[From Bill Powers (2003.03.08.1057 MST)]
Rick Marken (2003.03.08.0830)--
>Attached is my spreadsheet version of Econ004c (cleverly names Econ004c.xls).
>Because I'm trying to keep all calculations within cells, the order of
>calculation may not be quite the same as that in Econ004c. Therefore, it
would
>be nice to test Econ004c.xls to see if it behaves the same as Econ004c.
I've been able to get Excel and run your program. I changed the
initializations (green entries) that differed from those in my Pascal
program, and when I ran the spreadsheet, it ended up with very nearly the
same values as my program did (I use double precision floating point --
maybe you use a lower precision. The differences are trivial.
Pascal Excel
Rp 1028.33 1028.23
Rw 985.84 985.84
Rk 985.82 985.25
I was fooled at first -- I thought "Hey. his program runs a lot faster than
mine." It took a while to realise that it wasn't finished when it stopped
-- I had to keep clicking on the Run button until the values stopped
changing. Is there a way to change the number of iterations that run before
it stops? I am a TOTAL neophyte when it comes to Excel.
It would also be nice if the program could be run to the end while the bar
graphs are showing, instead of having to go back to the main screen and
select dynamic display again. I see that during the run, first Rw goes
negative,then it goes back positive and Rp goes negative, and finally Rp
goes positive and everything reaches its reference level. The Pascal
version does exactly the same thing.
It looks to me as though you have reproduced the Pascal version perfectly.
I am very impressed with the display and the ease of altering parameters,
so much better than the Pascal version. I have almost finished a Delphi
version, and it, too is a lot easier to use because of those nice built-in
editing functions.
When borrowing is not allowed, my program ends up with the same values for
Rp and Rw, but Rk is 54+ units too large. Yours is 53+ units too large!
That's out of 1000. This could be the result of rounding errors or a
mistake. I've gone over the code and don't see the mistake. If there is one
it's probably something like subtracting the amount before instead of after
a test to see if it should be subtracted. Be nice to get it fixed if we
could, but we can live with it if we have to.
Check over the initialized values so yours are the same as mine, and I
think you'll find the sams things I did. If not we're in a space warp.
Nice work, Rick. I have more respect for the spreadsheet approach now.
Best,
Bill P.
I
···
presume a good test would be one that would drive reserves below zero. This
should not happen when no credit is allowed. Is there a set of reference
values for the various control systems that we could use to test to see
whether
the two programs produce the same results when credit is and is not allowed?
Dick Robertson (2003.03.06.1932CST) --
>Yup, it works! That is I see the lovely bar graphs, rising and falling.
Econ004c.xls makes nice rising and falling bar graphs, too. Could you give
it a
try and see if it works OK for you. You might need to adjust the zoom (to 75%,
say) to see all systems and graphs simultaneously on a PC).
Best regards
Rick
--
Richard S. Marken
MindReadings.com
marken@mindreadings.com
310 474-0313
