I started around 2016, so I was just a beginner back then. Everyone starts somewhere, and it took me a while to get better at problem solving :)

I see a steady and constant growth in your rating graph, so you should be proud of yourself and keep it up. I honestly think everyone with enough practice can reach quite high ratings and make it to national/international competitions. Some people learn faster than others, and some people just have more time to practice, which most of the time sets the line between "winners and losers." But I will tell you a secret. By doing competitive programming, at least during school, you are only winning. I have seen much greater benefits of it for non-competition related things.

Regarding your sleep schedule, I am not an expert in the topic because my sleep schedule changes a lot, but make sure to get enough sleep (7 to 8 hours is usually pretty good). I think you should look into other posts for practice resources, I personally have found solving problems from OI contests and AtCoder work quite well for IOI prep.

I would say, in the long run, it's probably consistency. Of course, there are some really talented people out there who can get very good without spending as much time, but I think in the general sense solving problems consistently will get you the best at it.

Ik you didn't ask me but the answer is high IQ.

Why did you choose that username?

I was stuck in newbie for a while when I chose this username. Ironic — as soon as I switched to this username, my rating started to increase lol.

Do you think that undergraduate students who already have APIO/IOI medals should train and participate in ICPC, or should they spend their time to learn something else instead? Same question for students who have no prior medals.

I joined university-level and regional ICPC at MIT just to get together with friends and have some nostalgic times. Given MIT's strong team, we had no intention to make it to the WF. However, if you have time to practice and enjoy doing competitive programming, I'd say an ICPC participation or a medal would be nice. Yes, you don't really use IOI/ICPC-level skills during a development job, but I have seen a lot of respect from experienced people (e.g., interviewers for companies) people due to my background in competitive programming. People know that it's a pretty hard field. It can also open up new connections, work opportunities at sponsors, etc. But again, only if have time and like doing it. I personally had no time to do more competitive programming.

I see a nice growth, congrats! For newbie/pupil level, the next step is to be consistently able to solve A and B problems in Div 2 (and sometimes C). These problems tend to be implementation heavy and not require knowledge of more advanced algorithms/data-structures. Sometimes they just want you to make a small observation, which improves by better intuition, which in its place requires practice. Thus, my personal suggestion is:

1. Make sure to understand C++ STL quite well, since it's used often in A/B tasks (set, map, queue, stack, etc.).
2. Go through the past contests one by one and solve problems A and B from there. If problem C also has high solving rate, give an attempt to that too.
3. For each problem, mark: implementation heavy, observation, both, or neither. Make sure you are doing balanced amount of them. It's also useful to know which ones you are better at.
4. Also join AtCoder ABC contests. I used to do contests from other platforms as well but I am not sure if they still have contests.

There's also a few other useful guides I'd invite you to check out:

• https://mirror.codeforces.com/blog/entry/66909
• https://mirror.codeforces.com/blog/entry/116371
• https://mirror.codeforces.com/blog/entry/118882

Of course, I was also stuck at different levels during different times of my learning. Luckily enough, we can see the reason of being stuck due to how Codeforces problems are structured. There's a large difference from problems A/B to C in Div 2. They usually require a stronger observation or use of basic-to-intermediate algorithms and data structures (e.g., graphs, trees, binary search, etc.).

Honestly, I don't remember how I dealt with this back in time, but now I'd try to stick to a structured training to get past this level. First, I'd create a list of topics that appear in problem C and choose a balanced set of problems for each (ad-hoc included). Start with around 5-10 problems per topic, and you'll notice that you are more comfortable in some than others while solving them, which means you should keep adding problems to the harder lists. Collections like A2OJ ladders can help to choose problems. To find their topic, look at their tags -- I think it's useful for levels below expert to know the topic of the problem since it's the intuition-development stage. I'd also check out blogs I mentioned in my other comment.

I was lucky enough to experience both tech and finance industry through internships. My immediate next step, for now, is doing software engineering work in the financial industry. I liked the faster paced work and being trusted with important projects.

I have also gotten into master's program, which might come up in the future.

When I did competitive programming, I rarely proved my solutions. Most of the time, it was simply because I could just code it up and know if it works. In real world, this is not considered great practice since the tests can never be fully thorough.

