Segmentation fault when using minimize_blockmodel_dl for community detection
Hi Tiago: I got Segmentation fault when using minimize_blockmodel_dl to discover communities from a graph. The following is the output in the terminal: ------------------- *** Error in `*** Error in `python*** Error in `python': double free or corruption (out): 0x00007fe6a8000c30 *** *** Error in `python': double free or corruption (!prev): 0xpython': double free or corruption (!prev): 0x0000000051abe1a0 *** Segmentation fault ------------------- This happens when I specify the parameter parallel = True for the method minimize_blockmodel_dl. I am wondering whether this segmentation fault has something to do with the synchronization between threads when specifying the parallel = True parameter. The graph-tool I am using is 2.7 (commit ebd75c28, Thu Sep 17 23:25:50 2015 +0200). Is this fixed in the latest version? Thanks. Best Regards~ Hao -- View this message in context: http://main-discussion-list-for-the-graph-tool-project.982480.n3.nabble.com/... Sent from the Main discussion list for the graph-tool project mailing list archive at Nabble.com.
Hi Tiago: I update the graph-tool to the latest stable version ('2.12 (commit 1d6bf978, Fri Nov 6 15:54:18 2015 +0100)'). However, when I try to use minimize_blockmodel_dl to discover communities in parallel, I still get segmentation fault as shown below. --------------------------- *** Error in `python': double free or corruption (fasttop): 0x000000000419d360 *** Segmentation fault -------------------------- The OS is Ubuntu 14.04 trusty, and the python version is 2.7.6. Thanks. Best Regards~ Hao ------------------------------------------------------------------------------- Hao Wu Discovery Analytics Center, Virginia Tech Bradley Department of Electrical and Computer Engineering Virginia Tech, Arlington, VA 22203, USA ------------------------------------------------------------------------------- On Mon, Feb 22, 2016 at 12:03 PM, Tiago de Paula Peixoto <tiago@skewed.de> wrote:
On 18.02.2016 18:31, hao wrote:
The graph-tool I am using is 2.7 (commit ebd75c28, Thu Sep 17 23:25:50 2015 +0200). Is this fixed in the latest version?
Yes. You should always try the newest release first.
Best, Tiago
-- Tiago de Paula Peixoto <tiago@skewed.de>
_______________________________________________ graph-tool mailing list graph-tool@skewed.de http://lists.skewed.de/mailman/listinfo/graph-tool
On 25.02.2016 16:18, Hao WU wrote:
Hi Tiago:
I update the graph-tool to the latest stable version ('2.12 (commit 1d6bf978, Fri Nov 6 15:54:18 2015 +0100)'). However, when I try to use minimize_blockmodel_dl to discover communities in parallel, I still get segmentation fault as shown below. ---------------------------
*** Error in `python': double free or corruption (fasttop): 0x000000000419d360 ***
Segmentation fault
Can you give a complete example of how you are calling the function? Best, Tiago -- Tiago de Paula Peixoto <tiago@skewed.de>
Hi Tiago: For example: ---------------- import graph_tool.all as gt overlap_blocks = gt.minimize_blockmodel_dl(g, overlap = True, parallel = True) ---------------- where g is a instance of gt.Graph. When I remove the parameter parallel = True (since default value is False), everything works well. However, setting parallel = True generates the segmentation fault like I mentioned in my last email. Thanks. Best Regards~ Hao ------------------------------------------------------------------------------- Hao Wu Discovery Analytics Center, Virginia Tech Bradley Department of Electrical and Computer Engineering Virginia Tech, Arlington, VA 22203, USA ------------------------------------------------------------------------------- On Thu, Feb 25, 2016 at 10:58 AM, Tiago de Paula Peixoto <tiago@skewed.de> wrote:
On 25.02.2016 16:18, Hao WU wrote:
Hi Tiago:
I update the graph-tool to the latest stable version ('2.12 (commit 1d6bf978, Fri Nov 6 15:54:18 2015 +0100)'). However, when I try to use minimize_blockmodel_dl to discover communities in parallel, I still get segmentation fault as shown below. ---------------------------
*** Error in `python': double free or corruption (fasttop): 0x000000000419d360 ***
Segmentation fault
Can you give a complete example of how you are calling the function?
Best, Tiago
-- Tiago de Paula Peixoto <tiago@skewed.de> _______________________________________________ graph-tool mailing list graph-tool@skewed.de http://lists.skewed.de/mailman/listinfo/graph-tool
On 25.02.2016 17:31, Hao WU wrote:
Hi Tiago:
For example:
----------------
import graph_tool.all as gt
overlap_blocks = gt.minimize_blockmodel_dl(g, overlap = True, parallel = True)
----------------
where g is a instance of gt.Graph. When I remove the parameter parallel = True (since default value is False), everything works well. However, setting parallel = True generates the segmentation fault like I mentioned in my last email. Thanks.
Do you observe the same with overlap=False? -- Tiago de Paula Peixoto <tiago@skewed.de>
overlap = False seems to work well. No segmentation fault issue. ------------------------------------------------------------------------------- Hao Wu Discovery Analytics Center, Virginia Tech Bradley Department of Electrical and Computer Engineering Virginia Tech, Arlington, VA 22203, USA ------------------------------------------------------------------------------- On Thu, Feb 25, 2016 at 11:33 AM, Tiago de Paula Peixoto <tiago@skewed.de> wrote:
On 25.02.2016 17:31, Hao WU wrote:
Hi Tiago:
For example:
----------------
import graph_tool.all as gt
overlap_blocks = gt.minimize_blockmodel_dl(g, overlap = True, parallel = True)
----------------
where g is a instance of gt.Graph. When I remove the parameter parallel = True (since default value is False), everything works well. However, setting parallel = True generates the segmentation fault like I mentioned in my last email. Thanks.
Do you observe the same with overlap=False?
-- Tiago de Paula Peixoto <tiago@skewed.de> _______________________________________________ graph-tool mailing list graph-tool@skewed.de http://lists.skewed.de/mailman/listinfo/graph-tool
participants (3)
-
hao -
Hao WU -
Tiago de Paula Peixoto