on codeforces for more info, the first lesson is free (｡◕‿‿◕｡)
Consider rotating coordinate plane by $$$45$$$ degrees when you encounter Manhattan distance problem
Before I learned this trick, I had heard this phrase several times, but I never understood what it means. Turns out it is a pretty cool and easy trick. In this blog, I explain the trick in $$$2$$$ different ways, and show how to use it in practice on one of the recent problems. I tried to make this blog approachable, tested it on a $$$1200$$$~ rated participant
Let's look at this problem. I recommend thinking about it on your own for $$$10$$$-$$$15$$$ minutes.
Summary of the statement: you are given n = $$$2e5$$$ points. You have to find $$$3$$$ points, so that the Manhattan distance between each pair is d ( d is always even). Definition of Manhattan distance: $$$dist = |x_1 - x_2| + |y_1 - y_2|$$$
The solution above, in my opinion, is not trivial to think of neither to implement. But I think the trick in a way helps with both of the issues.
There are 2 ways to think about it, in a purely algebraic, or geometrical form. I suggest you read both of them.
It is a bit hard to work with the formula of Manhattan distance $$$dist = |x_1 - x_2| + |y_1 - y_2|$$$. It has 4 variables, and 2 summands of absolute values. Let's try to simplify it.
1) Let's start with introducing 2 new variables for convenience:
<=> $$$|x_d| + |y_d|$$$2) Let's divide our formula into 2 cases:
case where signs of $$$x_d$$$ and $$$y_d$$$ are the same (for this section I treat $$$0$$$ as positive number)
case where signs of $$$x_d$$$ and $$$y_d$$$ are the oposite:
For now part after " then " is the same, but in the next step we will transform both of them.
3) Under these 2 conditions we can put everything under one $$$abs()$$$ function.
When signs of $$$x_d$$$ and $$$x_y$$$ are the same, it is clear how to do it, just summarize them inside of the $$$abs()$$$ function.
And if the signs are different, let's make them the same, and then summarize them.
4) Now the most interesting part: let's try combining conditions together once again.
So we know that we can use the formula $$$|x_d + y_d|$$$ when signs of $$$x_d$$$ and $$$y_d$$$ are the same. But what happens to it when the signs are opposite?
We can see that $$$(|x_d + y_d| \color{red}{=} |x_d| + |y_d|)$$$ no longer holds true. To be more specific:
We can make a similar statement with formula $$$|x_d - y_d|$$$: when signs of $$$x_d$$$ and $$$y_d$$$ are the same, the following is true:
We can make a very interesting conclusion:
When signs are the same:
When signs are the opposite:
5) Now we have everything in place. Let's summarize it in human words.
So:
We can make a conclusion that to combine this 2 formulas we can always take the biggest one, and it will be the one we want to use.
So we end up with this beautiful formula
6) Let's rewrite it in terms of $$$x$$$ and $$$y$$$
And shuffle it around
7) lets define 2 new variables $$$s=x+y$$$ , $$$t = x-y$$$, with these variables we get:
8) One final thing we should understand is that this formula no longer uses $$$x$$$ or $$$y$$$. It means that we can calculate values $$$s$$$ and $$$t$$$ for each point and work only with them.
As you can see, the formula became a lot more pretty. And also we can look at it in an interesting way: the distance between 2 points can be caused by the difference in coordinate $$$s$$$ or $$$t$$$ (or it can be both). This property can be used to solve the problem above.
I suggest you try to think about this problem on your own for 10-15 minutes once again, you may see this problem in a different light.
1) In the picture below you can see a random point $$$\color{#A00}{P}$$$ I chose. All points that are at a Manhattan distance of $$$3$$$ from point $$$\color{#A00}{P}$$$ are located on the red square. Also $$$\color{#A00}{W}$$$ represents one of the sides of the red square.
Notice that the red square is not parallel to $$$OX$$$ and $$$OY$$$ axes. Because of it is a bit hard to visualize how it looks without drawing.
But it should not necessarily be like that. Let's make a new coordinate system with axes $$$OS$$$ and $$$OT$$$, where sides of the square are parallel (or perpendicular) to the new axes. We can achieve that by having $$$OS$$$ go in $$$\vec{(1, 1)}$$$ direction, and $$$OT$$$ in $$$\vec{(1, -1)}$$$ direction.
From this point I will use $$$black$$$ color to represent points in $$$XY$$$ system, and $$$\color{blue}{blue}$$$ color for $$$ST$$$ system.
2) Now let's think about the unit of measure, that we will use in our new $$$ST$$$ coordinate system. We will use $$$\color{blue}{(s, t)}$$$ notation to define points.
Now we have everything ready to locate every corner of the red square and point $$$\color{#A00}{P}$$$ in our new $$$ST$$$ coordinate system.
3) I think it is a good moment to learn how to transform points from $$$XY$$$ system to $$$ST$$$ system. To do that let's think in terms of vectors, I think it would be easier to understand this way.
One of the corners of the red square has coordinates $$$(2, 3)$$$ in $$$XY$$$ and coordinates $$$\color{blue}{(5, -1)}$$$ in $$$ST$$$. Let's try to make this transition between coordinate systems through formulas.
a) At first, lets understand how basis vectors of $$$XY$$$ $$$\vec{(1, 0)}$$$ and $$$\vec{(0, 1)}$$$ transfer to $$$ST$$$.
b) Then we should understand that any point can be described using basis vectors and origin point $$$(0, 0)$$$
So point $$$(2, 3)$$$ in $$$XY$$$ system can be described as
c) And now the cool part: let's use the formula from b, but instead of using $$$XY$$$ coordinates for basis vectors, let's use their $$$ST$$$ coordinates that we got in a
So the point $$$(2, 3)$$$ in $$$XY$$$ system converts to $$$\color{blue}{{(5, -1)}}$$$ in $$$ST$$$
It's a hustle to go though all of these operations every time we want to transfer a point from $$$XY$$$ to $$$ST$$$, so we should put all of the above into simple formula.
