GMAT Crashing

[smdrake]

I have installed the windows version of GMAT onto my Windows7 system computer. I then proceeded to run through several tutorials using GUI and GMAT functioned as the tutorials said it would. When It came time to create my own example however GMAT would crash and the program would close. My colleague downloaded the Mac version of GMAT and he has been able to successfully run the same example script. There were occasions, however, where GMAT also crashed for him as well.

The Error message for Windows and Mac reads:
**GMAT.exe has stopped working** (Checking for a solution and will notify you when one is found)
Then the program proceeds to exit and the GMAT file is closed.

Is there a compatibility issue with GMAT for Windows and Macs?

Any information on this matter would be much appreciated.

[jjkparker]

Hi, thanks for trying GMAT!

We try to fix every crash we can, but some of these always fall through the cracks. Can you tell us what you did to cause the crash? Do you have a script that you were trying to run? If we can get that information, we can try to fix it in the code.

GMAT should be very stable on Windows, and should be usable (though maybe a bit less stable) on Mac. We're definitely interested in fixing any issues you find.

- Joel

[smdrake]

We have discovered that the scripts in which a plot was generated, GMAT crashes. However the scripts that do not require a plot do not crash GMAT.
We are using 64-bit Windows 7 computers.

[jjkparker]

Just to confirm, you're saying that removing the XYPlot from your script stops the crash?

- Joel

[smdrake]

Yes, for our Windows7 computers, removing the XYplot and the other plots stops the crash.

[DJCinSB]

Actually, in GMAT the XY Plot and the OpenGL ("OrbitView") plots are completely different code, so we need to know if the issue occurs with XY Plots and no OrbitView plots, or with OrbitView plots but no XY Plots, or only when you have both running at once.

On Linux, I do sometimes see a crash that sounds like the one you are describing, and (again, on Linux) that crash goes away when I remove the OrbitView plots, but keep the XY plots in place. If we can figure out which component is causing the crash for you, it'll help track it down. (So far, we haven't been able to reproduce it on any of our Windows 7 boxes, so we're still looking for a way to see it.)

[dev-wcs]

It would be helpful to know where it is crashing on the Mac, as I have seen some crashes with plotting before. When it crashes, you should be able to get information on the type of crash and where it crashed from the pop-up window (alternatively, run the Console app (inside /Applications/Utilities) to get that info).

Wendy

[teubert]

I'll let you know next time it crashes on my mac. It hasn't crashed for a long while though.

[JackCrenshaw]

I, too, have been having a lot of crashes on my XP machine. As a total newbie, I'm perfectly prepared to accept that the problem is pilot error, but I sure wish I could make it stop.

As near as I can tell, my problems all have to do with the DC solver. If I guess right on the parameters in the Vary block, I get good results. If not -- if the solver hits some conflict between trial values and the trial value parameters (min, max, step, etc.) I'm seeing one of two problems: Either GMAT seems to stick (as in an endless loop), or it simply stops working (program is not responding). I'm working the problems as best I can, but it's slow going.

Can you tell me the DC algorithm being used? I know that in one case, the solver was approaching the desired Achieve goal, just adding thruster burn time in max step quantities. But the first time it chose a value less than the max step value, it HUGELY overshot the trial value, yielding an orbit apogee 10-12 times larger than the target goal. I'm attaching a graph of the solver's convergence.

One possible contributing factor: I'm simulating a lunar mission (who ain't?), and the earth-based, elliptic orbit is very close to parabolic/hyperbolic. As you know, a hyperbola doesn't have an apogee, but depending on the implementation of your code that computes osculating Keplerian parameters, they might go wonky if we cross the parabolic threshold.

Maybe I should use C3 instead of apogee?

Speaking of which: Does your calculation of Keplerian elements use universal variables, or the classic two-body formulas?

[jjkparker]

Hi, Jack.

Well, GMAT should never outright crash, even if the DC gets stuck. This definitely sounds like something we should investigate. Would you mind attaching your script, so we can try to recreate the behavior?

If you'd prefer not to make it public, you can email it to me (see my profile) or gmat@gsfc.nasa.gov.

Joel