Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated

Clique path networks


Proved construction exclusion, not a resolution of Problem 617. The hereditary caps exclude a further family containing graphs with claws, not just the line graphs already excluded in Line-graph obstruction.

Fix an integer r≥3r\ge3. Throughout,

pr(rq+b)=r(q2)+qb(0≤b<r),Dr(n)=(n2)−(r−1)pr(n).p_r(rq+b)=r\binom q2+qb\quad(0\le b<r),\qquad D_r(n)=\binom n2-(r-1)p_r(n).

A graph satisfies the caps if e(H[W])≤Dr(∣W∣)e(H[W])\le D_r(|W|) for all WW. In fact the proofs below only use sets of order at most 2r−12r-1.

1. The class and the conclusion

Start with a loopless multigraph on a set of branch vertices, replace every edge by an internally disjoint path of length at least three, and replace every vertex of the resulting graph by a nonempty clique. Two bags are completely joined exactly when their underlying vertices are adjacent. Arbitrary positive, unequal bag orders are allowed.

Theorem. If the resulting graph HH satisfies the caps, then

χf(H)≤r,∣H∣≤rα(H).(1)\boxed{\chi_f(H)\le r,\qquad |H|\le r\alpha(H).} \tag{1}

There is no edge budget, order bound, or independence-number restriction.

More generally one may add edges between branch vertices provided those direct branch edges form a disjoint union of cliques. The conclusion is unchanged. Parallel subdivided paths are allowed, including paths whose endpoints lie in the same direct branch clique. No other edges between bags are allowed.

Here χf(H)≤r\chi_f(H)\le r will be proved by constructing a probability distribution on independent sets in which every vertex has probability 1/r1/r of being selected. Multiplying this distribution by rr gives a fractional coloring of total weight rr; taking expected cardinality gives the second conclusion in (1).

In particular no color of an E617 counterexample can itself be one of these graphs. If a color is a disjoint union of components in this class, the same conclusion follows by summing the span inequalities.

2. Local consequences of the caps

There is no Kr+1K_{r+1}, and an outsider has at most one neighbor in a KrK_r: both assertions follow from Dr(r+1)=(r2)+1D_r(r+1)=\binom r2+1. Consequently every bag has order at most rr, every nonisolated branch bag has order at most r−1r-1, and every internal path bag has order at most r−2r-2.

For the last assertion, an internal bag of order r−1r-1 together with one vertex in a neighboring bag is a KrK_r. A vertex of its other neighboring bag has r−1≥2r-1\ge2 neighbors in that clique, a contradiction. The same argument shows that a branch bag of order r−1r-1 has at most one neighboring bag. If it is incident with a subdivided path, its branch vertex therefore has multidegree one and no additional direct branch neighbor.

Any adjacent pair of internal bags has total order at most r−1r-1. If its total were rr, a vertex immediately before this pair would force the first bag to have order at most one, and a vertex immediately after it would force the second to have order at most one. This would give r≤2r\le2.

For three consecutive bags of orders a,b,ca,b,c, their only nonedges are the acac pairs between the first and third bags. Thus

ac≥(r−1)(a+b+c−r).(2)ac\ge(r-1)(a+b+c-r). \tag{2}

If the right side is negative this is automatic; otherwise it follows from the cap, since pr(n)≥n−rp_r(n)\ge n-r. The adjacent-pair clique bounds also show that these triples have order at most 2r−12r-1.

Write R=r−1R=r-1. A useful rearrangement of (2) is

r−b−c≥a−acR.(3)r-b-c\ge a-\frac{ac}{R}. \tag{3}

3. Odd paths

Consider a path of length 2k+12k+1, with endpoint bag orders a,fa,f and internal total SS. Put D=kr−SD=kr-S.

For length three, write its bag orders as (a,x,y,f)(a,x,y,f) and δ=r−x−y\delta=r-x-y. The internal-pair bound gives δ≥1\delta\ge1. Applying (2) at both ends and rearranging gives

ax≤a+δ(R−a),fy≤f+δ(R−f).ax\le a+\delta(R-a),\qquad fy\le f+\delta(R-f).

