Context Based Predictive Information

. 2019 Jun 29;21(7):645. doi: 10.3390/e21070645

Algorithm 1 Context-based predictive information: pseudocode.

1:
Input: $x^{N}$ , $y^{N}$ , c, $ϵ$
2:
Implement on $x^{N}$ , $y^{N}$ the first stage of the I/O CT algorithm to obtain a full I/O context tree
3:
fori in 1 to $| c | - 1$ do
4:
Implement the following stages of the I/O CT algorithm
5:
with $c_{i - 1}$ , $c_{i}$ , $c_{i + 1}$ , and obtain $S_{c_{i - 1}}$ , $S_{c_{i}}$ , $S_{c_{i + 1}}$
6:
Calculate ${\hat{P I}}_{c_{i - 1}}, {\hat{P I}}_{c_{i}}, {\hat{P I}}_{c_{i + 1}}$
7:
if | $S_{c_{i - 1}}$ | = | $S_{c_{i + 1}}$ | then
8:
$d v 2 \leftarrow 0$
9:
else
10:
$d v 2 \leftarrow | \frac{\partial^{2} {\hat{P I}}_{c_{i}}}{\partial | S_{c_{i}} |^{2}} |$
11:
end if
12:
if $d v 2 > ϵ$ then
13:
return $c_{i}$ , ${\hat{P I}}_{c_{i}}$
14:
end if
15:
end for
16:
return $c_{| c |}, 0$