Record four 2026 lower bounds for the Shannon capacity of C₇

[Chessing234]

Refs #168 (item 2).

The C₉ page still stopped at Polak–Schrijver $367^{1/5}\approx 3.2578$ (2018). Four later constructions, all checked against the arXiv texts rather than a summary:

• Itty–Rosin–Carstensen–Reichman, arXiv:2607.21517: independent set of size $134753$ in $C_7^{\boxtimes 10}$, so $\Theta(C_7)\ge 134753^{1/10}\approx 3.258020$.
• Gao, arXiv:2607.27869: recursive product giving $\Theta(C_7)\ge 3.258789153908\ldots$ from an independent set in $C_7^{\boxtimes 200}$.
• Buys–Polak–Zuiddam, arXiv:2607.29681: $\Theta(C_7)\ge 3.258805369885\ldots$, Lean-formalized.
• Tandon, arXiv:2608.30273: heterogeneous recursion, $\Theta(C_7)\ge 3.25883262\ldots$ from an independent set in $C_7^{\boxtimes 500}$.

The README cell is updated to the last of these, the bound table keeps the history in chronological order, and the other odd-cycle bounds from BPZ are noted on the page rather than split into new constants.

Construction/certificate links from the papers are in the comments column. AI was used to locate the papers; every numerical claim above was read from the source.

Test plan

• Recomputed $134753^{1/10}$ and $367^{1/5}$ locally
• Checked each bound against the arXiv abstract/theorem statement
• README table cell matches the last lower-bound row
• No open PR already covering C₉

Made with Cursor