|
|
|
|
|
|
|
|
|
Comparison between
default kernel 1 model and with dissolution |
|
Minimal |
|
|
n |
kernel 1 |
dissolution |
|
Arbitrary |
|
Additional arbitrary comparison |
12 |
56815 |
108941 |
|
13 |
154919 |
328310 |
|
n |
halt |
kernel 1 |
kernel 2 |
diss |
14 |
471162 |
1002596 |
|
12 |
32000 |
36050 |
38147 |
85000 |
15 |
|
3431545 |
|
|
|
56815 |
|
108941 |
|
|
For small
values... |
|
n=12..14 |
|
n |
Kernel 1. |
Dissolution. |
12 |
56815 |
108941 |
13 |
154919 |
328310 |
14 |
471162 |
1002596 |
|
|
|
|
|
|
But for bigger
ones... |
|
n=15.. |
|
n |
kernel 1 |
dissolution |
12 |
56815 |
108941 |
13 |
154919 |
328310 |
14 |
471162 |
1002596 |
15 |
36000000 |
3431545 |
|
Note:
The value for n=15 and kernel 1 model could not be obtained after 5 hours,
with the memory all the time at 100%. We include a value of 10 hours to
illustrate the trend |
|
|
Conclusion:
For small instances of n, the cost of the dissolution operation penalizes
the model with dissolution.
However, with bigger instances of this NP complete problem, other models
fill the whole memory with the exponential growth of space; the dissolution
option permits releasing useless non-solution membranes, making the problem
more treatable. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|