If we tried to simulate what Mirko is doing, we would have exceeded both the time and memory limit because the words that Mirko reads out on the screen can consist of hundreds of thousands (or even millions) of characters, even for small ~K~. That's why we won't simulate Mirko's process, but count the number of letters A and B in the following way:

• initially, ~\text{number_of_A} = 1~ and ~\text{number_of_B} = 0~
• in one step we get one letter A for each letter B we've had so far, and one letter B for each letter A and each letter B we've had so far; therefore: ~\text{new_A} = \text{number_of_B}~ and ~\text{new_B} = \text{number_of_A} + \text{number_of_B}~