Author |
Message |
Fred Sachs (sachs)
Member Username: sachs
Post Number: 10 Registered: 01-2004
| Posted on Sunday, July 11, 2004 - 04:20 pm: | |
When I output a history file using export, can I choose the name of the file and can I insert column headings? Using the versus command, can I label the relevant axis with a text title other than the variable name? sort of an "as" phrase? I would be really andy for parameter studies to be able to specify multiple stage variables. Scan parameter 1 over stage range i and parameter 2 over stage range j. then we can output data for a surface plot of the dependent variable vs. two parameters. this would really save time. I can scan two parameters by using the mod function to create sub stages, but that is clunky and requires resorting the data for plotting. |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 183 Registered: 06-2003
| Posted on Monday, July 12, 2004 - 02:46 pm: | |
1. You can choose the file name in a history export exactly as in other plots: EXPORT FILE="name". In the default output format, each data field is presented as a separate block, with a curly-bracketed comment giving the "As" name and the actual text of the computed expression. If you give a format string (q.v.), then stages are presented one per line, and you can label values as you please. There is no provision for column headers in this case. 2. Versus will label the abcissa with the text of the "versus" expression. There is no overriding label facility. 3. You can specify multiple staged parameters, but you must do the cycling manually. If you want two values each of a and b, specify four stages and give them values a = staged(a1,a1,a2,a2) b = staged(b1,b2,b1,b2) We will give some thought to how to generalize the staging facility. |
Fred Sachs (sachs)
Member Username: sachs
Post Number: 13 Registered: 01-2004
| Posted on Monday, July 12, 2004 - 02:54 pm: | |
Rather than having to cycle manually, I use the index i=stage #, and the second staging specificed by j=mod(i,nsub) where nsub is the number of points in the inner loop. The total number of stages would be nsub*n for the outerloop index. This saves having to come up with all the combinatorial possible parameters |
Fred Sachs (sachs)
Member Username: sachs
Post Number: 14 Registered: 01-2004
| Posted on Tuesday, July 13, 2004 - 02:52 pm: | |
I had thought that versus was going to use another staged variable for the x axis rather than basically using the stage number. That could be very convenient! |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 185 Registered: 06-2003
| Posted on Tuesday, July 13, 2004 - 04:54 pm: | |
I'm not sure what you mean. If you plot "history(data) versus thing", then the x-axis of the plot will be the value of thing, and labeled "thing". There will be one plot point per stage. The abcissa of the point will be the value of "thing" at that stage, and the ordinate of the point will be the corresponding value of "data". The trace will connect the points in the order of the stages. The only refinement I can see that could be made is to re-order the points by the value of "thing", so that the line doesn't retrace if "thing" is not monotonic across stages. |
|