We need the ADMM algorithm to be tested to make sure that it works and is numerically stable in each scenario. Here are a few scenarios that are worth being considered:
easy inputs (say, current_counts with two segments, short lengths (e.g., including less than 10 signals), and low variation) and difficult inputs (say, current_counts with more than 5 segments, long lengths, and large variation);
small lambdas (lambda = 0) and large lambdas (say, lambda > 10);
low degrees (say, degree = 0,1,2,3) and high degrees (say, degree > 3).
We also need other functions (including generate_D and get_mu) to be tested in various scenarios.
We need the ADMM algorithm to be tested to make sure that it works and is numerically stable in each scenario. Here are a few scenarios that are worth being considered:
current_counts
with two segments, short lengths (e.g., including less than 10 signals), and low variation) and difficult inputs (say,current_counts
with more than 5 segments, long lengths, and large variation);lambda = 0
) and large lambdas (say,lambda > 10
);degree = 0,1,2,3
) and high degrees (say,degree > 3
).We also need other functions (including
generate_D
andget_mu
) to be tested in various scenarios.