array_2D = new ushort * [nx];
// Allocate each member of the "main" array
//
for (ii = 0; ii < nx; ii++)
array_2D[ii] = new ushort[ny];
// Allocate "main" array
array_3D = new ushort ** [numexp];
// Allocate each member of the "main" array
for(kk=0;kk<numexp;kk++)
array_3D[kk]= new ushort * [nx];
for(kk=0;kk<numexp;kk++)
for(ii=0;ii<nx;ii++)
array_3D[kk][ii]= new ushort[ny];
the values of numexp,nx and ny is obtained by user..
Is this the correct form for dynamic allocation for a 3d array....We know that the code is working for the 2D array...If this is not correct can anyone suggest a better method?
I think the simplest way to allocate and deal with a multidimensional array is to use one big 1d array (or better yet a std::vector) and provide an interface to index into correctly.
This is easiest to think about first in 2 dimensions. Consider a 2D array with "x" and "y" axis
x=0 1 2
y=0 a b c
1 d e f
2 g h i
We can represent this using a 1-d array, rearranged as follows:
y= 0 0 0 1 1 1 2 2 2
x= 0 1 2 0 1 2 0 1 2
array: a b c d e f g h i
So our 2d array is simply
unsigned int maxX = 0;
unsigned int maxY = 0;
std::cout << "Enter x and y dimensions":
std::cin << maxX << maxY
int array = new int[maxX*maxY];
// write to the location where x = 1, y = 2
int x = 1;
int y = 2;
array[y*maxX/*jump to correct row*/+x/*shift into correct column*/] = 0;
The most important thing is to wrap up the accessing into a neat interface so you only have to figure this out once
(In a similar way we can work with 3-d arrays
z = 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2
y = 0 0 0 1 1 1 2 2 2 0 0 0 1 1 1 0 0 0 1 1 1 2 2 2
x = 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2
array: a b c d e f g h i j k l m n o p q r s t u v w x
Once you figure out how to index into the array correctly and put this code in a common place, you don't have to deal with the nastiness of pointers to arrays of pointers to arrays of pointers. You'll only have to do one delete [] at the end.
Looks fine too me, so long an array of arr[numexp][nx][ny] is what you wanted.
A little tip: you can put the allocation of the third dimension into the loop of the second dimension, aka you allocate each 3rd dimension while the parent subarray gets allocated:
ushort*** array_3D = new ushort**[nx];
for(int i=0; i<nx; ++i){
array_3D[i] = new ushort*[ny];
for(int j=0; j<ny; ++j)
array_3D[i][j] = new ushort[nz];
}
And of course, the general hint: Do that with std::vectors to not have to deal with that nasty (de)allocation stuff. :)
#include <vector>
int main(){
using namespace std;
typedef unsigned short ushort;
typedef vector<ushort> usvec;
vector<vector<usvec> > my3DVector(numexp, vector<usvec>(nx, vector<ushort>(ny)));
// size of -- dimension 1 ^^^^^^ -- dimension 2 ^^ --- dimension 3 ^^
}
Related
I've been looking at this for a while and I have some ideas on what this code could be doing, but I'm not sure if I correctly understand what the syntax of the code does.
The code iterates through a 2D array of unsigned chars, its meant to fill the array with 0's unless the spot in the array represents; the bottom, or the sides. If that is the case fill the spot in the array with a 9 instead.
The part I'm confused about is the statement (pField[y*fieldWidth + x] =) I believe this is a conditional statement, I understand the logic after, my question is specifically about this conditional, how should it be interpreted using if statements if possible?, If its not a conditional statement, what kind of statement is it?
pField = new unsigned char[fieldWidth*fieldHeight]; // Create play field buffer
for (int x = 0; x < fieldWidth; x++) // Board Boundary
for (int y = 0; y < fieldHeight; y++)
pField[y*fieldWidth + x] = (x == 0 || x == fieldWidth - 1 || y == fieldHeight - 1) ? 9 : 0;
The code is using a 2-dimensional array that is allocated in memory as a 1-dimensional array. The expression y*fieldWidth + x is calculating a 1D array index from a pair of 2D indexes.
