C++ type with 200 digits after decimal point?

Question

I have a project for my school, and the goal of it is create simple "Virtual Machine" who execute operations.

To do that, we can store some numbers with types, they are: int8, int16, int32, float, double, and big decimal.

The only type i don't know is the "big decimal", it is describe with: "The maximum of digits to display after the decimal point is are float(7), double(15), bigdecimal(200)."

So big decimal type can have 200 digits after the decimal point.

Any type in C++ ca do that? If i have to deal with strings it's more complex for the operations...

Thanks you in advance !

PS: "• all (external) libraries, except the STL are explicity forbidden"

Probably it is a duplicate. Check [this](https://stackoverflow.com/questions/18060242/how-to-store-extremely-large-numbers) — Jónás Balázs, Jul 21 '17 at 15:05
you'll have to deal with strings & implement whatever arithmetic operations your professor asks you to handle. — Haleemur Ali, Jul 21 '17 at 15:05
"• all (external) libraries, except the STL are explicity forbidden" — Jean Walrave, Jul 21 '17 at 15:13
@JónásBalázs: Not a dupe, since OP is forbidden from using an existing library. — einpoklum, Jul 21 '17 at 15:20
@HaleemurAli: most BigDecimal implementations do not use strings to store such numbers. They use a binary representation, often using BigIntegers internally. — Rudy Velthuis, Jul 21 '17 at 23:00

score 3 · Accepted Answer · edited Sep 13 '18 at 08:05

3

There is no "big-decimal" type in C++; from your description of the problem, it seems you are subposed to implement your own type with very-high precision.

There is more than one way to do this. Here are some hints/potential directions:

If d₁ ... d_k are digits, then the k-digit number 0.d₁d₂...d_k is equal to 10^-k * d₁ d₂ ... d_k (i.e. think mantissa and exponent).
200 = lots. Don't try to manually fit your number in enough basic types to get 200 digits.
The standard library has several classes you could use. Still, you can hardly go wrong by basing your work on an std::vector. Of course, what exactly you want to put in that vector is a different question.

edited Sep 13 '18 at 08:05

phuclv

37,963
15
156
475

answered Jul 21 '17 at 15:28

einpoklum

118,144
57
340
684

Implementing a proper BigDecimal type is not trivial. There are many things to consider. Especially rounding is not so easy, but also things like multi-precision division, multiplication, etc. may be quite a daunting task. Probably more so than all the rest of the VM. – Rudy Velthuis Jul 21 '17 at 23:04
@RudyVelthuis: OP is not supposed to implement a general-purpose BigDecimal, just something at the school project level. – einpoklum Jul 21 '17 at 23:39
Even a simple implementation (that does not implement stuff like Karatsuba or Burnikel-Ziegler or other optimizations) is not so easy. Especially rounding and division, but also parsing and displaying won't be very easy, not even with something like BCD or strings. I personally find it a little too much for a school project. – Rudy Velthuis Jul 22 '17 at 01:31

C++ type with 200 digits after decimal point?

1 Answers1