declaration of size of vector<vector<int> >

[Erik Borgstr?m]

Hi,

Yes, I have followed some of the discussion on vector<vector<int> >,
but perhaps I'm just blind. I want to use a vector<vector<int> > of
outer size n and inner size m. Are these correct ways to achieve this?

1.
vector<vector<int> > v(n);
for(int i=0;i<n;i++){
v[i] = vector<int>(m);
}

2.
vector<int> smallvec(m);
vector<vector<int> > v(n, smallvec);

3.
vector<vector<int> > v(n, vector<int>(m));

Also, what's the preferred way?

thanks,
Erik

[Unforgiven]

Erik Borgstr?m wrote:

Hi,

Yes, I have followed some of the discussion on vector<vector<int> >,
but perhaps I'm just blind. I want to use a vector<vector<int> > of
outer size n and inner size m. Are these correct ways to achieve this?

1.
vector<vector<int> > v(n);
for(int i=0;i<n;i++){
v[i] = vector<int>(m);
}

2.
vector<int> smallvec(m);
vector<vector<int> > v(n, smallvec);

3.
vector<vector<int> > v(n, vector<int>(m));
As far as I can see, they're all correct.
Also, what's the preferred way?

I'd prefer the last one, as it avoids the need of creating any temporary
variable names, and is only one line.

I wouldn't use number 2 when defining a global variable, since you'd end up
with a global variable smallvec that never gets used after v is constructed.

--
Unforgiven

[Alberto Barbati]

Erik Borgstr?m wrote:

1.
vector<vector<int> > v(n);
for(int i=0;i<n;i++){
v[i] = vector<int>(m);
}

2.
vector<int> smallvec(m);
vector<vector<int> > v(n, smallvec);

3.
vector<vector<int> > v(n, vector<int>(m));
They are all correct, but they all copy n times a vector of size m
filled with zeros. This is a waste of time. Indeed, solution 1 is even
worse, as it wastefully creates and destroys n temporary vectors of size
m (solution 2 and 3 require only one temporary).
Also, what's the preferred way?

Dunno, but I would use this:

vector<vector<int> > v(n);
for(int i=0;i<n;i++)
v[i].resize(m);

Alberto

[Erik Borgstr?m]

Hi Alberto and unforgiven,

Thanks a lot for your advices. So I guess I'll use the 'resize' alternative :)

/erik

Alberto Barbati <Al************@libero.it> wrote in message news:<co********************@twister2.libero.it>.. .

Erik Borgstr?m wrote:

1.
vector<vector<int> > v(n);
for(int i=0;i<n;i++){
v[i] = vector<int>(m);
}

2.
vector<int> smallvec(m);
vector<vector<int> > v(n, smallvec);

3.
vector<vector<int> > v(n, vector<int>(m));

They are all correct, but they all copy n times a vector of size m
filled with zeros. This is a waste of time. Indeed, solution 1 is even
worse, as it wastefully creates and destroys n temporary vectors of size
m (solution 2 and 3 require only one temporary).

Also, what's the preferred way?

Dunno, but I would use this:

vector<vector<int> > v(n);
for(int i=0;i<n;i++)
v[i].resize(m);

Alberto