Dividing by a,fa,f and using x+y=r−δx+y=r-\delta yields

δ≥(r−2)afR(a+f)−af.(4)\delta\ge \frac{(r-2)af}{R(a+f)-af}. \tag{4}

The denominator is positive. If a,f≤r−2a,f\le r-2, then (R−a)(R−f)≥1(R-a)(R-f)\ge1, so (4) implies δ≥af/r\delta\ge af/r. If a=Ra=R and f≤r−2f\le r-2, (4) gives δ≥f(r−2)/(r−1)>f−1\delta\ge f(r-2)/(r-1)>f-1; integrality gives δ≥f≥af/r\delta\ge f\ge af/r. The other case is symmetric.

The sole remaining case is a=f=Ra=f=R. The two adjacent internal bags then have orders one, and the branch-degree observation in Section 2 makes this an entire isolated component with bag orders (r−1,1,1,r−1)(r-1,1,1,r-1). Its order is 2r2r, its independence number is two, and it is rr-colorable: use colors 1,…,r−11,\ldots,r-1 on the first bag, color rr on the next bag, color 11 on the third bag, and colors 2,…,r2,\ldots,r on the last bag. Such components can be handled separately.

For every odd path of length at least five, there is a stronger bound:

D≥min⁡(a,f)+k−2.(5)\boxed{D\ge\min(a,f)+k-2.} \tag{5}

Write the end triples as (a,b,c)(a,b,c) and (d,e,f)(d,e,f). The middle path between c,dc,d has odd length 2h+12h+1, where h=k−2≥0h=k-2\ge0; write TT for its internal total. Matching all its bags in consecutive pairs, and then matching only its internal bags, gives respectively

c+d+T≤(h+1)R,T≤hR.c+d+T\le(h+1)R,\qquad T\le hR.

All these pairs consist of internal bags of the original path. Therefore

hr−T≥h+(c+d−R)+.hr-T\ge h+(c+d-R)_+.

Using (3) at both ends now gives

D≥a+f−ac+fdR+(c+d−R)++h.D\ge a+f-\frac{ac+fd}{R}+(c+d-R)_++h.

Assume a≤fa\le f. Since the coefficient of ff is nonnegative, replace ff by aa on the right. After subtracting a+ha+h the result is

a(1−c+dR)+(c+d−R)+≥0,a\left(1-\frac{c+d}{R}\right)+(c+d-R)_+\ge0,

using 0≤a≤R0\le a\le R. This proves (5).

Thus, except for the explicitly removed isolated components, every odd path satisfies S+af/r≤krS+af/r\le kr. For length at least five it even satisfies S+min⁡(a,f)≤krS+\min(a,f)\le kr.

4. Even paths

For a path of length 2k≥42k\ge4, let TT be the total of all its bag orders and put E=T−krE=T-kr. Then

E≤(a+f−r)+.(6)\boxed{E\le(a+f-r)_+.} \tag{6}

For length four, write the orders as (a,b,c,d,f)(a,b,c,d,f). The two end triple inequalities (2) give

E≤c(a+f−R)R.E\le\frac{c(a+f-R)}R.

If a+f≤Ra+f\le R, this is nonpositive. Otherwise a+f−Ra+f-R is a positive integer and c≤R−1c\le R-1, so integrality gives E≤a+f−R−1=a+f−rE\le a+f-R-1=a+f-r.

For length six, write the orders as (a,b,c,z,d,e,f)(a,b,c,z,d,e,f). Applying (2) to the three triples centered at b,z,eb,z,e gives

E≤cd−(R−a)c−(R−f)dR.E\le\frac{cd-(R-a)c-(R-f)d}{R}.

The numerator is bilinear in c,d∈[0,R−1]c,d\in[0,R-1]. Its maximum on this rectangle is

max⁡{0,(R−1)(a+f−r)},\max\{0,(R-1)(a+f-r)\},

as is seen at the four corners. This proves (6).

