Designing a Transition Matrix

Each column (or row, pick convention) sums outgoing probabilities or counts.

Try this in Python

T = [[0.7, 0.2], [0.3, 0.8]]
print(sum(T[0]), sum(T[1]))

Key takeaways

• Markov stationary vector needs eigen solve—outside quick contests often.
• Nonlinear updates cannot be single matrix—don't force.
• Document row vs column vector multiply convention.