The array represents a rectangle. The code is assigning a 9 to the 1D array elements that represent the rectangle’s left, right, and bottom edges (but not the top edge), and a 0 in the 1D array elements representing the rest of the rectangle.
For example, a 5x5 rectangle would look like this:
9 0 0 0 9
9 0 0 0 9
9 0 0 0 9
9 0 0 0 9
9 9 9 9 9
The corresponding 1D array elements would look like this:
x = 0 1 2 3 4 | 0 1 2 3 4 | 0 1 2 3 4 | 0 1 2 3 4 | 0 1 2 3 4
y = 0 0 0 0 0 | 1 1 1 1 1 | 2 2 2 2 2 | 3 3 3 3 3 | 4 4 4 4 4
---------------------------------------------------------
pField[] = 9 0 0 0 9 9 0 0 0 9 9 0 0 0 9 9 0 0 0 9 9 9 9 9 9
The ternary ?: operator can be rewritten using an if statement like this:
int value;
if (x == 0 || x == fieldWidth - 1 || y == fieldHeight - 1)
value = 9;
else
value = 0;
pField[y*fieldWidth + x] = value;
This code allocates space for a 2D array into a 1D array and then sets each element in the array. Regarding specifically the code you question, on the left hand side of the array the logic translates from 2D indices into 1D index. On the right hand side of the assignment the value is calculated by concatenating boolean predicates into a resulting boolean value that is then selected between returning 0 or 9 via a ternary operator and is wrote to the unsigned char array element.
I would like to find a mapping f:X --> N, with multiple discrete natural variables X of varying dimension, where f produces a unique number between 0 to the multiplication of all dimensions. For example. Assume X = {a,b,c}, with dimensions |a| = 2, |b| = 3, |c| = 2. f should produce 0 to 12 (2*3*2).
a b c | f(X)
0 0 0 | 0
0 0 1 | 1
0 1 0 | 2
0 1 1 | 3
0 2 0 | 4
0 2 1 | 5
1 0 0 | 6
1 0 1 | 7
1 1 0 | 8
1 1 1 | 9
1 2 0 | 10
1 2 1 | 11
This is easy when all dimensions are equal. Assume binary for example:
f(a=1,b=0,c=1) = 1*2^2 + 0*2^1 + 1*2^0 = 5
Using this naively with varying dimensions we would get overlapping values:
f(a=0,b=1,c=1) = 0*2^2 + 1*3^1 + 1*2^2 = 4
f(a=1,b=0,c=0) = 1*2^2 + 0*3^1 + 0*2^2 = 4
A computationally fast function is preferred as I intend to use/implement it in C++. Any help is appreciated!
Ok, the most important part here is math and algorythmics. You have variable dimensions of size (from least order to most one) d0, d1, ... ,dn. A tuple (x0, x1, ... , xn) with xi < di will represent the following number: x0 + d0 * x1 + ... + d0 * d1 * ... * dn-1 * xn
In pseudo-code, I would write:
result = 0
loop for i=n to 0 step -1
result = result * d[i] + x[i]
To implement it in C++, my advice would be to create a class where the constructor would take the number of dimensions and the dimensions itself (or simply a vector<int> containing the dimensions), and a method that would accept an array or a vector of same size containing the values. Optionaly, you could control that no input value is greater than its dimension.
A possible C++ implementation could be:
class F {
vector<int> dims;
public:
F(vector<int> d) : dims(d) {}
int to_int(vector<int> x) {
if (x.size() != dims.size()) {
throw std::invalid_argument("Wrong size");
}
int result = 0;
for (int i = dims.size() - 1; i >= 0; i--) {
if (x[i] >= dims[i]) {
throw std::invalid_argument("Value >= dimension");
}
result = result * dims[i] + x[i];
}
return result;
}
};
I have been stuck with this problem for two days and I still can't get it right.
