Related
Given a number, n, I need to efficiently find how many times this number is a multiple of all powers of 4 less than the given number.
For examples:
16 is a multiple of 4, and 16, so the result would be 2.
64 is a multiple of 4, 16, and 64, so the result would be 3.
256 is a multiple of 4, 16, 64, and 256, so the result would be 4.
14 is not a multiple of any power of 4, so the result would be 0.
35 is not a multiple of any power of 4, so the result would be 0.
Bitwise operations are preferred, and this is in a very tight loop so it is inside of a bottleneck that needs to be efficient. My code at the moment is the obvious answer, but I have to believe there is something more mathematical that can figure out the result in less steps:
power = 4;
while (power < n) {
result += !(n & (power - 1));
power *= 4;
}
You could use logarithms. A quick Google search for "fast log2 c++" brought up a pretty long list of ideas. Then your answer is log2(x)/2, and you'd have to find some way to make sure that your result is a whole number if you only want an answer for exact powers of 4.
If you are programming for an x86 processor, you can use BitScanForward & BitScanReverse to find the set bit, and use it to compute log2. The following code works in Visual Studio, for GCC or others, there are other ways to do inline assembly.
uint32_t exact_power_of_4_scan(uint32_t num)
{
unsigned long reverse;
unsigned long forward;
if (!_BitScanReverse(&reverse, num)) return 0;
_BitScanForward(&forward, num);
if (reverse != forward) return 0; // makes sure only a single bit is set
if (reverse & 0x1) return 0; // only want every other power of 2
return reverse / 2;
}
If you need a portable solution, table lookup might be the way to go, but is more complicated.
uint8_t not_single_bit[256] = {
1, 0, 0, 1, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1,
0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
0, 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, 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, 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, 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,
};
uint8_t log2_table[256] = {
0, 0, 1, 0, 2, 0, 0, 0, 3, 0, 0, 0, 0, 0, 0, 0,
4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
};
uint32_t exact_power_of_2(uint32_t num)
{
auto a = not_single_bit[num & 0xff];
auto b = not_single_bit[(num >> 8) & 0xff];
auto c = not_single_bit[(num >> 16) & 0xff];
auto d = not_single_bit[(num >> 24) & 0xff];
if (a + b + c + d != 3) {
return 0;
}
if (!a) {
return log2_table[num & 0xff];
}
if (!b) {
return log2_table[(num >> 8) & 0xff] + 8;
}
if (!c) {
return log2_table[(num >> 16) & 0xff] + 16;
}
return log2_table[(num >> 24) & 0xff] + 24;
}
uint32_t exact_power_of_4(uint32_t num)
{
auto ret = exact_power_of_2(num);
if (ret & 0x1) return 0;
return ret / 2;
}
Both are linear algorithms. The first will probably beat out looping for almost any value of num, but I haven't tested it. The second is probably only good for largish nums.
The mathematics would be to keep dividing by 4 until the result is no longer divisible by 4.
If you really want to do it with bitwise operations, techniques here can be used to count the number of trailing zero bits (i.e. the number of times a value is divisible by 2). Those can be adjusted to count pairs of trailing bits (i.e. divisibility by a power of 4 rather than 2).
Note that you will need to work with unsigned values to avoid certain cases of undefined or unspecified behaviours.
I would dispute your assertion that bitwise operations will make for a more efficient solution. It is not a given without testing, particularly with modern compilers.
I want to use pre-processor to fill some arrays in some ways. I can use preprocessor only for newly declared arrays. However, I need to change the array p which I declared and used before. Time optimization is very important for my purpose.
