#8 Strange lattice gas behavior

[Silvio Capobianco]

Hi All,
it's Silvio here.

The experiments:
hpp.py
hpp_pca.py
hpp_rotated.py
crashed after I pressed the Enter key with the following error:

COMPILING THE SCAN ...........................
Fatal Python error: (pygame parachute) Segmentation Fault
Aborted

However, this only happened the first time I launched them.
Then I ran diffusion.py --- which went fine.
I had expected a crash from fhp.py as well. However, it ran fine; and the others also did.

So I did a test by cleaning the .simpcache folder and launching hpp.py again.
Outcome: first launch crash, second launch all fine.
Repeated the test with hpp_pca.
This time, three failed attempts were followed by a success.
hpp_rotated and fhp worked just fine from the start.

It seems that some of the scans randomly fail to compile.
I will update this report as I find out the possible causes.

[Silvio Capobianco]

I have been testing other experiments from the Examples folder:
difference.py
greenberg_hastings.py
parity.py
polymer.py
All of these have this bizarre behavior: first execution crashes, later work.
Others that immediately work fine are:
diffusion.py
ising.py
life.py
stochastic_greenberg_hastings.py

[Ted Bach]

Humm. What version is this?

If it's from the rc-8 branch on 64 bit linux and you just upgraded, then perhaps clearing out the simpcache is the right thing.

Speaking of which, we should make that happen automatically after an upgrade.

[Ted Bach]

Never mind, I see now you said that you've cleared the simpcache.

[Silvio Capobianco]

Yeah, the problem spawns when simpcache is clean---or at least, does not contain "enough info" for the experiment.
By the way: the error only happens on the 64-bit machines. My 32-bit laptop runs all the experiment fine from the beginning (except some that seem to be broken at least since 0.7).