Basically, I have a 2D array with relations between certain numbers (in given range):
0 = the order doesn't matter
1 = the first number (number in left column) should be first
2 = the second number (number in upper row) should be first
So, I have some 2D array, for example this:
0 1 2 3 4 5 6
0 0 0 1 0 0 0 2
1 0 0 2 0 0 0 0
2 2 1 0 0 1 0 0
3 0 0 0 0 0 0 0
4 0 0 2 0 0 0 0
5 0 0 0 0 0 0 0
6 1 0 0 0 0 0 0
And my goal is to create a new array of given numbers (0 - 6) in such a way that it is following the rules from the 2D array (e.g. 0 is before 2 but it is after 6). I probably also have to check if such array exists and then create the array. And get something like this:
6 0 2 1 4 5
My Code
(It doesn't really matter, but I prefer c++)
So far I tried to start with ordered array 0123456 and then swap elements according to the table (but that obviously can't work). I also tried inserting the number in front of the other number according to the table, but it doesn't seem to work either.
// My code example
// I have:
// relArr[n][n] - array of relations
// resArr = {1, 2, ... , n} - result array
for (int i = 0; i < n; i++) {
for (int x = 0; x < n; x++) {
if (relArr[i][x] == 1) {
// Finding indexes of first (i) and second (x) number
int iI = 0;
int iX = 0;
while (resArr[iX] != x)
iX++;
while (resArr[iI] != i)
iI++;
// Placing the (i) before (x) and shifting array
int tmp, insert = iX+1;
if (iX < iI) {
tmp = resArr[iX];
resArr[iX] = resArr[iI];
while (insert < iI+1) {
int tt = resArr[insert];
resArr[insert] = tmp;
tmp = tt;
insert++;
}
}
} else if (relArr[i][x] == 2) {
int iI = 0;
int iX = 0;
while (resArr[iX] != x)
iX++;
while (resArr[iI] != i)
iI++;
int tmp, insert = iX-1;
if (iX > iI) {
tmp = resArr[iX];
resArr[iX] = resArr[iI];
while (insert > iI-1) {
int tt = resArr[insert];
resArr[insert] = tmp;
tmp = tt;
insert--;
}
}
}
}
}
I probably miss correct way how to check whether or not it is possible to create the array. Feel free to use vectors if you prefer them.
Thanks in advance for your help.
You seem to be re-ordering the output at the same time as you're reading the input. I think you should parse the input into a set of rules, process the rules a bit, then re-order the output at the end.
What are the constraints of the problem? If the input says that 0 goes before 1:
| 0 1
--+----
0 | 1
1 |
does it also guarantee that it will say that 1 comes after 0?
| 0 1
--+----
0 |
1 | 2
If so you can forget about the 2s and look only at the 1s:
| 0 1 2 3 4 5 6
--+--------------
0 | 1
1 |
2 | 1 1
3 |
4 |
5 |
6 | 1
From reading the input I would store a list of rules. I'd use std::vector<std::pair<int,int>> for this. It has the nice feature that yourPair.first comes before yourPair.second :)
0 before 2
2 before 1
2 before 4
6 before 0
You can discard any rules where the second value is never the first value of a different rule.
0 before 2
6 before 0
This list would then need to be sorted so that "... before x" and "x before ..." are guaranteed to be in that order.
6 before 0
0 before 2
Then move 6, 0, and 2 to the front of the list 0123456, giving you 6021345.
Does that help?
Thanks for the suggestion.
As suggested, only ones 1 are important in 2D array. I used them to create vector of directed edges and then I implemented Topological Sort. I decide to use this Topological Sorting Algorithm. It is basically Topological Sort, but it also checks for the cycle.
This successfully solved my problem.
A discussion along this line could be found in this question and also in here, but my case is slightly different, as I am dealing with a dynamically allocated memory.
also please note, memset does not quite work with double value.