#define Reverse(x) {x[63], x[62], x[61], x[60], x[59], x[58], x[57], x[56], x[55], x[54], x[53], x[52], x[51], x[50], x[49], x[48], x[47], x[46], x[45], x[44], x[43], x[42], x[41], x[40], x[39], x[38], x[37], x[36], x[35], x[34], x[33], x[32], x[31], x[30], x[29], x[28], x[27], x[26], x[25], x[24], x[23], x[22], x[21], x[20], x[19], x[18], x[17], x[16], x[15], x[14], x[13], x[12], x[11], x[10], x[9], x[8], x[7], x[6], x[5], x[4], x[3], x[2], x[1], x[0] }
int main()
{
int p[64] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
int q[64] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
//doThings
p= Reverse(q); // line A - Gives error
int s[64] = Reverse(q); // line B - Works properly
//doThings
}
I got this error:
Error 11 error C3079: an initializer-list cannot be used as the right operand of this assignment operator c:\users\ferda\documents\visual studio 2013\projects\consoleapplication3\consoleapplication3\consoleapplication3.cpp 393 1 ConsoleApplication3
Built in array type does not allow to assign to it another array or use aggregate initialization on already initialized array, you would have to use memcpy or for loop to update it with new values. Your code will compile if you use std::array instead. It provides operator=:
operator= (implicitly declared) overwrites every element of the
array with the corresponding element of another array (public member
function)
http://coliru.stacked-crooked.com/a/8e664210b7f7f73b
i am not sure if this will work as fast as you expect, gcc will generate lots of mov instructions : https://godbolt.org/g/tzUqC3. I suppose it might be faster to use a for loop which will require less cache memory. As always profile your code.
I'd like to know if I translated a piece of code correctly from C++ to Delphi.
It looks like it is working, but I have a feeling that I'm reading and writing into memory that I'm not supposed to using Delphi.
Given C++ code:
struct tile_map
{
int32 CountX;
int32 CountY;
uint32 *Tiles;
};
inline uint32
GetTileValueUnchecked(tile_map *TileMap, int32 TileX, int32 TileY)
{
uint32 TileMapValue = TileMap->Tiles[TileY*TileMap->CountX + TileX];
return(TileMapValue);
}
uint32 Tiles00[9][17] =
{
{1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1},
{1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1},
{1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1},
{1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1},
{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1},
{1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1},
{1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1},
{1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1},
{1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1},
};
// More tile map declarations ...
// uint32 Tiles01[9][17] = ...
// uint32 Tiles10[9][17] = ...
// uint32 Tiles11[9][17] = ...
tile_map TileMaps[2][2];
TileMaps[0][0].CountX = 17;
TileMaps[0][0].CountY = 9;
TileMaps[0][0].Tiles = (uint32 *)Tiles00;
TileMaps[0][1] = TileMaps[0][0];
TileMaps[0][1].Tiles = (uint32 *)Tiles01;
TileMaps[1][0] = TileMaps[0][0];
TileMaps[1][0].Tiles = (uint32 *)Tiles10;
TileMaps[1][1] = TileMaps[0][0];
TileMaps[1][1].Tiles = (uint32 *)Tiles11;
// Usage
int32 PlayerTileX = 2;
int32 PlayerTileY = 2;
uint32 TileMapValue = GetTileValueUnchecked(&TileMap[1][1], PlayerTileX, PlayerTileY);
Delphi translation:
program Project1;
{$APPTYPE CONSOLE}
type
Puint32 = ^uint32;
tile_map = record
CountX : int32;
CountY : int32;
Tiles : Puint32;
end;
Ptile_map = ^tile_map;
{$POINTERMATH ON}
function GetTileValueUnchecked(TileMap : Ptile_map; TileX, TileY : int32) : uint32; inline;
begin
result := TileMap^.Tiles[TileY * TileMap^.CountX + TileX];
end;
const //in the future these will be read from file, so const for now
Tiles00: array [0..8, 0..16] of uint32 =
(
(1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1),
(1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1),
(1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1),
(1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1),
(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1),
(1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1),
(1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1),
(1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1),
(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1)
);
// More tile map declarations ...
//Tiles01: array [0..8, 0..16] of uint32 = ...
//Tiles10: array [0..8, 0..16] of uint32 = ...
//Tiles11: array [0..8, 0..16] of uint32 = ...
var
TileMaps : array [0..1, 0..1] of tile_map;
PlayerTileX, PlayerTileY : int32;
TileMapValue : uint32;
begin
TileMaps[0][0].CountX := 17;
TileMaps[0][0].CountY := 9;
TileMaps[0][0].Tiles := Addr(Tiles00);
TileMaps[0][1] := TileMaps[0][0];
TileMaps[0][1].Tiles := Addr(Tiles01);
TileMaps[1][0] := TileMaps[0][0];
TileMaps[1][0].Tiles := Addr(Tiles10);
TileMaps[1][1] := TileMaps[0][0];
TileMaps[1][1].Tiles := Addr(Tiles11);
// Usage
PlayerTileX := 2;
PlayerTileY := 2;
TileMapValue = GetTileValueUnchecked(#TileMaps[1][1], PlayerTileX, PlayerTileY);
end.
