R
rajus
I want to store the (x,y) coordinates of about 10,000 points in a 2D
array.How can I store them and retrieve them later?
array.How can I store them and retrieve them later?
L
Lew Pitcher
rajus said:
I'm afraid that your question doesn't relate specifically to the use of
the C langage, and is off topic in this forum. You might try one of the
comp.programming groups where they discuss algorithms and data
structures.
But, before you go, why don't you think about a sparse matrix
implementation?
Something like
struct { long x; long y; } coord[10000];
HTH
J
jacob navia
You have 10 000 x 2 points i.e. 20 000 points to store.rajus said:
To minimize space you could use 12 bits (coordinates up
to 4096x4096) and store 5 points per 64 bits, making it
around 32 000 bytes for the 2D array.
Using only 10 bits (1024x1024 coordinates) you would store
6 points per 64 bits unit, around 26 666 bytes for the array.
You would need to develop a small package that reads any sequence of
10 or 12 bits in the array, making a function like
uint32 GetPointsCoords(int x,int y)
{
// Here you calculate the offset in the array
// of your point, and return a 32 bit result
// containing in the upper 16 bits the y coordinate
// and in the lower 16 bits the x coordinate
// for instance.
}
You can save yourself development time of course if you
just make
unsigned short array[10000][2];
This would take 40 000 bytes, an increase of 14 000
but no development effort, you just index the array:
x = array[2357][0];
y = array[2357][1];
With short in 16 bits, you would get enough coordinate space to go
to 65535 x 65535 screens. The largest screens today go to
2500 x 2000 so your software will last (for a while...)
jacob
A
Al Balmer
Of course, the extra code required to do this would probably make up
for the saved space. I like Lew's suggestion, using whatever size
integer is appropriate.
D
dcorbit
rajus said:
#include <stdlib.h>
#include <stdio.h>
typedef struct point {
double x;
double y;
} point, *pointAddr;
static size_t dim = 10000u;
int main(void)
{
size_t i;
/* The use of calloc to zero array of floats is not portable. */
/* Allocate array with malloc() */
pointAddr array = malloc(dim * sizeof( point) );
/* Check for success */
if (array == NULL) {
puts("FATAL ERROR: Out of memory.");
exit(EXIT_FAILURE);
}
/* Initialize array to a known state. */
for (i = 0; i < dim; i++) {
array.x = 0;
array.y = 0;
}
/* do whatever you like with the array here... */
return 0;
}
D
dcorbit
Lew said:
If this isn't a C question, then I don't know what one is.
With 10,000 points, a sparse array implementation is a total waste of
time and energy.
Even with doubles that's {typically} 16*10,000 = 160,000 bytes.
Considering 1 GB of PC3200 DDR2 ECC ram at $112:
http://www.ramseeker.com/pc/index.php
that's about 1/2 cent worth of memory. Hardly worth worrying about
sparse implementations.
IMO-YMMV.
D
dcorbit
rajus said:
Something a bit ambiguous about your post is:
"How can I store them and retrieve them later?"
You can obviously poke them into an array and pull them out when you
need them. The code I posted earlier is an example of how you might
prepare an array of 2d points for use.
But you might be talking about permanent storage. In such a case, the
C answer is to use fwrite() to save them to disk and fread() to read
them back into memory. This answer is not portable across different
systems because different systems have different binary formats. In
other words, if you write out binary data using fwrite() then you
should only read the data back into memory using fread() on machines
with similar architecture. So if you wrote the data out on a Windows
machine, it probably won't read in correctly on a SPARC machine.
Also, if you plan to have many sets of data (or need the binary
compatibility referred to above), you might want to store them in a
database.
We do not have enough information for a very clear answer if you want
to know about long term storage and retrieval.
D
dcorbit
Time to temper my remark. I keep forgetting that much of (most?) C
programming is actually embedded work. On an embedded system, you
might have very limited resources. In such a circumstance, a skiplist
might be a good way to store a sparse vector.
There are advanced codes available for sparse matrices and vectors, but
generally they are designed for huge data sets.
W
William Hughes
You got that one right!
- William Hughes
K
Keith Thompson
Suppose each (x,y) coordinate consists of a pair of integers, each of
which can be in the range 0..999999. If the OP wants to store
information about each point in the corresponding element of a 2D
array, the array would have to have one trillion (10**12) elements.
Some sort of sparse array representation is just about mandatory.
I think the OP wants to store information about a point, and use the
coordinates of the point to retrieve it later. I might consider using
a hash table, hashing the (x,y) coordinate pair to obtain the
retrieval key.
But we need more information from the OP. How are the (x,y)
coordinates represented (int, double, whatever)? What are the
possible ranges? What information to you need for each point? How
important is fast access to each point?
A
Andrey Tarasevich
rajus said:
Well, declare a 2D array of points. Store points in the array. Retireve them
later from the array. What exactly is the problem that makes you ask such a
trivial question?
J
jacob navia
Al said:
I do not think that a simple code for accessing a 2D 12 bit
integer array would make 14 000 bytes of code in C.
At most it would take 150-200 bytes of code, maybe more in RISC
processors, but never 12 000 bytes.
But that was a memory efficient solution, good for embedded
systems where memory considerations are important.
I proposed also a 16 bit solution, not knowing precisely the context.
What is important to emphasize is that integers of bit-size other
than 8-16-32 are possible, and can be a space saving in many situations.
I have seen proposals to add to C
int:12 a;
meaning a is a 12 bit integer. This, with support for array notation of
those integers would be interesting in systems where memory is important
jacob
A
Al Balmer
Trivial questions are not unwelcome here.
However, your question, "What exactly is the problem", is certainly
pertinent here, and the answer to that might make the original
question less trivial
