Frequently asked questions¶
Can any step function be modelled?
Unfortunately no. The intervals comprising a step function must have non-zero length and either be all left-closed right-open, or right-open left-closed. Also, there cannot be an infinite number of intervals.
What happens if I call
staircase.Stairs.layer() with default arguments?
The default arguments for start, end and value in
staircase.Stairs.layer() are all None, as per the method signature. However the internals of staircase will treat them as -infinity, infinity and 1 respectively. The result of this will be increasing the value of a step function by 1 everywhere. For a discussion around this choice please see The layer method and then Gotchas.
Why can’t I calculate the integral of my step function?
This is an issue which can arise when using a datetime domain. The result of an integral calculated on such a step function is expressed as a
pandas.Timedelta. Unfortunately this class has limitations which may be exceeded with integral calculations. A workaround may involve scaling your step function values down before calculating the integral, eg:
Can I work with
numpy.datetime too. However, pandas converts these types to its own
pandas.Timestamp type, and consequently staircase does too. If you need to convert a result back to your desired class then you can do so with
Can I use method chaining?
What is sc.inf?
staircase.inf is a singleton object of
staircase.Inf, which is used to represent the concept of infinity within staircase domains (regardless of domain type). You are welcome to use it and its negative counterpart (-staircase.inf) when specifying domain bounds but the use of None can be substituted in place.
>>> my_step_function.clip(-sc.inf, 10)
is equivalent to
>>> my_step_function.clip(None, 10)
Does my step function plot have infinite intervals?
A step function can have at most two infinite intervals, which trail off to negative and positive infinity respectively. When inspecting the step function with
staircase.to_frame() it will be obvious if the step function has infinite intervals. If they exist they will be listed in the first and last rows of the dataframe. You may also be able to infer an answer to this question from a plot.
In the above example, the step function on the left has two infinite intervals (each having a value of 0). The step function on the right only has one infinite interval, trailing off to negative infinity. For all domain values greater than 7 the step function is undefined. The difference between these two step function can be noticed from the above plot by a keen observer. If a plot is produced with an argument style = “hlines” then it will be impossible to identify infinite intervals in the plot unless arrows = True. See
staircase.plot() for details on these parameters.