Comparing the values of two vectors

[utab]

Hi there,

I would like to compare the values in two vectors of double. I can do
it by using iterators, I had an idea but wondering there is a library
facility to do that.

vector<double> a;
vector<double> b;

and that is enough for me to check the first three or four values of
these two vectors. Maybe in the future I will need something from STL
if existent ) .

Thanks in advance.

 

[Rolf Magnus]

Hi there,

I would like to compare the values in two vectors of double. I can do
it by using iterators, I had an idea but wondering there is a library
facility to do that.

Depends on what exactly you mean by "compare the values in two vectors". Do
you want to compare each element of a with the corresponding element of b?

vector<double> a;
vector<double> b;

and that is enough for me to check the first three or four values of
these two vectors. Maybe in the future I will need something from STL
if existent ) .

You could use std::equal, which returns true if both sequences are equal or
false otherwise.

 

[Richard Herring]

Hi there,

I would like to compare the values in two vectors of double. I can do
it by using iterators, I had an idea but wondering there is a library
facility to do that.

vector<double> a;
vector<double> b;

Doesn't your library documentation describe the std::vector comparison
operators?

and that is enough for me to check the first three or four values of
these two vectors. Maybe in the future I will need something from STL
if existent ) .

Thanks in advance.

if (a==b) //...
if (a<b) // ...
etc.

 

[Gavin Deane]

if (a==b) //...

But bear this in mind
http://www.parashift.com/c++-faq-lite/newbie.html#faq-29.17

operator== is defined for std::vector, but it is not necessarily a good
way to compare floating point numbers.

Gavin Deane

 

[Richard Herring]

But bear this in mind
http://www.parashift.com/c++-faq-lite/newbie.html#faq-29.17

operator== is defined for std::vector, but it is not necessarily a good
way to compare floating point numbers.

True. OTOH it's often exactly what's required. We can't tell without
knowing more about the OP's requirements..

 

[Robbie Hatley]

, regarding equality-checking doubles:

True [that it's a bad idea]. OTOH it's often exactly what's
required.

I've been burned by that before. Just what does "equal" mean
with doubles? "Equal" to +- 10^-5? "Equal" to +- 10^-10?
Often two variables that should be "equal" will compare
as inequal due to tiny round-off errors in the ninth or tenth
decimal places.

Idea:

#include <cmath>
bool MostlyEqual(double A, double B, int Precision)
{
if (Precision< 1) Precision= 1;
if (Precision>10) Precision=10;
double Max = A;
if (B > A) {Max = B;}
double Divisor = pow(10.0, (double)Precision);
double Difference = A - B;
if (B > A) {Difference = B - A;}
return (Difference < Max / Divisor)
}

 

[Richard Herring]

, regarding equality-checking doubles:

True [that it's a bad idea].

No. True that it's a bad idea to do it without understanding how
floating-point works.

I've been burned by that before. Just what does "equal" mean
with doubles? "Equal" to +- 10^-5? "Equal" to +- 10^-10?

Equal as in "==". If that isn't what you mean, don't call it "equal",
and write a function that expresses what you _do_ mean, as you have done
below.

Often two variables that should be "equal" will compare
as inequal due to tiny round-off errors in the ninth or tenth
decimal places.

Yes, when they are the result of arbitrary computation. No, when they
result from some restricted set of operations. For example, you can
store quite large integers *exactly* in doubles, and expect to get exact
integer results from adding and subtracting them.