During college, my algorithms classes focused on proving solutions rather than coding up, so I have done a lot of proofs since then, which I think improved my understanding a lot. I am probably not as speedy as I was in 2020, but I definitely approach problems in more structured ways and try to do little proofs if needed.

If you find proofs hard to understand, you might want to take a "Math for CS" class from one of the universities online. MIT has its class up for free: https://openlearninglibrary.mit.edu/courses/course-v1:OCW+6.042J+2T2019/about. After taking mathematically heavy proof classes, I found myself much better at proving statements/solutions.

It looks like you are can steadily stay in specialist level (congrats!), so the next step would be to train to master the expert rank. I think this is around the point Codeforces alone becomes insufficient to train, so I'd highly suggest also joining/solving AtCoder competitions. This website has a nice table to keep your AtCoder progress. I also suggest applying the same strategy I described in my comment above.

In addition, you should start getting comfortable with slightly more advanced topics, so combining the strategy with USACO Guide would work quite well (Silver, Gold).

I know a lot of Codeforces grandmasters and they are all extremely brilliant people, and all of them are very hard working. Some of them are very fast learners, and some others took their time to reach to the rank. I am not to speak too much since I couldn't reach that high, but talking to them, I would say it requires a lot of effort and consistency. More importantly, what this means is, you'd have to make some sacrifices -- not only the notion of "free time" disappears, but also you'll spend less time on other activities (gaming, watching movies, family time) for some period.

I personally prioritized other parts of my life and stopped doing competitive programming during college, but I am very happy that I spent a lot of time on it during high school -- probably one of the few things that got me to where I am.

recommend some math textbooks that may help in CP

For math practice, I highly suggest MIT's 6.042 class: https://openlearninglibrary.mit.edu/courses/course-v1:OCW+6.042J+2T2019/about. It's pretty good and goes over pretty much everything you'll need to know before reaching grandmaster. CP-Algorithms and USACO guide also has pretty good tutorials on topics you need.

where do you work ? what's your market track

I still think about longer-term plans, but for the time being, I'll be doing software engineering in the financial industry. But I also have gotten into a master's program, which will probably happen at some point.

why u solved only 600 problems. what other judges u use

I was a pretty bad competitor in terms of consistency, so I'd mostly never upsolve, which can explain why I am not a grandmaster. But I used to use AtCoder, oj.uz, Codechef, Hackerrank.

how can i be Golovanov399

There's perhaps a lot to learn from Alexander, but I think you should stay being yourself.

See my two comments above: one and two. Happy to answer a follow-up, though.

Please check out my other comments from above (it's getting harder to link), but I think the main step is figuring out what keeps you from doing better in contests. You will naturally be better at solving some types of problems than others for the time being, but if you realize that and practice correctly, you'll see big differences. For examples, I was great at math problems but really bad at graph. In fact, I'd avoid graph problems, which kept me from doing better. I saw a very noticeable progress as soon as I started doing a bit of graph practice. Thus, I suggest making a list of topics that you should be good at and starting out with 5-10 problems for each (of balanced difficulties). Mark your progress and efforts on each to see what's working better and worse, and then focus on your weak spots. Some topics, as you mentioned, like greedy/ad hoc/constructive are not very easy to practice, but your intuition for them will develop over time as you solve problems (not just problems in those categories but pretty much all).

How much math do I need and how can I reach that level?

Realistically, the math section from USACO Guide — Gold covers topic-wise what you need. And it will probably be enough math until you reach to IM/GM ranks. However, math also requires a lot of intuition, and looking into some topics from MIT's 6.042 might help with that.

I would love to know the way you were practicing.

Honestly, not great, but it was fun times! I spent a lot of my time working with friends and teaching underclassmen during school. In theory, I could probably spend more time trying to get better ranks, but everyone has different goals and I'm happy I could make a lot of great friends by forming a tight competitive programming community in my country.

Other than the logistical requirements (SAT, English scores, etc.) I think they care much more about what kind of person you are and who you want to become further. Here's an official blog on what the admissions office is looking for: https://mitadmissions.org/apply/process/what-we-look-for/

Having an IOI medal helps, but it's not a requirement — I did not have one when I got into MIT. Having a very high SAT score helps too, but it's also not too strict — I believe I had 1440, but my friend had only 1320 (out of 1600). There's also bunch of small essays you'll have to write. These are probably the most important part of your application, since it's the only way (other than a potential interview, which comes later) the admission officers get to know you personally. mitadmissions blog has examples of great ones, but let me know if you need me to link some. Well-written letter of recommendations from your teachers also helps. In addition, it's important that your whole life does not revolve around doing competitive programming — you have to show them that you are also willing to take on real-world tasks, help others, be interested in other topics, etc. Again, the blog is very great to have a sense of what you should need. One common mistake is to make all of your essays about olympiads, so while do definitely mention them in your essays, don't give the impression that you didn't do anything else.

