modularize _is_communicating in cython backend

chainopy/_backend/_is_communicating.pyx

@@ -2,10 +2,9 @@ cimport numpy as cnp
 import numpy as np
 import cython
 
-
 @cython.wraparound(False)
 @cython.boundscheck(False)
-def is_communicating_cython(cnp.ndarray[cnp.float64_t, ndim=2] tpm, list states, str state1, str state2, int threshold):
+def _is_partially_communicating(cnp.ndarray[cnp.float64_t, ndim=2] tpm, list states, str state1, str state2, int threshold):
     cdef Py_ssize_t i, idx1, idx2
     cdef cnp.ndarray[cnp.float64_t, ndim=2] x = tpm
     cdef bint result
@@ -17,9 +16,18 @@ def is_communicating_cython(cnp.ndarray[cnp.float64_t, ndim=2] tpm, list states,
     idx2 = states.index(state2)
 
     for i in range(threshold):
-        if x[idx1, idx2] > 0 and x[idx2, idx1] > 0:
+        if x[idx1, idx2] > 0:
             return True
 
         x = x @ tpm
 
+    return False
+
+
+@cython.wraparound(False)
+@cython.boundscheck(False)
+def is_communicating_cython(cnp.ndarray[cnp.float64_t, ndim=2] tpm, list states, str state1, str state2, int threshold):
+    if (_is_partially_communicating(tpm, states, state1, state2, threshold) \
+        and _is_partially_communicating(tpm, states, state2, state1, threshold)):
+        return True
     return False

chainopy/markov_chain.py

@@ -296,7 +296,9 @@ def is_absorbing(self) -> bool:
 
         transient_states = set(self.states) - set(absorbing_states_)
         for i in absorbing_states_:
-            if all(self.is_communicating(state, i) for state in transient_states):
+            if all(_is_communicating._is_partially_communicating(state, i)
+                   for state in 
+                   transient_states):
                 return True
         return False