memoizing arbitrary resumable time series, trying not to re-invent the wheel

I've written a research tool that allows users to write arbitrary expressions to define time series calculated from a set of primary data sources. Many of the functions I provide carry state derived from previous values, such as EMA. In example:

EMA(GetData(Foo), 280)

State contained in the component functions of these expressions can be saved and resumed with AST node labeling at compile time. This allows a series to be resumed later when any of its root data sources, which are wrapped in simple bookmarks, contain more elements.

This is done because the primary data sources can be very large, the transforms can be compute-intensive, or the user might want to combine hundreds or thousands of such series for real-time high-dimensional analysis.

I'm finding it hard to believe there isn't already something out there that does exactly this, either an open-source project or a tool. Am I re-inventing the wheel here?