For length at least eight, write the end triples as (a,b,c)(a,b,c) and (d,e,f)(d,e,f) and retain the even middle path from cc to dd. Its corresponding excess EmidE_{\rm mid} satisfies (6) by induction. Equation (3) at the two ends gives

E≤Emid−c−d+ac+fdR.E\le E_{\rm mid}-c-d+\frac{ac+fd}{R}.

If c+d≤rc+d\le r, the right side is nonpositive. If c+d>rc+d>r, it is at most −r+(ac+fd)/R≤a+f−r-r+(ac+fd)/R\le a+f-r. This proves (6) in every case.

5. Coupling independent sets on a path

The following elementary probability lemma accounts for the endpoint dependencies, rather than treating them as independent without proof. Let L≥1L\ge1 and let p0,…,pL∈[0,1]p_0,\ldots,p_L\in[0,1] satisfy pi+pi+1≤1p_i+p_{i+1}\le1. Prescribe an endpoint joint probability qq satisfying the Fréchet bounds

(p0+pL−1)+≤q≤min⁡(p0,pL).(7)(p_0+p_L-1)_+\le q\le\min(p_0,p_L). \tag{7}

If L=2k+1L=2k+1 is odd, a distribution on independent sets of the path with these marginals and this joint probability exists exactly when

∑i=1L−1pi+q≤k.(8)\sum_{i=1}^{L-1}p_i+q\le k. \tag{8}

If L=2kL=2k is even, it exists exactly when

∑i=0Lpi−q≤k.(9)\sum_{i=0}^L p_i-q\le k. \tag{9}

Necessity follows by counting selected vertices in each of the four endpoint states. Here is a self-contained sufficiency proof.

The stable-set polytope of an odd cycle of order 2s+12s+1 is described by nonnegativity, the edge inequalities, and ∑xi≤s\sum x_i\le s. To check integrality, consider a fractional extreme point. The tight edge equations on fractional coordinates form paths, unless the whole cycle is at 1/21/2, which violates the sum bound. Each path has an alternating perturbation parameter. Two parameters can be perturbed while preserving the sum; one can also be perturbed if the sum bound is slack or its path has even order. In the only remaining case there is one odd-order fractional path and the sum bound is tight. Its sum is an integer plus tt or 1−t1-t, whereas all other coordinates and the total sum are integers, forcing tt to be integral. This is again a contradiction. Thus every feasible point is a convex combination of independent-set indicators.

For odd LL, add one vertex adjacent to both endpoints, forming an odd cycle, and give it marginal 1−p0−pL+q1-p_0-p_L+q. Conditions (7)--(8) make the vector feasible in the cycle polytope. The induced path distribution has an endpoint joint q′≥qq'\ge q, because the added vertex can occur only when both endpoints are absent. On the original path there is also a distribution with joint (p0+pL−1)+(p_0+p_L-1)_+: choose UU uniformly in [0,1][0,1], select even-indexed ii when U≤piU\le p_i, and odd-indexed ii when U>1−piU>1-p_i. Mixing these two distributions gives joint exactly qq.

For even LL, close the path with a new path of length three, whose two internal marginals have sum 1−q1-q, the first at most 1−p01-p_0 and the second at most 1−pL1-p_L. Such choices exist by (7). The resulting odd-cycle vector is feasible by (9). Its endpoint joint q′≤qq'\le q, since selecting both endpoints excludes both new vertices. The threshold distribution just described now has joint min⁡(p0,pL)\min(p_0,p_L), because the two endpoints have the same parity. Mixing again gives qq.

6. The global fractional coloring

Remove the isolated exceptional length-three components from Section 3. In each direct branch clique independently, select a branch bag vv with probability wv/rw_v/r, selecting none with the remaining probability. These are valid probabilities because the joined branch bags constitute a clique of order at most rr. For no direct branch edges this simply selects the branch bags independently.

