betaFeedback

WorksMathematics

An additive-square-free word over an alphabet of size 6

Abstract

An additive square over N\mathbb{N} is two consecutive nonempty blocks of the same length and same sum. For example, 1423 is an additive square, since 14 and 23 both have length 2 and sum 5. Solving a problem that has been open for more than 50 years, we give an example of an infinite word over a 6-letter subset of N that avoids additive squares.

Introduction

A word is a sequence of letters drawn from a set called its alphabet. A block is a finite run of consecutive letters. For integer letters, an additive square consists of two adjacent nonempty blocks with equal lengths and equal sums. For example, 1423 is an additive square because each block has two letters and each sums to five.

Can an infinite word on a finite set of integers avoid every additive square? Justin [3] [3], Brown and Freedman [1] [1], Pirillo and Varricchio [4] [4], and Halbeisen and Hungerbühler [2] [2] each posed the problem in different forms. Brown and Freedman conjectured that every bounded sequence of positive integers contains an additive square.

This note proposes an explicit word YY on six positive integers that avoids additive squares. The proof has two steps. First we build an auxiliary word ww by repeatedly replacing each letter by a block of qq letters. We show that a square in ww would give a square in ww with its block length divided by qq. This rules out a square of least positive block length. We then use a four-symbol construction of Peltomäki and Whiteland [5] to encode ww as YY, in such a way that any square in YY would give a square in $w`.

Positions, cuts, and sums. In a word X=x0x1x2⋯X=x_0x_1x_2\cdots, position nn holds the letter xnx_n. The cut at nn is the boundary immediately before xnx_n. The letters before this boundary form the prefix of length nn. For integer letters its sum is SX(n)=∑j=0n−1xjS_X(n)=\sum_{j=0}^{n-1}x_j, with SX(0)=0S_X(0)=0. Two adjacent blocks of length hh starting at position nn have three boundary cuts, nn, n+hn+h, n+2hn+2h. Their sums are equal precisely when

SX(n+h)−SX(n)=SX(n+2h)−SX(n+h).S_X(n+h)-S_X(n)=S_X(n+2h)-S_X(n+h).

We call hh the half-length of the square.

The six integers and the word

Set

D=258,H=2D+1,L=2⋅8511,U=L3+1.D=2^{58},\qquad H=2D+1,\qquad L=2\cdot85^{11},\qquad U=L^3+1.

The alphabet, in increasing order, is

{1, U+1, U+2, U+L+1, U+L2+1, (H+2)U+1}.(1)\boxed{\{1,\ U+1,\ U+2,\ U+L+1,\ U+L^2+1,\ (H+2)U+1\}}. \tag*{(1)}

Define the 85-symbol word pp to be

0120232123203231301020103101213121021232021013010203212320231210212320232132303132120.

For each symbol c∈{0,1,2,3}c\in\{0,1,2,3\}, let φ(c)\varphi(c) be pp with cc added modulo four to each digit. A substitution replaces each letter by its assigned word and joins the resulting words in order. Use this rule to apply φ\varphi eleven times to the two-letter word 01. The result is v=φ11(01)v=\varphi^{11}(01), of length LL. This substitution is due to Veikko Keränen, and is given in [5 Section 3].

The integer wiw_i in the rule below is specified in Section 4 and satisfies 1≤wi<D1\leq w_i<D. To encode wiw_i, take one copy of vv. Label the first wiw_i zeros in this copy by (H+2)U+1(H+2)U+1. Leave the remaining zeros up to and including the DD-th zero at U+1U+1, then label the next wiw_i zeros by 1. All later zeros receive U+1U+1. The symbols 1, 2, 3 receive the same labels in every copy.

Equivalently, write the requested position as n=Li+rn=Li+r with 0≤r<L0\leq r<L. Thus ii selects the copy and rr, its offset, counts how many letters precede position nn within the copy. Let c=vrc=v_r and let kk count the zeros among the first rr letters of vv. The rule for YnY_n is

Yn={(H+2)U+1c=0,0≤k<wi,1c=0,D≤k<D+wi,U+1c=0,in all other cases,U+2c=1,U+L+1c=2,U+L2+1c=3.(2)Y_n= \begin{cases} (H+2)U+1 & c=0,\quad0\leq k<w_i,\\ 1 & c=0,\quad D\leq k<D+w_i,\\ U+1 & c=0,\quad\text{in all other cases},\\ U+2 & c=1,\\ U+L+1 & c=2,\\ U+L^2+1 & c=3. \end{cases} \tag*{(2)}

For instance, with B=(H+2)U+1B=(H+2)U+1, the word YY begins

B, U+2, U+L+1, B, U+L+1, U+L2+1, U+L+1, U+2,…B,\ U+2,\ U+L+1,\ B,\ U+L+1,\ U+L^2+1,\ U+L+1,\ U+2,\ldots

Each word φ(c)\varphi(c) contains all four symbols. The last substitution step joins 2⋅85102 \cdot85^{10} such words, so vv has more than 2D2D zeros. Because 1≤wi<D1 \le w_i < D, all three labels for zero occur in the copy encoding wiw_i. The symbols 1,2,31,2,3 supply the other three labels. Thus YY uses all six integers in (1).

