Symbulator for calculators — versions 4 to 8

The original Symbulator, written in TI-Basic for the TI-89 Titanium and later ported to the TI-Nspire CX II CAS. Version 9 — the Python/SymPy port — lives at github.com/Symbulator/solver; this repository is the source it came from.

It exists because these files were, until now, in one folder on one disk. They are the reference for every question about what Symbulator is supposed to do, and several such questions have already been settled by reading them rather than by inference.

What is here

Folder	What it holds
v4/, v5/, v6/, v7/	TI-89 Titanium releases, as .tig and .89g group files
v8/	TI-Nspire CX II CAS releases, as .tns documents
decoded/	Readable text recovered from the two current versions

v8/lf.tns is Lars Frederiksen's Laplace library, adapted for the Nspire by Philippe Fortin, which versions 7 and 8 depend on for their transforms. It is their work, not mine, and is included here because version 8 does not run without it.

Reading the source

Version 8 (.tns) is a ZIP-like container whose Problem1.xml entry uses compression method 13 — TI's own envelope: a fixed-key 3DES header, then a derived 3DES counter-mode body, then raw deflate, then a TIXC0100 XML expansion. The counter mode is why the payload is not block-aligned, which is what makes it look like something other than encryption at first.

It decodes with tnstools:

python tnstools.py -tns v8/s.tns -out v8xml

decoded/v8_programs.txt is the result, split into the 30 programs, and it is a complete and faithful recovery — every line of the original is there.

Version 7 (.tig) is a plain ZIP of tokenised TI variable files. No open-source detokeniser exists for program bodies (tilibs and libticonv handle variable names only), so decoded/v7_programs_partial.txt recovers the printable strings and loses the token bytes. It is a reading aid, not a listing — good enough to find which program does what and to read the string literals, and not good enough to quote as evidence. Where version 7 and version 8 agree, trust the version 8 text.

Questions this has already answered

Kept as a record of why it is worth having, and as a list of things not to re-derive:

• TR reads its sources in the time domain; FD reads them in s. symbv8s5 transforms the value of an e or j element when the tool is tr: a function of t through t2s, a constant as value/s, and a value that refers to another element's answer left alone, because a controlled source is a relation rather than a waveform. The Python port had only ever inverse- transformed its answers, which made every transient result one integration short.
• {...} is t2s(...), and only in FD. symbv8si does exactly two substitutions and only when the tool is fd. Not to be confused with [...], which is pr(...) and applies everywhere.
• A mutual inductance couples r elements as well as l ones. symbv8s8 adds the mutual term without a jw factor when the analysis is AC, because a value in jΩ is already an impedance. The port stamped only the henries form, so a coupled pair written in jΩ silently carried no current in the secondary.
• A delta of three sources cannot be solved, and the original's answer is to describe it with two. The third adds no information and one more unknown.

Licence

Symbulator has always been free. From June 2013 it was offered under a Creative Commons BY-NC-SA licence; it is now open source under the MIT licence, which supersedes it.

That covers my own code. v8/lf.tns is Lars Frederiksen's and Philippe Fortin's, under whatever terms they set, and is redistributed here by their long-standing permission to host it.

Not included. The roughly 300 solved problems that accompanied version 8 are worked solutions to problems from copyrighted textbooks. They were offered for teaching, which is not the same as publishing them in a public repository, so they are deliberately absent and .gitignore keeps them out.