An endpoint pair in different branch cliques has joint probability af/r2af/r^2; one in the same branch clique has joint zero. Put pi=wi/rp_i=w_i/r on every subdivided path. Adjacent marginals sum to at most one. For length three, Section 3 proves (8) for joint af/r2af/r^2, hence also for zero. For longer odd paths, (5) proves (8) for every joint allowed by (7). For even paths, (6) proves (9) for every allowed joint.

Apply the coupling lemma separately on each path. Conditional on the already sampled endpoint states, sample its interior according to the corresponding conditional distribution. Events of zero endpoint probability need no choice. Interior paths have no edges between them, so all the selected bags together form an independent set of the bag graph, and every bag has marginal wv/rw_v/r.

Choose one uniformly random vertex from each selected bag. Every vertex then has marginal 1/r1/r. This proves (1), including the direct-branch extension.

There is a useful local form. Start with a distribution on independent sets of the remaining graph in which every vertex has marginal exactly 1/r1/r. Any clique path of length at least four, whose internal bags have no neighbors outside their own and adjacent bags, can be inserted without changing this outside distribution. The endpoint joint may be arbitrary. The endpoint bags need only be disjoint cliques; their outside neighborhoods need not be homogeneous, and edges between the endpoint cliques are allowed.

Every graph with χf≤r\chi_f\le r admits the required exact-marginal normalization: divide fractional-color weights by rr, pad with the empty independent set, and independently thin the selected vertices to reduce each marginal to 1/r1/r. This normalization can change the initial distribution. A length-three path is not asserted to have the universal extension property.

Scope and the remaining structural gap

The initial odd-subdivision span proof and the enlarged fractional proof were separately audited. The arguments use only the cap, elementary probability, and the odd-cycle polytope proof above; no external classification theorem or finite computation is invoked.

The class contains, for example, positive clique blowups of the graph obtained by replacing each edge of K4K_4 by a path of length three. Those graphs can have claws, so this is not a restatement of the line-graph exclusion. The construction also permits an unbounded root and unbounded residual independence number.

No surviving least-color core has been proved to admit this decomposition. Its private-clique support description does not imply that exception vertices can be organized into the internally disjoint paths used here. Deleting a strip can leave a smaller graph with fractional chromatic number above rr even when every proper induced set has Hall ratio at most rr. Thus rank minimality cannot be silently replaced by fractional coloring minimality. This theorem excludes a parameterized construction mechanism; it does not prove any new unrestricted E617 case.

7. Allowing paths of length two

Theorem. Let r≥12r\ge12. Start with a loopless multigraph, replace every edge by an internally disjoint path of length at least two, and replace every vertex by a nonempty clique, with complete joins exactly between consecutive bags. If the resulting graph GG satisfies the hereditary caps, then

χf(G)≤r,∣G∣≤rα(G).(10)\boxed{\chi_f(G)\le r,\qquad |G|\le r\alpha(G).} \tag{10}

Parallel root edges and unequal bag sizes are permitted. Unlike the length-at-least-three theorem, this extension does not permit direct edges between branch bags. A correlated distribution on the branch bags replaces their independent selection.

Minimum degree and a common correlation parameter

First assume δ(G)≥r\delta(G)\ge r. Put R=r−1R=r-1, let MM be the largest bag order, and set d=r−Md=r-M. The clique cap gives M≤rM\le r. A largest bag has at least d+1d+1 common neighbors outside itself. Together they form an (r+1)(r+1)-set whose nonedges all lie among those d+1d+1 neighbors. Hence

d(d+1)≥2R.d(d+1)\ge2R.

For r≥12r\ge12, this forces d≥5d\ge5, and therefore

R≤d(d+1)/2≤(d−1)2.R\le d(d+1)/2\le(d-1)^2.

Define

θ=MRd.\theta=\frac{M}{Rd}.

Then 0<θ<10<\theta<1 and

θ≤d−2r+d−2.(11)\theta\le\frac{d-2}{r+d-2}. \tag{11}

After cross multiplication, the latter inequality is exactly r((d−1)2−R)≥0r((d-1)^2-R)\ge0.

