Use of monitor points for convergence check

Checking convergence by just looking at Picard residuals is not enough. The 
user should be able to select a number of monitor points and the variables to 
monitor on them. For example I may be able to select to monitor u and v at node 
10 and p at node 50. The values of these variables should be displayed on the 
screen as the code runs.

It is a better idea to use coordinate values instead of node numbers, in 
selecting monitor points. For example I may want to monitor u and v at 
coordinate (0.5,0.5,0.5). The code should first locate the element that 
encloses this point or maybe find the nearest node to this selected point and 
monitor the necessary variables.

The user will use the input file to enter the number of variable monitors 
(nMonitor), the variables to watch (monitorVars array of size nMonitor) and the 
coordinates of the monitor points (monitorCoord matrix of size nMonitor x 3).

Original issue reported on code.google.com by cuneytsert on 17 Sep 2012 at 8:22

tdlukulu / CUDA-CFD

Use of monitor points for convergence check #12