Languages like Java have Big Decimal libraries supporting basic arithmetic operations like additions, subtractions, multiplications and divisions. However, scientific problems usually requires mathematical functions like sin, cos, etc. In this problem, you’re to write a Big Decimal Calculator. There are 15 commands: • Function with two arguments: add, sub, mul, div, pow, atan2 • Functions with only one argument: exp, ln, sqrt, asin, acos, atan, sin, cos, tan For trigonometric functions, angles are always in radians. Input There will be at most 100 lines. Each line begins with the function name, followed by arguments, then the precision p (1 ≤ p ≤ 50). Each argument is formatted as one or more digits, followed by a dot ‘.’, then by one or more digits. The integer part cannot be omitted, but the last two parts can be omitted together. There can be an optional negative sign before an argument. Each input number contains at most 20 digits. In function ‘pow’, ‘exp’, ‘ln’ and ‘sqrt’, all the arguments are strictly positive; in function ‘asin’ and ‘acos’, the integer part of the arguments are always zero. Output For each line, print the answer, rounded to p decimal places (Don’t use scientific notation!). It is guaranteed that the result will be a finite number, and its integer part will not exceed 10 digits. Note: You may notice that this problem is not language-neutral. I mean, some programming languages have advantages over some others. This is intentional: real-world software development is like this. Choosing programming languages, libraries and the overall architectures can be vital. There are quite a lot of literatures on this topic (for example, Fast multiprecision evaluation of series of rational numbers by Bruno Haible , Thomas Papanikolaou), but that’s overkill for this problem. The time limit for this problem is rather large, and the test cases are quite gentle: the goal of this problem is to write a working program, not a perfect one, so try to write a concise code, which is usually faster to write and easier to debug. For a much more practical literature, look at this: http://www.tc.umn.edu/.ringx004/sidebar.html. That small article presents how to translate arguments to make the series expansion converge faster. Though you will not find the whole solution, you’ll have some nice ideas. Sample Input add 1.357 4.6279 10 sub 1.357 4.6279 10 mul 1.357 4.6279 10 div 1 103 30 pow 12.2 12.15 20 atan2 2.45 1.77 30 exp 10.98 50 ln 21.065 50
2/2 sqrt 2 40 asin 0.81 30 acos 0.47 35 atan 0.618 40 sin 3.1415 25 cos 2.0113 50 tan 1.78987 30 Sample Output 5.9849000000 -3.2709000000 6.2800603000 0.009708737864077669902912621359 15822384813181.61872382001683484036 0.945162277467215967394902628052 58688.55427461755601946329091442988532551237342326651423 3.04761289543097985660178308429069456872534888053139 1.4142135623730950488016887242096980785697 0.944152115154155950477697775653 1.08150554878078090500864808815790029 0.5535497640327316544572642343482671646331 0.0000926535896606714405662 -0.42639511018918176703311006536787403871085921161347 -4.491415179046604916096895094786