Closed heathergaya closed 1 year ago
Thanks for reporting. Just to manage expectations: how long does the example run until the error occurs? Would it also fail with fewer iterations? Do we need to fix the random seed to make the problem easier to reproduce? Meanwhile, I enjoy a bit of extra heat from my laptop 🙃
Could you please give me a rough idea where igraph is used in this setup?
Can you reproduce this with the latest version, 1.4.2?
I ran this code for 20 minutes using an ASan-instrumented build of R and igraph, and was unable to reproduce the problem.
nimble
uses few igraph functions. I don't have an exhaustive list, but the only non-trivial functions I could find so far were topological.sort()
and permute.vertices()
. The latter received some relevant bugfixes last year, but I believe these were already included in 1.3.5.
While the assertion failure clearly indicates an issue with igraph, tracking this down will be quite difficult without a small example. The most helpful thing to check would be whether you can reproduce the issue with 1.4.2 (or 1.4.1), @heathergaya, let us know how long it takes to reproduce the issue, and whether it reproduces consistently.
One possible explanation here is that your computer ran out of memory, and the failure you saw was due to an unchecked OOM condition. Please check how much memory this calculation is using relative to how much you have available.
I am closing this since we are unable to reproduce the issue. @heathergaya, feel free to re-open if you can consistently reproduce it with the latest igraph version.
This old thread has been automatically locked. If you think you have found something related to this, please open a new issue and link to this old issue if necessary.
I am running an MCMC in NIMBLE using the parallel package. This model works fine when run not in paralle. When using clusterEvalQ, I received the error:
Here is the code I was running (the three .txt files referenced are attached to this bug report):
nd.txt ni.txt nc.txt Version information