David Heffernan's comments has been helpful and others seem to agree that the code is correct, so I will mark this as answered.
I have a .cpp file that must include Console.h. In the file I'm trying to create a map (used later on for a game).
Error C2086: 'int nMapArray[15][20]: redefinition
#include "Console.h"
#include <Windows.h>
#include <stdint.h>
// Map dimensions
#define MAP_WIDTH 20
#define MAP_HEIGHT 15
// Tile Types
#define TILE_FLOOR 0
#define TILE_WALL 1
// Map declaration
int nMapArray[ MAP_HEIGHT ][ MAP_WIDTH ];
// Map Layout
int nMapArray[ MAP_HEIGHT ][ MAP_WIDTH ]=
{
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 1, 0, 1, 1, 0, 0, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
};
I know that I'm only supposed to declare nMapArray once but I'm not sure which one to discard. If I discard int nMapArray[ MAP_HEIGHT ][ MAP_WIDTH ]; then it will generate two errors:
LNK2019: unresolved external symbol
and
LNK1120: unresolved externals
Did a bit of googling but I still can't find the fix so help would be appreciated.
EDIT:
Ok so following the advice of many to get rid of the first inisialisation. From here I get two errors:
error LNK2019: unresolved external symbol "public: virtual class IConsole & __thiscall Win32Console::Color(unsigned short)" (?Color#Win32Console##UAEAAVIConsole##G#Z) referenced in function "void __cdecl DrawTile(int,int)" (?DrawTile##YAXHH#Z)
and
error LNK1120: 1 unresolved externals
Full code:
#include "Console.h"
#include <Windows.h>
#include <stdint.h>
// Map dimensions
#define MAP_WIDTH 20
#define MAP_HEIGHT 15
// Tile Types
#define TILE_FLOOR 0
#define TILE_WALL 1
// Map Layout
int nMapArray[ MAP_HEIGHT ][ MAP_WIDTH ]=
{
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 1, 0, 1, 1, 0, 0, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
};
void DrawMap( void );
bool IsPassable( int x, int y );
void DrawTile( int x, int y );
int main( void )
{
console.SetTitle( "Article Two Demo" );
// Declare the player's position
int nPlayerX = 4;
int nPlayerY = 4;
// Main game loop
while( true )
{
// Draw the map
DrawMap();
// Draw the player to the screen
console.Color( RED );
console.Position( nPlayerX, nPlayerY );
console << '#';
// Input phase - Wait for the player to do something
KEYPRESS sKeyPress = console.WaitForKeypress();
// Process the input
switch( sKeyPress.eCode )
{
// Move up
case CONSOLE_KEY_UP:
// Can we move to the tile above?
if( IsPassable(nPlayerX, nPlayerY-1) )
{
// Move up
nPlayerY--;
}
break;
// Move left
case CONSOLE_KEY_LEFT:
// Can we move to the tile to the left of the player?
if( IsPassable(nPlayerX-1, nPlayerY) )
{
// Move left
nPlayerX--;
}
break;
// Move right
case CONSOLE_KEY_RIGHT:
// Can we move to the tile to the right of the player
if( IsPassable(nPlayerX+1, nPlayerY ) )
{
// Move right
nPlayerX++;
}
break;
// Move down
case CONSOLE_KEY_DOWN:
// Can we move to the tile below the player?
if( IsPassable(nPlayerX, nPlayerY+1) )
{
// Move down
nPlayerY++;
}
break;
// Escape key
case CONSOLE_KEY_ESCAPE:
// Quit the program
return 0;
// Ignore any other keys
default:
break;
}
}
// If execution gets here, the program is done
return 0;
}
// IsPassable Function ///////////////////////////////////////////////////////////////////
//
// This function analyzes the coordinates of the map array specified and returns
// true if the coordinate is passable (able for the player to occupy), false if not.