Ultimately, as amazing place MIT is, I think there's more important things than education (let alone where you study) such as your values and attempts to become a better individual while sticking to those. But if you do get in, you should be very happy, because a big, fun journey is waiting for you.

What’s the probability I get into MIT if I apply next year?

It's always hard to say! I applied to 20 universities in the US, and I got rejected from 19 of them. There's definitely a lot of luck involved, but I see that you are a grandmaster, which tells me you are ready to take on academic challenges and probably even have national/international achievements, so also make sure to start building up your personal non-academic profile.

I downvote everyone who's name is Amir_Parsa

It's hard to say exactly, but here's my best guess.

• Informatics and math competitions. While I didn't have an IOI medal when I got in, I still had good amount of other awards, so they definitely helped.
• Promoting and teaching for informatics competitions. I probably spent more time teaching than practicing myself, but I'm happy I did that (though different people have different goals!). When I started competitive programming, I believe the max rank in Azerbaijan was expert, and there was only a few of them. Now there's even a grandmaster!
• Coming from an interesting background/family. My father is a journalist/human-rights defender in a country with no free speech, so certain things hasn't been easy while growing up, which allowed me to develop better understanding of the world a bit earlier on.
• Good letters of recommendations. I always had great personal relationship with my teachers and my school principal, so I believe they wrote me a quite good letter.

However, note that everyone who gets in have their own unique reasons. I suggest reading the MIT Admissions blog to get some more diverse sense.

My math background definitely helped me a lot. I think for master rank, you only need to know the math section from USACO Guide Gold, but knowing a bit more about proofs/induction can also help to formulate better ideas. See comment above: https://mirror.codeforces.com/blog/entry/130195?#comment-1156444

Here's a great overall resource for CS classes of any type: https://github.com/ossu/computer-science

For math, I'd suggest MIT's 6.042 and for algorithms, I'd suggest 6.006 and 6.046 from MIT. Stanford and CMU also has pretty good courses. Find links in the github repo.

Math background definitely helps, especially if you did bunch of combinatorics. Here's a nice combinatorics book to read when you have time: https://archive.org/details/olympiad-combinatorics.

In general though, ad hoc problems are mostly about having a good intuition, which comes from recognizing similar ideas and making dots connect -- which you get better with more exposure to problems and solutions. I don't do competitions anymore, but sometimes I still read a problem, think about it for 3-5 minutes, and then read the solution immediately. My goal is not to be able to solve it but give myself more exposure on problems and solutions.

Also, what do you do when you get stuck on a problem and can't think of any new approaches/ideas for a while?

During the competitions, it's harder done than said, but refreshing your mind on the problem, or even switching problems a bit helps. I remember in IZHO 2020, I was very stuck on a problem and went to the bathroom, washed my face with some cold water and thought a bit about non-competition stuff for 2-3 minutes. When I went back, I thought of new ideas and was able to do quite well. Also, it's a good idea to have a little checklist of general topics. Some really great problems have very interesting twists. E.g., IOI 2012 Ideal City problem comes down to a much simpler idea you usually don't think about, so it's good to keep a list of areas (e.g., greedy, dp, graph, tree, binary/ternary search etc.) and see if you can transform the problem in one of those.

As you can see on my profile, it also took me more than a year to reach pupil rank but then I quickly started getting better. I'd like to make an analogy to the temperature of water vs time graph:

I think practice is like that too -- when you begin, you start immediately learning a lot of things and being able to solve very simple problems, but then you feel stuck there. The key is to keep practicing and learning. Please see my comments above for some tips on practicing for Div 2. A/B level problems, in addition to some other blogs by others I linked. Make sure to learn the language well (e.g., for C++, STL is super important) and do a lot of practice. I strongly suggest starting solving AtCoder ABC problems as well.

