[Random number] online gaming mat equipment and thoughts

Students with no skills are more stupid than those without any skills ---- Franklin

 

When you play games, you will always have a lot of experiences. When you perform a slight search on the Internet, you will find a lot of comments about pad equipment. Many people believe that pad equipment is useful. This is a problem !!

OK. Let's simplify the problem. Because the probability of building and merging equipment is complex, it is necessary to simplify the problem. Q:

 

　　When you throw a coin consecutively, what is the probability of failure N + 1 after N times of failure? Will it be higher than 50%(!!! This is a problem we really need to solve ).

 

Coin throwing is a random event. theoretically, the probability of a successful attempt is 50% (the portrait is facing up or not), and there is no relation between the two attempts. otherwise, it will not be called a random event. I was not good at probability theory, but I still had some consciousness in my mind. I don't believe that continuous failure can significantly increase the probability of success!

However, the theory is too poor. I cannot prove that the probability of N + 1 is still so high ..... well, I can only write code to see what the simulated real situation looks like:

#include <stdlib.h>#include <time.h>#include <iostream>#include <Windows.h>#include <assert.h>#pragma comment(lib, "advapi32.lib")#define RANDOM_TIMES1000000#define FAIL_TIMES5#define FAIL_PERCENT50#define PERCENT_MAX100//#define C_RANDOM#ifdef C_RANDOM//nop#else#define RAND_MAX65535static HCRYPTPROVhProvider = 0;static const DWORDdwLength = 2;static BYTEpbBuffer[dwLength] = {};#endifstatic void random_init(){#ifdef C_RANDOMsrand((int)time(0));#elseDWORD result =::CryptAcquireContextW(&hProvider, 0, 0, PROV_RSA_FULL, CRYPT_VERIFYCONTEXT | CRYPT_SILENT);assert(result);#endif}static void random_close(){#ifdef C_RANDOM//nop#else::CryptReleaseContext(hProvider, 0);#endif}static int _random(){unsigned short _rand_value = 0;#ifdef C_RANDOM_rand_value = (unsigned short)rand();#else DWORD result = ::CryptGenRandom(hProvider, dwLength, pbBuffer);assert(result);_rand_value = *(unsigned short*)pbBuffer;#endifreturn _rand_value;}static long random_count = 0;static int random_result(){int times = 0;while(times < FAIL_TIMES){int _num = _random();random_count++;if((1.0f*_num/RAND_MAX) > (1.0f*FAIL_PERCENT/PERCENT_MAX)){times++;}else{times = 0;}}random_count++;return _random();}int main(int argc, char* argv[]){random_init();long times_total = 0;long times_fail = 0;for(int times = 0; times < RANDOM_TIMES;++times){int x = random_result();if((1.0f*x/RAND_MAX) > (1.0f*FAIL_PERCENT/PERCENT_MAX))times_fail++;times_total++;}std::cout<<"total: "<<times_total<<std::endl;std::cout<<"fail: "<<times_fail<<std::endl;std::cout<<"random_count: "<<random_count<<std::endl;system("pause");random_close();return 0;}

 

Two random numbers are used here. One is the standard C random number, and the other is CryptGenRandom. in Windows, there is no/dev/random and/dev/urandom, so which API is used instead.

I don't want to explain the code too much. The more important ones are the macros. I have nothing to do with changing the macros and running them. Let's take a look at the results:-) the two results of CryptGenRandom on my side:

 

Total: 1000000 fail: 499497random_count: 62905788 press any key to continue... total: 1000000 fail: 499914random_count: 62979706 press any key to continue...

In addition, the running result of the Standard C random number is as follows:

 

Total: 1000000 fail: 500330random_count: 63103246 press any key to continue...

We can see that the simulation results are similar several times:After N consecutive failures, the probability of N + 1 times remains unchanged.This is called a random event .:-)

 

But the problem is not over yet. here we need to set up a random number with a very good quality and there is no association between two random numbers. in fact, the random number of standard C is difficult to achieve this (pseudo-random number generator ). pseudo-random number, which may be cracked and predicted. The real random number will not:-D.

Therefore, when online games perform random queries related to RMB, you can consider real random numbers or RNG with a better quality, libc rand/rand_r is enough ~~~~

PS:

I remember a plan. The probability of dropping a monster was set to be too low (20% or 25%), and then it was said that there were more than 40 monsters, and none of them fell .... later it was replaced by rand_r, and the effect was much better.

 

Refer:

Http://msdn.microsoft.com/en-us/library/aa379942 (v = vs.85). aspx