//
bool IsPassable( int x, int y )
{
// Before we do anything, make darn sure that the coordinates are valid
if( x < 0 || x >= MAP_WIDTH || y < 0 || y >= MAP_HEIGHT )
return false;
// Store the value of the tile specified
int nTileValue = nMapArray[y][x];
// Return true if it's passable
if( nTileValue == TILE_FLOOR)
return true;
return false;
}
// DrawMap Function //////////////////////////////////////////////////////////////////////
//
// This function draws the entire map to the screen.
//
void DrawMap( void )
{
for( int y = 0; y < MAP_HEIGHT; y++ )
{
for( int x = 0; x < MAP_WIDTH; x++ )
{
DrawTile(x, y);
}
}
}
// DrawTile Function /////////////////////////////////////////////////////////////////////
//
// Draws a map tile for the map coordinates specified.
//
void DrawTile( int x, int y )
{
console.Position( x, y );
switch( nMapArray[y][x] )
{
case TILE_FLOOR:
console.Color( GREY );
console << '.';
break;
case TILE_WALL:
console.Color( GREY );
console << '#';
break;
}
}
//////////////////////////////////////////////////////////////////////////////////////////
In Console.h I haven't put the layout in (because I'm not entirely sure how to do that).
Get rid of the first one, the one without the initialisation.
That solves the compiler issue, the linker errors are another matter. The reason they appear is simply because the compilation phase is working once the double declaration is fixed.
Then I suggest you post another question with more details on the linker problems.
I use the following function to return a weight matrix of values between 0 and 1 depending on how close to the center a particular matrix position is. Also after a threshold the values are all 1 (it is like plateau with the points closer to center having value 1 and points away from center after a threshold of distance linearly fall from 1 to 0 at the edges)?
cv::Mat2f getWeightsMatrix(int N, int M, float r){
cv::Mat2f weights = cv::Mat2f(N,M);
int i,j;
for(i=0;i<N;i++){
for(j=0;j<M;j++){
if(i<=floor(N*(1-r)/2)){
if(j<=floor(M*(1-r)/2)){
weights[i][j]=((float)(i/N-j/M)/(1-r));
}
else{
weights[i][j]=(2*(float)(i/N)/(1-r));
}
}
else if (i>=floor(N*(1+r)/2)){
if(j>=floor(M*(1+r)/2)){
weights[i][j]=(((float)((N-i)/N))-((float)((M-j)/M)))/(1-r);
}
else{
weights[i][j]=(2*(float)((N-i)/N)/(1-r));
}
}
else{
if(j<=floor(M*(1-r)/2)){
weights[i][j]=(2*(float)(j/M)/(1-r));
}
else if(j>=floor(M*(1+r)/2)){
weights[i][j]=(2*(float)((M-j)/M)/(1-r));
}
else{
weights[i][j]=1;
}
}
}
}
cout << weights << endl;
return weights;
}
Now my problem is that I am having some casting issues and only values 0 and 1 are being returned (no floats). Also my matrix size displayed by the cout is 20x10 when I call the function with N=10, M=10 and r=0.5.
Please helP!
EDIT: This is the output
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0;
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0;
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0;
0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0;
0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0;
0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0;
0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0;
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0;
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0;
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
Your matrix consists only if integers because when you do basic operations on integers, the results gets rounded automatically.
For example: when you write
weights[i][j]=((float)(i/N-j/M)/(1-r));
The result of i/N is rounded to a integer, j/M is also rounded to an integer, and finally, the division by (1-r) is also rounded. Your (float) cast is a good idea but it's applied too late.
You can do several things:
Cast inside elementary operations, for instance, float(i)/N instead of i/N
Use float numbers instead of integers: write 1.0 instead of 1
Use floats inside your loops and statements
For example:
for(float i = 0; i < N-0.5; i++) {
for(float j = 0; j < M-0.5; j++) {
if(i <= floor(N*(1.0-r)/2.0)) {
// ...
It's important you understand that because of floating point precision, a test such as i < N might or might not pass when float i = N. This is why I did a little trick by substracting 0.5 from your loop bounds N and M.