This part of the Auto-generate project is concerned with recognizing the type/definition of the sequence and then, in particular, extracting the coefficients. Please see the talk page for the discussion.
Aim and scope
We need to know
- whether a given sequence is of "allowed type" ("currently": linear recurrence with constant coefficients)
- whether the data (DATA & other: link to index/Rec#order.. & signature; ...) is sufficient to compute anything
- whether it is of some particular sub-class:
- polynomial : yields particularly simple formulae & programs
- large data / coefficients: in computations of data, care must be taken to avoid corrupt results. Use qualitative and/or quantitative indicators for this. Fix limits on index, size of a(n), add consistency checks (a(n) mod p or similar) to detect unexpected errors. Report such errors.
- linear recurrence of low orders: special formulae & methods exist for 1st, 2nd,... order linear recurrences, etc.
- what (other) consistency checks should be added: compare information from scan of NAME, "hand made" FORMULAs, PROGRAMs...
(detailed considerations concerning the above points)
How the type of a sequence is determined
- use ggf on DATA ?
- use LINKS to /index/Rec#... & signature ?
- scan NAME or FORMULA for simple patterns ?
Compare the result obtained from the various sources above.
Here go the scripts. We split up nontrivial parts (functions) for easier editing & documentation in separate subsections when this seems useful.