This is a humorous post. Do not take this seriously, unless you want to.

Share your impossible-to-implement solutions to the AI cheating problems on Codeforces here!

I had an idea that involving requiring usage of special applications that would record keystroke frequency, test whether it is running in a VM, or (with all the privacy problems it causes) monitor eye movements!

Well, this is almost impossible to implement, but what other fun ideas do you have?

So I have a script I've written for my own use that does simple compilation jobs.

Through time, I've been adding functionalities to it, and then it occurred to me that I can publish it so others can benefit and improve it.

So I've added some sanity check and made it free (as in freedom) software on Github

Currently it has the following functions:

Arguments:
  -h | --help               Show this help
  -f | --file FILE          The file to compile, without the .cpp suffix
  -i | --input-file FILE    Feed input to your program from FILE
  -c | --compare PATH       Batch compare with data in PATH
                            This option assumes that the input and output files
                            are named name_id.in and name_id.out, where name is 
                            the argument to -f and id increases from 1
  --sanitize                Use G++ sanitize options
  --verbose                 Be more talkative
  --version                 Print version and copyright information.

Just execute the compile.sh file in the repo and that's it!

Please help me improve it and I hope that it can become more powerful through community efforts.

Note that currently some options are hardcoded, they do work for me but may not work for you:

• It defines macro with -DLILYWHITE for debug use.
• It compiles in C++14

Thanks!

This is an ancient post, but I do see a need of OI mode contests in recent actions from time to time, so maybe it’s high time we considered this again.

I think this will be beneficial for people who are preparing for OIs, whether it is CNOI mode (no realtime judge feedback) or IOI mode. Since partial scoring is really different from the current CF contests.

I think maybe Codeforces IOI Round #x can be held in parallel with the normal Codeforces Rounds.

Apart from that, it would be great to see more kinds of contests. In current CF mode contests, heavy implementation is nearly impossible given the number of problems crammed into a short timespan. While in a 5h-3 tasks IOI contest, setters can put heavy implementations in. There can be even more possibility like answer-submitting problems (you download the inputs, work out the answer in whichever way you like, and submitting the output).

I have a friend who is rated purple on CF but only 1700 on AtC.

So how do they really map?

I can usually get to C/D in contests. However, I often get stuck at As for overestimating its complexity.

Apart from this, I code really slow and buggy at A, but code way better at Cs and Ds

So what should I do to solve this interesting problem of me?

So I was goofing around and playing computer games during lockdown and I've came across this problem.

Imagine I am playing a text-adventure game that has $$$n$$$ stories, by going through a story I gain $$$s_i$$$ points. After a story ends, I will face some choices, each choice are marked with a score range $$$[l_i, r_i]$$$ and I can only choose it if my current score is within the range. The choices will then lead me to another story.

I am standing at the beginning which is also a story, call it story $$$0$$$, the game ends when I reach the end of one story and cannot find a choice that is open for me (i.e my score is fit for the range limit).

Question: What's the maximum score I can get given that I know the structure of the game (how the choices are placed and their requirement, and the score increases for the stories) and I play optimally?

Is this question solvable within polynomial time?

The virtual participation system is great. However, sometimes I want to simulate a contest more realistically, and simply testing system tests becomes a problem as some problem setters tend to have weak pretests.

So here is my solution: virtual participation will only test pretests by default, but you can still turn on system test mode. And when the participation ends, a "Run System Test" button will appear and you can run systests for your own submission, just like how TopCoder works.

First things first, Codeforces blog system is awesome, however recently I had found a problem with it.

Given a segment of length $$$1$$$, randomly pick two points $$$A$$$ and $$$B$$$, you are tasked to find the expected length of segment $$$AB$$$.

For Chinese readers, Chinese statement is here:

Someone says that it is related to the theoretical proof of Chtholly Tree, but I simply cannot understand why...

Suppose we have an indefinite equation about $$$x,y,z$$$.

If $$$(x,y,z)=(2,2,2)$$$ satisfies the equation, is there other such $$$(x,y,z)$$$ groups with $$$x,y,z \geq 2$$$ satisfy the equation?

Notice, that unlike Fermat's last theorem, $$$x,y,z$$$ does NOT have to be pairwise equal.

I think that the answer is NO, but I am unable to prove it or construct such groups. Please help me :D

int n;
cin>>n;
if (n==0)//deal with it
for (double i=0;i<=1/(double)n;i+=1e-9)//do something

It seems that, this code runs faster when n becomes larger...

When n = 10⁹, the code will be executed once. When n = 1, it runs 10⁹ times.

So I estimated it as O(n^- 1).

But, this means that this algorithm runs even faster than O(1).

How is it possible?