There is currently code in tf.simplify which attempts to round transfer function coefficients and keep them as integers. This causes significant pain to people who have real values like 1.231e-2 in their tranfer function description.
I have added a flag to the tf object which allows one to bypass this part of the simplification. However, this will still cause issues when doing "normal" operations with these objects, as they are simplfied on creation.
If we don't do this, we need to figure out how else to normalise the transfer functions so that we can get repeatable output. Is tf([1], [3, 3]) better than for instance tf([0.3333], [1, 1])?
There is currently code in
tf.simplify
which attempts to round transfer function coefficients and keep them as integers. This causes significant pain to people who have real values like 1.231e-2 in their tranfer function description.I have added a flag to the
tf
object which allows one to bypass this part of the simplification. However, this will still cause issues when doing "normal" operations with these objects, as they are simplfied on creation.If we don't do this, we need to figure out how else to normalise the transfer functions so that we can get repeatable output. Is
tf([1], [3, 3])
better than for instancetf([0.3333], [1, 1])
?