Since branch bags are pairwise anticomplete, use the following mixture. With probability 1−θ1-\theta, select them independently with probabilities w/rw/r, where ww is the bag order. With probability θ\theta, choose a common uniform U∈[0,1]U\in[0,1] and select every branch bag with U≤w/rU\le w/r. Each branch bag still has marginal w/rw/r. Endpoint bags of orders a≤f≤Ma\le f\le M have joint selection probability

j=af+θa(r−f)r2.(12)j=\frac{af+\theta a(r-f)}{r^2}. \tag{12}

We check that this single distribution can be extended along every path.

Length two and length three

For a path with bag orders (a,b,f)(a,b,f), put E=a+b+f−rE=a+b+f-r. Its only nonedges are the afaf endpoint pairs, so its cap and pr(n)≥n−rp_r(n)\ge n-r give E≤af/RE\le af/R. Since

θ=MR(r−M)≥fR(r−f),\theta=\frac{M}{R(r-M)}\ge\frac{f}{R(r-f)},

equation (12) gives

rj≥af+af/Rr=af/R≥E.rj\ge\frac{af+af/R}{r}=af/R\ge E.

This is precisely the even-path coupling requirement (9).

For a length-three path (a,x,y,f)(a,x,y,f), put δ=r−x−y\delta=r-x-y. The end-triple caps, as in Section 3, give

δ≥(r−2)afR(a+f)−af.\delta\ge\frac{(r-2)af}{R(a+f)-af}.

The coupling requirement is rj≤δrj\le\delta. To check it for arbitrary unequal endpoints, set A=R−aA=R-a, B=R−fB=R-f. Then A≥B≥d−1A\ge B\ge d-1, and the preceding bound implies

rδ−afa(r−f)≥f(AB−1)(B+1)(R2−AB)≥B−1R+B≥d−2r+d−2≥θ.\frac{r\delta-af}{a(r-f)} \ge\frac{f(AB-1)}{(B+1)(R^2-AB)} \ge\frac{B-1}{R+B} \ge\frac{d-2}{r+d-2} \ge\theta.

For the second inequality, (z−1)/(R2−z)(z-1)/(R^2-z) increases with zz, so replace AA by BB, then use f=R−Bf=R-B. The third follows by monotonicity in BB. This proves the required upper bound for jj.

Paths of length at least four extend every feasible endpoint joint, by Sections 3--5. Sample all branch states first, then sample each path interior conditional on its two endpoint states. The interiors have no cross edges, so the samples combine into independent bags with marginal w/rw/r. Choosing one uniform vertex from each selected bag gives each vertex marginal exactly 1/r1/r. This proves (10) when the minimum degree is at least rr.

Removing the minimum-degree assumption

If G−vG-v already has an independent-set distribution with marginals 1/r1/r and dG(v)≤r−1d_G(v)\le r-1, then the event that no neighbor of vv is selected has probability at least 1−dG(v)/r≥1/r1-d_G(v)/r\ge1/r. Add vv on a randomized subevent of probability exactly 1/r1/r. This preserves all existing marginals and gives the required marginal at vv.

Repeatedly delete vertices of degree at most r−1r-1. In the remaining induced graph the bags may have smaller orders or disappear. A nonempty bag with quotient degree zero or one cannot remain: it and its sole neighboring bag, if any, form a clique of order at most rr, making its vertices' degree at most r−1r-1.

Thus a surviving original internal bag retains both its neighboring bags. Propagating along its original path forces the whole path and both branch bags to survive. The nonempty remainder is still a clique blowup of a root subgraph with every path of length at least two. Apply the minimum-degree construction to it, then restore deleted vertices in reverse order by the preceding extension argument. If the remainder is empty, start with the empty-set distribution.

This proves the fractional conclusion itself, not merely the unweighted span inequality. It does not replace induced-ratio minimality by fractional minimality. No least-color budget or independence restriction is used in this theorem, and no general E617 core is known to have the required subdivision structure.

The correlation bounds, arbitrary-endpoint coupling, preservation of the class under taking the rr-core, and reverse low-degree extension were checked by the author and are author-recorded.