Anyway, I am trying to use std::fill to fill a dynamically allocated 2D array --
#include <iostream>
#include <algorithm>
using std::cout ; using std::endl ;
using std::fill ;
int main()
{
double **data ;
int row = 10, col = 10 ;
data = new double*[row] ;
for(int i = 0 ; i < col ; i++)
data[i] = new double[col];
// fill(&(data[0][0]),
// &(data[0][0]) + (row * col * sizeof(double)), 1); // approach 1
// fill(&(data[0][0]), &(data[0][0]) + (row * col), 1); // approach 2
// fill(data, data + (row * col * sizeof(double)), 1); // approach 3
// fill(&(data[0][0]),
// &(data[0][0]) +
// ((row * sizeof(double*)) +
// (col * sizeof(double))), 1); // approach 4
for(int i = 0 ; i < row ; i++) {
for(int j = 0 ; j < col ; j++)
cout << data[i][j] << " " ;
cout << endl ;
}
for(int i = 0 ; i < row ; i++)
delete [] data[i] ;
delete [] data ;
}
Approach 1: What I understand, the approach 1 should be the correct code, I am starting from the beginning &(data[0][0]) and the end of the array should be located at &(data[0][0]) + (row * col * sizeof(double)), but when I run, I get this error, but the array has been filled --
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
*** Error in `./test': free(): invalid next size (fast): 0x0000000000da3070 ***
Aborted (core dumped)
Approrach 2: However, according to this post, the approach 2 is recommended, but I do not quite understand this code, since sizeof(double) is missing, and I am getting this output --
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0
*** Error in `./test': free(): invalid next size (fast): 0x0000000000bf5070 ***
Aborted (core dumped)
Approach 3: I am not sure why this does not compile, data and &(data[0][0]) should have the same meaning, right?
Approach 4: I am not sure if this is correct or not.
How do I do it ?
Does std::fill give any extra benefit over two nested loops?
Unlike a stack allocated 2D array, a dynamic 2D array is not guaranteed to be a contiguous range. It is however a contiguous range of pointers, but each pointer in the array may point to non-contiguous memory areas. In other words, the first element of data + i + 1 may not necessarily follow the last element of the array pointed by data + i. If you wonder why a stack-allocated 2D array is contiguous, it is because when you declare something like
double data[10][20];
then the compiler understands it as array of 10 (contiguous) elements, each element of type double[20]. The latter type is also an array, which guarantees contiguous elements, so the double[20] elements (i.e. 20 double one after the other) are stacked one after the other in the memory. double[10][20] is is strikingly different from double**.
That's why std::fill or std::memset gives you headaches, as they both assume a contiguous range. Therefore in your case a nested loop seems to be the simplest way of filling the array.
In general, it is much better to use a 1D array in which you "mimic" the 2D access, exactly for the reasons mentioned above: data locality. Data locality implies fewer cache missed and better overall performance.
Pointer arithmetic requires that a pointer be incremented only to the extent that the result still points at the same array (or one past the end).
You allocate each row as a separate array in your for loop:
for(int i = 0 ; i < col ; i++)
data[i] = new double[col]; // this creates a distinct array for each row
Since each row array you allocate is col elements, the maximum value that can legally be added to &(data[0][0]) is col. But in each of your examples of std::fill usage you add more to the pointer than you are allowed.
Given the way you are allocating the array, there's no way for you to pass raw pointers to a single call to std::fill in order to initialize the entire 2D array. Either you must use more than one call to std::fill (which defeats the purpose of using std::fill), or you must create an Iterator type that knows how to deal with the separate allocation of each row, or you must change the way you allocate the array.
I would recommend allocating the whole array at once as a single dimensional array and then writing some additional code to make it work like a two dimensional array. This has a number of benefits:
The standard library contains a convenient way of dynamically allocating a single dimensional arrays: std::vector
Using std::vector means you no longer need to use naked new and delete, which fixes the exception safety problem your code has.
A single allocation generally has better performance characteristics than many allocations (Of course, there are cases where separate allocations are better).
