Using powershell (on windows) I generated regular expression problems in which the minimal elements in the loop varies.
I expected that the time for cvc4 to solve these problems would increase linear with the minimal elements in the loop.
However, I obtained
minimal elements in loop
time (ms)
0
36,3953
1
47,4381
2
48,905
5
63,6582
10
133,5371
20
319,4708
50
2284,578
100
11657,9906
200
74854,9273
This performance seems prohibitive for real-world problems where long string are quite common.
Is the performance fundamentally non-linear?
If not, could the performance be made linear?
Using powershell (on windows) I generated regular expression problems in which the minimal elements in the loop varies. I expected that the time for cvc4 to solve these problems would increase linear with the minimal elements in the loop. However, I obtained
This performance seems prohibitive for real-world problems where long string are quite common.
Is the performance fundamentally non-linear? If not, could the performance be made linear?
Details:
The ps1 script to generate Problems was
The output for this generation script with 10 as parameter is
The script to measure the time was (it includes a run-in measurement to incorporate caching effects)