We can think about it like this:
We can summarize it as
Notice that we had variables $$$s$$$ and $$$t$$$ in the algebra section that we defined exactly like that.
4) So now we know how to convert points from $$$XY$$$ to $$$ST$$$, but it does not help us yet. Let's redraw a picture, get rid of $$$XY$$$ coordinate system (we won't need it anymore) and make $$$OS$$$ go to the right, and $$$OT$$$ go up.
5) Now interesting things start to happen: remember that red square still represents the Manhattan distance from $$$P$$$ equal to $$$3$$$, but now it looks different. Manhattan distance formula should also look different, it should represent a square that is parallel to axes $$$S$$$ and $$$T$$$. We can try to deduce what formula it will be by analyzing the picture above.
a) At first, let's try to describe the square the most straightforward way
b) let's try to simplify equation $$$-1 \leq t \leq 5$$$
<=> $$$-3 \leq t-2 \leq 3$$$ <=> $$$|t-2| \leq 3$$$Let's right down simplified formula
c) We should think about combining the statements together.
<=> $$$|s-2| = 3$$$<=> $$$|t-2| = 3$$$so we get this:
I think here it is clear that we can use $$$max$$$ function to combine these $$$2$$$ statements
Values $$$(2, 2)$$$ represent coordinates of the point $$$P$$$ in $$$ST$$$ system.
So the general formula will look like this:
Notice that it is the exact same formula that we got in the algebra section of the blog.
From this point we can follow the editorial for the algebra approach. Or we can be more interesting: let's try to solve the problem above with visual approach (just like we did in the editorial without the trick), but this time it will be a bit different: because we don't have to deal with diagonals, the implementation will be much easier.
And also, but this is just my opinion, I find it a bit easier to come up with this approach while using the trick, I only thought of this solution after thinking in terms of $$$ST$$$ coordinates. I know that there is little difference, but I find it a lot easier to imagine it in my head.
I think that the approaches I provided for the problem above are not that difficult, but that task was marked as $$$\color{red}{2400}$$$ problem. When I saw that problem in the contest, I also had no idea how to solve it. But after learning this trick, it seemed pretty easy to me. I hope you also don't find this problem hard anymore after reading this blog.
Also I give online classes, $$${\$25}$$$ per hour for one on one lessons, $$${\$8}$$$ per hour for lessons in groups of 3, free trial lesson. Contact me on Codeforces if you are interested or for more info.
I am on Windows, I don't want to use WSL or Visual Studio. I want the compiler to tell me about stupid bugs. But I cannot figure out a way to do it.
The problem: I cannot figure out a way to catch both of the bugs below:
#include<iostream>
#include<vector>
#include<set>
using namespace std;
int main() {
//case 1
vector<int> V(5);
cout << V[10] << endl;
//case 2
set<int> Set;
cout << *Set.begin();
}
Clang allows me to use sanitizers on windows. I was not able to find a way to make them work in mingw. Here are my flags:
I can successfully catch the case 1, in the error window I can clearly see what the mistake and the line where it happens.
But I get no errors when I have the case2 bug.
I found 2 useful blogs on codeforces on this topic. There I found about the -D_GLIBCXX_DEBUG flag, but it does not work for me with clang, so I used mingw.
Case1 is worse than in clang case, because I don't see the specific line
But the case2 is better than with clang, at least I know that something is wrong and what type of mistake I made, but sadly I don't see the specific line.
I feel like I am doing something very wrong.
If you have a setup on windows that allows you to easily identify both of the bugs, I would be very interested to see how you did that.
I think this info can be useful to a lot of people who use windows and avoid using IDEs like visual studio (like people with low end pcs). I have seen too many people, including me, who manually search for this kind of bugs, and I think it should not be done like that.
Basically the title.
Why the problems seem so much easier if you go to sleep and return to it the next day? It happened to me numerous times. I struggle for hours with a problem, I give up and move on with my day, I return to it in a day or two, usually I don't remember the statement, but after reading it I solve it almost effortlessly. It is like my brain solves the tasks on its own throughout the day, but I find it hard to believe that it is able to do that if I don't even remember the statement. Sometimes I wonder why I bother thinking about a problem at all if I get stuck for more than 10 minutes. I know that almost certainly I will be able to solve it the next day (if I pick an appropriate difficulty for me obviously).
I have pretty much 0 knowledge on how the brain works, and I wonder why this weird thing happens. I know that it is a common knowledge to "sleep on the problem", I was told that in school for example, but I have not seen anyone talking on how MUCH it affects the results.
I also wonder, if it possible to master this "power" lol?
It sounds weird, but I have an "ultimate move" during the contest to go to the toilet, for some unholy reason I consistently was able to find creative ideas in a span of several minutes I spent there. It seams like forgetting about the problem, and then returning to it works just like turning a device off and on, seems to fix most of the problems. But it is so hard to do it during a contest, especially when there is almost no time left. I also know that skipping a problem and then returning to it might be effective when you are stuck, but the effect is so much weaker for me. It helps, but just a bit.
У меня 560 баллов из 600, но я не прошёл. Интересует, есть ли кто-то с меньшим/таким же количеством баллов, кто смог пройти? Хочу узнать, то ли тоталы у всех, то ли им не понравилось что я про себя написал.
