Series

Ehrenfeucht–Fraïssé games — the series

8 essays on one idea, from the one that introduces it to the one that assumes the rest.
  1. 3 rounds on chains of 4 and 5. Two chains of dots with pebbles placed in turn, and the transcript of a play: Spoiler picks an element of one chain, Duplicator answers in the other, and the pebbles must keep the same order.

    A game that decides what can be said

    Two players take turns pointing at elements of two structures; if the second can survive k rounds, then no sentence with k quantifiers tells the structures apart — a statement about infinitely many formulas, settled by a finite search.

    part 1 · logic
  2. Three properties that leave the middle, and one that cannot. Four measured curves of the share of random graphs having a property, plotted against the number of points: three first-order properties running to zero or one, and the parity of the edge count sitting on a half throughout.

    Nearly always, or nearly never

    Toss a coin for every pair of points and ask whether the graph that results has some property. For a property a first-order sentence can state, the answer in the limit is never a genuine probability — it is zero or it is one, and the game is what proves it.

    part 2 · logic
  3. What a sentence of depth 2 can reach. Two rings of points, of 14 and 19 points, each with a run of 9 consecutive points marked as the neighbourhood a sentence of depth 2 can inspect.

    The distance a sentence can see

    A first-order sentence with three quantifiers cannot notice anything about a graph beyond a fixed distance from the points it names. That single limitation is why it cannot say connected, and why the failure survives every attempt to add more quantifiers.

    part 3 · logic
  4. A 6-cycle and two 3-cycles: refinement cannot tell them apart. Two graphs side by side — one cycle and two smaller cycles — with the same number of points, the same number of edges and every point of the same degree.

    The game the algorithm was playing

    Change what Duplicator has to offer — a whole bijection instead of one element — and the game stops measuring first-order logic and starts measuring colour refinement, the algorithm every practical graph-isomorphism test begins with. Two subjects that grew apart are one game with the moves relabelled.

    part 4 · logic
  5. Where a word of m letters stops being distinguishable from one of m + 1. A row for each quantifier depth and a cell for each length, shaded where Duplicator survives the game between words of that length and one longer.

    A language that can name a set

    Allow a sentence to quantify over sets of positions as well as positions, and on words the answer changes completely: the sets buy exactly the languages a finite automaton recognises. Whether the number of letters is even is the smallest example of what the sets are for.

    part 5 · logic
  6. Two graphs every count agrees on, and one question that does not. Two sixteen-point graphs drawn on a four-by-four grid, with one point marked and its six neighbours highlighted in each, the neighbours forming two triangles in one and a six-cycle in the other.

    One gadget defeats every refinement

    Colour refinement fails on two triangles against a hexagon; its two-dimensional version fixes that and fails on a pair of strongly regular graphs. For every k there are two graphs the k-dimensional version cannot separate, and they are built from one local piece whose only symmetry is a parity.

    part 6 · logic
  7. Who survives with two pebbles and who with three, over 4 rounds. A table of three pairs of graphs with, for each, whether the duplicating player survives a two-pebble game and a three-pebble game played to a fixed depth.

    The boundary at three variables

    Restrict a sentence to two variable names and it can still be arbitrarily long, because the names are reused. What it cannot be is deep: every satisfiable two-variable sentence has a small model, so asking whether one is satisfiable is a bounded search. Allow a third name and the question becomes undecidable.

    part 7 · logic
  8. A subword can be checked with one quantifier block; a factor cannot. Subword aba: monoid 7, J-trivial, degree 3; factor aba: monoid 12, aperiodic, not J-trivial; witness abab / abba.

    One block of quantifiers, read off an algebra

    Which properties of words can be stated with existential quantifiers alone — 'there are positions, in this order, carrying these letters' — and Boolean combinations of such statements? Imre Simon answered in 1975: exactly those whose syntactic monoid is J-trivial. Of the 57,068 languages that automata of up to four states recognise, 848 pass the test, and every one of them is decided by which words of length three or less it contains as scattered subwords.

    part 8 · logic

All series