Each entry is a single-file interactive page on a long-standing problem of the form
what's the extremal configuration that… — the largest set of points whose
pairwise distances obey some rule, the smallest chromatic number that colours a metric
graph, the sparsest line cover of a plane, the largest cap inside 𝔽₃ⁿ. They share a
scaffold and a sensibility, and each one frames its problem in whatever visual language
the underlying math actually rewards.
This page is the hub: a family-resemblance table of what's built, and an explicit
roadmap of what's next.
Built
| name |
problem |
era |
technique |
status |
Click a column header to sort. Click a row to open the entry. Era is the year
of the most recent resolution or breakthrough.
adjacent ·
elements — a
periodic-table mandala. Strictly chemistry, not extremal geometry, but it borrows the
same single-file canvas scaffold and rounds out the visual family.
sibling hub ·
conjectures —
a searchable index of ~136 open mathematical conjectures across nine fields, each with a
context page (statement, logical form, what a counterexample would look like, current
evidence, further reading). Where this pack builds one deep interactive page per problem,
conjectures indexes the whole landscape of "find-a-counterexample" open problems — several
of the pack's own entries (erdős, runner, hadwiger, borsuk, viazovska) appear there too.
Roadmap
Next
-
/elekes/ — sum-product bound. Erdős–Szemerédi 1983
(same year as ST, sister problem with shared machinery); progress over decades;
current bound around |A|^{4/3 − ε}.
-
/orchard/ — orchard visibility / ordinary lines.
Green–Tao 2013 resolved the asymptotic.
-
/kepler/ — the dimension-3 sphere packing optimum
(Hales 1998, formalised by Flyspeck 2014). Natural prequel to
viazovska; the computer-assisted-proof
angle pairs with hadwiger's SAT story.
Other candidates
-
/cohen-pohoata-zakharov/ — explicit 2024 improvement
on Heilbronn's upper bound, via polynomial method. Could spin off from
heilbronn as a technique-focused
explainer.
-
/frankl-wilson/ — the rank-of-incidence-matrix
theorem that drove the original Kahn–Kalai counterexample on
borsuk. Adjacent to capset; could anchor
a "finite-field linear-algebra toolkit" page.
-
/cohn-elkies/ — the linear-programming bound behind
viazovska. An interactive "magic function"
explorer (drag the auxiliary function, watch the LP bound move) would make the
modular-form punchline land.
Branching ideas (adjacent to the main pack)
-
/dvir/ — a dedicated explainer for the
polynomial-method, walking Dvir's five-page Kakeya proof in interactive steps.
Shares material with kakeya; would be the
technique's "from scratch" lecture-hall page.
-
/polymath/ — the Polymath collaborative projects.
Polymath16 (the de Grey graph reduction) is the most natural entry; tells the
"math by github" story.
-
/erdos-problems/ — an interactive index of all the
Erdős open problems with progress trackers. Linked from
erdős and
guthkatz; mirrors the
erdosproblems.com taxonomy.
House rules
Every entry has
- One single-file static HTML page in its own top-level directory (
/<name>/index.html).
- Header with crumb to mino.mobi, h1, subtitle, sister-site crossref strip.
- Format bar (theme + reveal-all) where colour or hover-driven viz benefits from it.
- At least one interactive viz tab; optional growth, history, play, and docs tabs.
- A distinct accent colour from its siblings.
- An entry on the root landing page in the
data category, age hot.
- Math validated by a smoke test inside the deploy commit body.
What we avoid
- No server-side compute. Everything static HTML + canvas JS. The polynomial-method demos tempt one toward server-side SAT solvers; resist.
- No problems that need > 500 vertices to demonstrate. The page becomes click-uncolourable. Reveal-only static rendering is fine — see hadwiger's de Grey graph.
- No log-log growth charts for problems with two data points. A prose timeline reads better.
- No overclaiming. Be honest when a procedural construction doesn't quite reach the claimed bound — the page is still educational.