8
$\begingroup$

I have contacted support, CASE:4075709 UPDATE I have received reply from wolfram support. They have confirmed that this is indeed a bug. There are lots of sets of data that can cause the Kernel to crash in this situation.

I was trying to solve a complex problem, in which I have to take some random points and calculate the minimum distance for the tour around the points:

SeedRandom[100];
FindShortestTour[RandomReal[100, {35, 2}]]

This works fine if I run it for only a few times, however, if I use something that calculates this for more than a few times, like the following, I run into trouble.

RepeatedTiming[FindShortestTour[RandomReal[100, {35, 2}]], 3]

This would run fine for the first time or so, but running it for the second time without restarting the Kernel would cause the Kernel to crash, turning every In[] and Out[] into In[*] and Out[*], erasing every variable assigned, and gives out no result.

Since this stuff crashes, there is no tracing available for my situation. But since FindShortestTour works for as many as 10000 points, this shouldn't happen for merely 35 points. Why does my kernel crash?

So I have tried it again and again. The CPU load and memory usage rises during calculation, but not more than 10%. And sometimes the program actually runs. I have been unable to find a way to solve this-reinstalling Mathematica didn't seem to help.

Opening the Kernel directly and running this code would cause the program to exit.

Update

My solution to this problem is to install an older version of Mathematica, 11.0 works fine on my computer. I suppose there is a better solution for this-as @Kuba has suggested, installing an older version is not really a good way, as it does not solve the problem, but to escape from it.

I am using Mathematica 11.3.0 on Windows 10.

$\endgroup$
  • 4
    $\begingroup$ I don't get any crashes in M10, M11.1 or M11.3 on OSX. What Mathematica version and OS are you using? Could you add a SeedRandom[..] call so that it is repeatable? $\endgroup$ – Carl Woll Jun 4 '18 at 2:50
  • 3
    $\begingroup$ @CarlWoll It crashes for almost every random I use. I'm using Mathematica 11.3 on Windows 10 $\endgroup$ – t-smart Jun 4 '18 at 2:53
  • 2
    $\begingroup$ It is the nature of crashes that whether they are actually triggered may depend on effectively random things, such as the memory contents. Just because many people can't trigger the crash does not mean that there isn't a serious bug present. $\endgroup$ – Szabolcs Jun 4 '18 at 8:12
  • $\begingroup$ Unfortunately I am still unable to reproduce the crash on MMA 11.2 Win7-64, even when using your SeedRandom value. You can report it to Wolfram Support (they may have more insight), but I am not sure that we can help you further. $\endgroup$ – MarcoB Jun 4 '18 at 21:13
  • 2
    $\begingroup$ @t-smart Once you reported the issue and you get a case number from Wolfram, would you consider updating your question and include a bug header like Kuba did here? Also, add the [bugs] tag. I think 3 confirmed crashes is enough. $\endgroup$ – halirutan Jun 5 '18 at 13:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.