I think the same advice as in the comment above also applies, but happy to answer more questions! https://mirror.codeforces.com/blog/entry/130195?#comment-1157213

Hey. I just wanted to point something out:

Why are your templates changing with each submission? [link-1, link-2, link-3]

To be honest, copy-pasting 1600-1700 rated problems will not help you. Be truthful to yourself, and try to solve Div2-B (Also, if you are able to solve 1600 in practice, you should definitely by able to solve Div2-B in contest). Increasing number of solved problems by copy-pasting is just pathetic. Don't do it.

How solve div2 problem C in 6 min? cheater

Congrats on reaching pupil! It's good to hear that you are also practicing math. It sounds like you are having difficulty with implementation-heavy problems. Unfortunately, the only way to attack this weakness is by doing implementation problems. Problems of rating range 1600-1700 might be a bit too high -- for now, I'd focus on being able to quickly implement 1200-1500 range problems and actually coding them after figuring out the solution. It helps to write out some logic on paper -- a plan of what you'd be coding, and then actually doing the coding. Once you can consistently solve Div 2. A and B quickly and sometimes solve C, I think you should be reaching to a higher rank. Then USACO Guide can also help to master some more advanced topics.

Please also check other comments for suggestions -- I'd like to keep repetition to minimal.

I started off using CodeBlocks on Windows. To be honest, it was pretty good -- you don't need much for a coding competition, just a text editor and a run button. Later on, I switched to Ubuntu and started using Atom as my editor and used custom scripts to compile and run (side note: Atom stopped development).

Right now, I am on Arch with GNOME (Wayland) and I use VSCodium as my editor for pretty much everything, but I rarely do contests anymore. I do a lot of C/C++, Rust, and web development though.

I see lots of contests and a nice grow from newbie to specialist -- good job! I think it's about time for you to start practicing to be able to solve Div 2. C problems (or easy Div2. D). You'll have to find your weaknesses to get better. I'd suggest a strategy similar to this comment: https://mirror.codeforces.com/blog/entry/130195?#comment-1156435

Find which topics you need more practice in by experimenting and grind them, and you should be feeling much more comfortable with them shortly. Don't only think about why the solution is the way it is, but also how one could start thinking about the solution -- the thought process starts to matter a lot more in this level. Happy to answer more questions!

You did hit me with the hardest question here! Firstly, I will start by saying that you can make whichever university you go to very special for you -- in some of them, this will be easier and I think MIT is one of them. For me, MIT was a place of many different journeys and I grew in ways I didn't expect there.

I enjoyed two things the most: the people and the teaching opportunities. You could put the name "nerd" on everyone at MIT, but you'd be surprised about the diversity of people and their interests. Even if it's something extremely niche, you'll see someone doing it. Keep meeting different people and you'll find yourself doing things you have never even thought of before in your life. I met so many people and had so many different experiences with them (late-night discussions, long coding sessions, hacking [1], dumb stuff around campus, partying, etc.) and I'm very grateful for these, and I sure hope that these connections and friendships last forever.

I also got myself into helping to teach some classes by first becoming a lab assistant for a programming class. Basically, a lot of students come for help and you do 1-to-1 sessions with them to help them understand the material better or debug their code. First, my plan was to keep doing this so I can become a TA (teaching assistant) so I can get free funding for master's, but in no time I realized that I actually love teaching a lot, so I spent around 3 years at MIT helping to teach 4-5 different classes. Teaching is quite hard because the main point is to not give away the answer but help the student find the answer themselves, but the satisfaction I got from getting them there and seeing them being so happy was far more than any "Accepted." Being a TA, you also get to develop material for classes, and even though one might say "these are MIT students they can figure out," we tried quite hard to make it easy and even were very careful about wording of the content. I also enjoyed educational content creation a lot.

[1] https://handbook.mit.edu/hacking

Does it have alternatives elsewhere?

I think it sure does. Honestly, some advanced classes might be taught better in other universities since MIT has great researchers but sometimes they prefer to put less effort on teaching. I remember referring to Stanford, CMU, and some European university courses for additional resources. I think great people exists everywhere, and it will take a bit of effort to find compatible people with you, but remember that college is a temporary place, so make it fun wherever it is.

Eh, I prefer vectorizing manually by aligning the memory and using AVX intrinsics ( _mm256 vectors). /s