Message Boards Message Boards

GROUPS:

Mathematica 11.3 randomly quits kernel on Thinkpad T470s and Ubuntu 16.04

Posted 3 months ago
518 Views
|
5 Replies
|
2 Total Likes
|

I have already contacted the technical support, but not heard back yet. Since this problem is quite frustrating, and I am relying on Mathematica on a daily basis, I am posting this here in the hope to get a quicker response.

A few weeks ago I got a new laptop (Thinkpad T470s) and installed Ubuntu 18.04 together with Mathematica 11.3. Very quickly did I run into the issue that Mathematica quit the kernel at seemingly random stages during a "Simplify" or larger symbolic expressions, or simply larger symbolic computations (it has happened in the context of "Series", "Simplify" and in several applications of the xTensor package). I then installed Ubuntu 16.04, as this is officially supported by Wolfram, with the exact same issue (same for other ubuntu based distributions).

Some sample code (and I can provide many more examples) that causes Mathematica 11.3 to quit the kernel after a good 5 seconds:

var = {g11n[1] -> g11n1, g11n[2] -> g11n2, g12n[1, 1] -> g12n11, 
   g12n[2, 1] -> g12n21, g12n[2, 2] -> g12n22, g21n[1] -> g21n1, 
   g21n[2] -> g21n2, g22n[1, 1] -> g22n11, g22n[2, 1] -> g22n21, 
   g22n[2, 2] -> g22n22};
Hnew[Ps_] := 
  H0n[Ps] + 
    G/r (f10n[Ps] + L/r^2 Sum[S[i] f11n[i, Ps], {i, 1, 2}] + 
       1/r^2 Sum[S[i] S[j] f12n[i, j, Ps], {i, 1, 2}, {j, 1, i}]) + 
    G^2/r^2 (f20n[Ps] + L/r^2 Sum[S[i] f21n[i, Ps], {i, 1, 2}] + 
       1/r^2 Sum[
         S[i] S[j] f22n[i, j, Ps], {i, 1, 2}, {j, 1, i}]) //. {S[1] ->
      S1, S[2] -> S2};

reprule = {A -> 
     g10n + L/r^2 Sum[S[i] g11n[i], {i, 1, 2}] + 
      1/r^2 Sum[S[i] S[j] g12n[i, j], {i, 1, 2}, {j, 1, i}], 
    B -> g20n + L/r^2 Sum[S[i] g21n[i], {i, 1, 2}] + 
      1/r^2 Sum[S[i] S[j] g22n[i, j], {i, 1, 2}, {j, 1, i}], 
    S[1] -> S1, S[2] -> S2, var} // Flatten;
PSnew = g0 + G/r A + G^2/r^2 B //. reprule;
SeriesCoefficient[Hnew[PSnew], {G, 0, 2}]

Now, for obvious reasons, Mathematica shouldn't crash over this. Since this seems to be a problem on several different Ubuntu versions (of which most are officially supported by Wolfram), I suppose it is due to the new laptop, though, I really cannot think of what it might be (some different type of RAM that Mathematica can't handle for some reason?). So maybe you have an idea for how to attempt to solve the issue. Thanks!

5 Replies
Posted 3 months ago

I tried the code from your post on my MacBook Pro 2018 with Mathematica 11.3, and after running for a few seconds the kernel quit without producing a result. Looks like it's not your OS or machine - seems more likely that your code is tickling a bug in the MMA kernel.

Posted 3 months ago

Hm, I have many more different examples (mainly simplifications of longer symbolic expressions) that cause Mathematica to quit the kernel; so can it really be the same issue every time? Or, in other words, how likely is it that I run into two different issues back to back?

This was due to a bug in version 11.3, fixed in version 12. I would not be surprised if the other problematic examples referred to are also encountering the same issue. If some are small enough to post I can check them.

Posted 3 months ago

The following shows the same problem (Mathematica quits the kernel after a few seconds):

-(1/((c^2 + (-2 + r) r) (1 + r^2 (I iv + rv)^2) (r^2 + 
    c^2 Cos[k]^2)^2)) I E^(I m p + t wi - 
   I t wr) ((1/(
    1 + c^2 (iv - I rv)^2 Cos[k]^2))(r^2 + c^2 Cos[k]^2) (-1 + (2 r)/(
       r^2 + c^2 Cos[
          k]^2)) (-c m (r (-2 + r (c^2 + (-2 + r) r) (I iv + rv)^2) + 
          c^2 (c^2 + r^2) (I iv + rv)^2 Cos[k]^2) (I RaI[r] + 
          RaR[r]) (I SaI[k] + SaR[k]) + (I wi + 
          wr) (-r (r^3 + c^4 r (iv - I rv)^2 + 
             c^2 (2 + r + r^3 (iv - I rv)^2 + 2 r^2 (I iv + rv)^2)) + 
          c^2 (c^4 (I iv + rv)^2 + c^2 (-1 + r^2 (I iv + rv)^2) + 
             r (2 - r + 2 r^2 (I iv + rv)^2)) Cos[k]^2) (I RaI[r] + 
          RaR[r]) (I SaI[k] + SaR[k]) + 
       r (c^2 + (-2 + r) r) (c^2 + r^2) (I iv + rv) (-1 + 
          c^2 (I iv + rv)^2 Cos[k]^2) (I SaI[k] + 
          SaR[k]) (I Derivative[1][RaI][r] + Derivative[1][RaR][r]) - 
       c^2 (c^2 + (-2 + r) r) (I iv + rv) (1 + r^2 (I iv + rv)^2) Cos[
         k] (I RaI[r] + RaR[r]) Sin[
         k] (I Derivative[1][SaI][k] + Derivative[1][SaR][k])) + (
    1/(-1 + c^2 (I iv + rv)^2 Cos[k]^2))
    2 c r Sin[
      k]^2 (1/2 c r (c^2 + (-2 + r) r) (I iv + rv) (-2 + 
          c^2 (I iv + rv)^2 + c^2 (I iv + rv)^2 Cos[2 k]) (I SaI[k] + 
          SaR[k]) (I Derivative[1][RaI][r] + 
          Derivative[1][RaR][r]) - (I RaI[r] + 
          RaR[r]) ((-c r (-2 + 
                r (c^2 + (-2 + r) r) (I iv + rv)^2) (I wi + wr) + 
             c^2 m (1 + c^2 (I iv + rv)^2) Cot[k]^2 - 
             c^3 (I iv + rv)^2 Cos[
               k]^2 ((c^2 + r^2) (I wi + wr) + c m Cot[k]^2) + 
             m r (-2 + r + 2 r^2 (iv - I rv)^2 + c^2 r (I iv + rv)^2 +
                 r^3 (I iv + rv)^2) Csc[k]^2) (I SaI[k] + SaR[k]) + 
          c (c^2 + (-2 + r) r) (I iv + rv) (1 + 
             r^2 (I iv + rv)^2) Cot[k] (I Derivative[1][SaI][k] + Derivative[1][SaR][k]))));
ComplexExpand[% // Re] // Simplify

This second example does not seem to crash, at least not in the 12.0 kernel I ran it in. Gives a big result and takes several minutes (I was not paying close attention so I do not know how many).

Reply to this discussion
Community posts can be styled and formatted using the Markdown syntax.
Reply Preview
Attachments
Remove
or Discard

Group Abstract Group Abstract