Non-reproducible graph

[l0b0]

The results in the graph seem to have a small amount of noise. In four runs I got three slightly different results:

❯ diff <(./result/bin/unsure "4~6 * 1~2") <(./result/bin/unsure "4~6 * 1~2")
6c6
<       11.1 | ▒▒▒▒▒▒
---
>       11.2 | ▒▒▒▒▒▒
14,15c14,15
<        8.5 | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
<        8.2 | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
---
>        8.5 | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
>        8.2 | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
❯ diff <(./result/bin/unsure "4~6 * 1~2") <(./result/bin/unsure "4~6 * 1~2")
❯ diff <(./result/bin/unsure "4~6 * 1~2") <(./result/bin/unsure "4~6 * 1~2")
❯ diff <(./result/bin/unsure "4~6 * 1~2") <(./result/bin/unsure "4~6 * 1~2")
6c6
<       11.2 | ▒▒▒▒▒▒
---
>       11.1 | ▒▒▒▒▒▒
23c23
<        5.6 | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
---
>        5.6 | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒

[marralesfios]

The website says that the computations use the Monte Carlo method, which depends on random sampling and is thus nondeterministic.