there are infinite ways of doing it

for example you could draw directly the needed pixels
Code:

loadPixels();
for(int i=start; i<end; i++){
   for(int j=start1; j<end1; j++){
	pixels[j*width+i]=mycolor;
   }
}
updatePixels(); 

or write a custom block class, which is probably the best way iin most scenarios.

Although he mentioned the idea of drawing pixels for your graph. I would definetly go with the rect... should be much easier.
Maybe you can post an example pic or something...

i dont get why you want to assign more than one coordinate to a point?
That doesnt make a lot of sense to me. How should the result look like?

i made up a quick sketch, hope that helps to understand

int gridSize = 20;

void setup() {
 size(400,600);
 background(255);
 smooth();
 noFill();
 stroke(120,100);

 //the grid
 for (int i = 0; i <width; i+=gridSize) {
   for (int j = 0; j < height; j+=gridSize) {
     rect(i,j,gridSize,gridSize);
   }
 }
}  


void draw(){

//drawing the bars with a given position, value and color
bar(2,5,color(#ff0000));
bar(4,3,color(#ffff00));
bar(6,8,color(#0000ff));


}


void bar(int pos, int value, color c){
   for (int i = 0; i <value; i++) {
   fill(c);
   rect(gridSize*pos, (height-gridSize)-(gridSize*i),gridSize,gridSize);
 }
}