Minor bugs in granule generator

[GoogleCodeExporter]

1) There is inaccuracy in algorithm for small granules, when checking for 
granule intersection in the newly generated set. Sometimes intersecting 
granules are generated. However, the number of such granules is small compared 
to the (slightly) intersecting granules from different sets. The latter is due 
to the fact that intersection between the sets is tested through dipoles, and 
thus is inherently inaccurate. Since the latter is not considered a bug but as 
a designed feature (at least, it is very hard to correct this), the mentioned 
bug should not compromise any already obtained results.

2) When granule size is larger than the minimum particle dimension, only the 
warning is produced. But then another error (like integer overflow) happens due 
to appearing negative numbers. So this warning should be changed to error.

3) When granule size is comparable to the particle size, using volume 
correction can silently adjust the granule size by several times. Usually, this 
is not what meant by the user. So a warning should be produced, if the 
adjustment is more than, say, 10%.

Original issue reported on code.google.com by yurkin on 3 Sep 2010 at 2:31

• Blocking: #112, #112

[GoogleCodeExporter]

Fixed by revision r959.

Original comment by yurkin on 5 Sep 2010 at 6:31

• Changed state: Fixed

[GoogleCodeExporter]

Original comment by yurkin on 3 Aug 2014 at 4:58

• Now blocking: #112, #112