Solver does not find correct solution???

Hello,
I have recently had a problem with excel's solver (until last week I never
had a porblem). When fitting a data set (trying to minimize my target cell),
the solver says that either it has found a solution or that the data doesn't
converge. When I run the solver again, my target cell is further reduced or
the cells don't converge again. When running my spreadsheet on a colleagues
computer, there is no problem, so I know the problem isn't with my data.
Any help or suggestions would be greatly appreciated!
Thanks in advance.

experiment626 -

Check (and perhaps adjust) the settings (precision, tolerance, convergence,
etc.) on Tools | Solver | Options.

- Mike
www.mikemiddleton.com

"experiment626" <[email protected]> wrote in message
news:[email protected]...
> Hello,
> I have recently had a problem with excel's solver (until last week I never
> had a porblem). When fitting a data set (trying to minimize my target
> cell),
> the solver says that either it has found a solution or that the data
> doesn't
> converge. When I run the solver again, my target cell is further reduced
> or
> the cells don't converge again. When running my spreadsheet on a
> colleagues
> computer, there is no problem, so I know the problem isn't with my data.
> Any help or suggestions would be greatly appreciated!
> Thanks in advance.

H Mike, Thanks for your response, but I have already looked at that and I
have not changed any of my settings from when the solver used to work for me
with no problems. Any other suggestions?
Thanks in advance.

For non-linear non-convex problems, Solver, like most other
optimization packages, cannot guarantee a global optimum. It might
work depending on the options selected as well as the starting values
for a specific solution attempt.

As far as not changing any setting from when solver used to work, how
could you possibly do that? I used Solver a few times yesterday, and I
couldn't tell you what my option settings were! It is hard to imagine
someone being in the habit of noting or otherwise memorizing the
settings for max.time, max.iterations, precision, tolerance,
convergence, estimates, derivaties, search, etc.

On your colleague's computer and your own, open the workbook you are
having problems with. Compare the values of *all* the cells in the
model. Open the solver dialog box, click Options. Note the values of
*all* 12 settings.

Are there any differences between the two?

Now, what happens if you ask Solver to solve the problem?

--
Regards,

Tushar Mehta
www.tushar-mehta.com
Excel, PowerPoint, and VBA add-ins, tutorials
Custom MS Office productivity solutions

In article <[email protected]>,
[email protected] says...
> H Mike, Thanks for your response, but I have already looked at that and I
> have not changed any of my settings from when the solver used to work for me
> with no problems. Any other suggestions?
> Thanks in advance.
>

Hi Tushar Mehta,
Thanks for your response. I have really checked and rechecked the settings
many times, nothing has changed. I use the solver many times daily with
always the exact same model on different data. The settings are the same
EVERY time, they never change. To make things more frusterating, when my
collegue logs onto my computer (he has a login account on my computer) he can
run the solver without the problem I have when I am logged in. I understand
that the solver can not guarantee a golbal optimum, but usually it works for
several hours before finding a solution. My problem is that is now quits
after about 5 minutes and says that it has found a solution. It is not
possible for a solution to be found so quickly with my data. When i run the
solver again, it calculates for a few minutes and then quits again, I could
do that all day... As i said, I just started having this problem last week.
I have tried COMPLETELY uninstalling office, I have tried installing a
different office version, always with the same problem. Any other
suggestions? Thank you!!!!

> For non-linear non-convex problems, Solver, like most other
> optimization packages, cannot guarantee a global optimum. It might
> work depending on the options selected as well as the starting values
> for a specific solution attempt.
>

Tushar,

As a side note, the nonlinear nonconvex problems can now reach a proved
global optimum, thans to the new technology progress. Of particular
approach is the one based on convexification and branch-and-bound
methods. This guaranteed approach is also availabe in an Excel addin
solver, namely What's Best. If you are interested, you might want to
check out the product website at www.lindo.com. As I know, there is a
free version with limited capacity.

Regards,
Tony