Here's a simple wrapper class to make a 1D array look like a 2D array:
class Matrix {
std::vector<double> data;
int cols;
public:
Matrix(int row, int col)
: data(row * col)
, cols(col)
{}
auto begin() { return data.begin(); }
auto end() { return data.end(); }
struct index_type { int row; int col; };
double &operator[](index_type i) {
return data[i.row * cols + i.col];
}
int row_count() const { return data.size()/cols; }
int column_count() const { return cols; }
};
Using this you can rewrite your code:
#include "Matrix.h"
#include <iostream>
#include <algorithm>
using std::cout ; using std::endl ;
using std::fill ;
int main()
{
Matrix data(10, 10);
fill(data.begin(), data.end(), 1.0);
for(int i = 0 ; i < data.row_count() ; i++) {
for(int j = 0 ; j < data.column_count() ; j++)
cout << data[{i, j}] << " " ;
cout << endl ;
}
}
Does std::fill give any extra benefit over two nested loops?
Using loops is less readable because loops could do lots of other things, and you have to spend more time figuring out what any particular loop is doing. For this reason one should always prefer using STL algorithms over manual loops, all else being equal.
// fill(&(data[0][0]),
// &(data[0][0]) + (row * col * sizeof(double)), 1); // approach 1
Pointer arithmetic automatically considers the size of the array elements. You don't need sizeof(double). Multiplying by sizeof(double) here is the same as multiplying by sizeof(double) inside []. You wouldn't do: data[i * sizeof(double)], so don't do data + (i * sizeof(double)).
Your example code uses &(data[0][0]). Think about if this the same or different from data[0]. Consider both the type and the value of the expressions.
I agree with the above comments. You have allocated 10 separate arrays so you can't initialize these with a single std::fill call.
Moreover, when you perform arithmetic operations on pointer of non-void types, a compiler automatically multiply your results by sizeof of a given type. However, when you use functions like memset or memcpy, you actually have to multiply number of elements by sizeof of a given types and pass it to one of these functions. It's because these function operate on bytes and they accept pointers of void type. Therefore it is impossible for compiler to take care of adjusting of sizes, because the void type has no specified size.
I'm trying to do the following in CUSP:
A=[
1,1,0,0;
2,2,2,0;
0,3,3,3;
0,0,4,4];
B=[1,1,1,1]';
disp(mldivide(A,B));
which is
X=[0.9167,0.0833,-0.5000,0.7500]
On the other hand I get a strange answer from CUSP
#include <cusp/dia_matrix.h>
#include <cusp/krylov/cg.h>
#include <cusp/print.h>
int main()
{
cusp::dia_matrix<int,float,cusp::host_memory> A(4,4,10,3);
A.diagonal_offsets[0] = -1;
A.diagonal_offsets[1] = 0;
A.diagonal_offsets[2] = 1;
for (int i = 0;i <3;i++)
{
for (int q = 0 ;q < A.num_cols;q++)
{
A.values(q,i)=q+1;
}
}
//copy
cusp::dia_matrix<int,float,cusp::device_memory> AA = A;
cusp::array1d<float,cusp::device_memory> BB(A.num_rows,1);
cusp::array1d<float,cusp::device_memory> XX(A.num_rows,0);
cusp::print(AA);
cusp::print(XX);
cusp::print(BB);
cusp::krylov::cg(AA,XX,BB);\
cusp::print(XX);
return 0;
}
The result looks like
sparse matrix <4, 4> with 10 entries
0 0 1
0 1 1
1 0 2
1 1 2
1 2 2
2 1 3
2 2 3
2 3 3
3 2 4
3 3 4
array1d <4>
0
0
0
0
array1d <4>
1
1
1
1
array1d <4>
-39.9938
-53.436
87.9025
-30.1429
The last one doesn't look quite right. Anybody know what I'm doing wrong? Am I using the code wrong or are we supposed to have a really good guessed solution + use a preconditioner?
The conjugate gradient method is only valid for use in symmetric positive definite matrices. Your matrix isn't symmetric. That is why it isn't (and cannot) producing a valid solution. Either use an appropriate, well conditioned SPD matrix, or use a different numerical method.