Forcing the half-length to be a multiple of qq

Set q=820=260q = 8^{20} = 2^{60} and M=243M = 2^{43}. Our aim is to build a periodic sequence gg such that the residues of the letters of ww modulo MM force the beginning, middle, and end of a square to lie at the same offset in their respective length-qq blocks.

For a statement EE, the Iverson bracket [E][E] is 1 when EE is true and 0 when EE is false. Given a residue xx modulo qq, take its representative between 0 and q−1q-1. Write this representative in base 8 as x0+8x1+⋯+819x19x_0 + 8x_1 + \cdots+ 8^{19}x_{19}, using leading zeros as needed. Each digit xjx_j belongs to {0,…,7}\{0,\ldots,7\}. Define

F(x)=∑j=0193j[xj≥4]+320∑j=019xj2.F(x) = \sum_{j=0}^{19} 3^j [x_j \ge4] + 3^{20}\sum_{j=0}^{19} x_j^2.

The first sum records which digits lie in the upper half {4,5,6,7}\{4,5,6,7\}. The second uses the squares of all twenty digits.

Lemma 1. If x+z≡2y(modq)x+z \equiv2y \pmod q and F(x)+F(z)≡2F(y)(modM)F(x)+F(z) \equiv2F(y) \pmod M, then x=y=zx=y=z in Z/qZ\mathbb{Z}/q\mathbb{Z}.

Proof. There are twenty digits, each at most 7, so the sum of their squares is at most 20⋅72=98020 \cdot7^2 = 980. The first sum in F(x)F(x) is less than 3203^{20}. Hence 0≤F(x)<981⋅320<M/20 \le F(x) < 981 \cdot3^{20} < M/2. The integer F(x)−2F(y)+F(z)F(x)-2F(y)+F(z) is divisible by MM and has magnitude less than MM, so it is zero.

For each digit position jj, put

cj=[xj≥4]−2[yj≥4]+[zj≥4].c_j = [x_j \ge4] - 2[y_j \ge4] + [z_j \ge4].

Reducing the equality of the FF-values modulo 3203^{20} gives

c0+3c1+32c2+⋯+319c19≡0(mod320).c_0 + 3c_1 + 3^2c_2 + \cdots+ 3^{19}c_{19} \equiv0 \pmod{3^{20}}.

Here the units coefficient means c0c_0, the coefficient of 30=13^0=1 in this particular sum. Every other summand is divisible by 3, so 3 divides c0c_0. Since −2≤c0≤2-2 \le c_0 \le2, we get c0=0c_0=0. Divide by 3 and repeat to get c1=0c_1=0, then c2=0c_2=0, and so on. If yj≥4y_j \ge4, the equation cj=0c_j=0 requires both xj,zj≥4x_j,z_j \ge4. If yj<4y_j<4, it requires both xj,zj<4x_j,z_j<4.

Thus the three digits at position jj lie in a set of four consecutive integers, and ∣xj−2yj+zj∣≤6\lvert x_j-2y_j+z_j\rvert\le6. Apply the same divisibility argument in base 8 to ∑j=0198j(xj−2yj+zj)≡0(mod820)\sum_{j=0}^{19}8^j(x_j-2y_j+z_j)\equiv0\pmod{8^{20}}. Each coefficient is zero, giving yj=(xj+zj)/2y_j=(x_j+z_j)/2. The first sums in F(x)−2F(y)+F(z)F(x)-2F(y)+F(z) have already cancelled, so the squared digits must satisfy

0=∑j(xj2+zj2−2yj2)=12∑j(xj−zj)2.0 = \sum_j (x_j^2+z_j^2-2y_j^2) = \frac{1}{2}\sum_j (x_j-z_j)^2.

Each square in the last sum vanishes. Hence xj=zj=yjx_j=z_j=y_j for each jj, as required.

Let grg_r be the representative in {1,…,M}\{1,\ldots,M\} of F(r+1)−F(r)F(r+1)-F(r) modulo MM, where the arguments of FF are read modulo qq. This makes gg periodic with period qq. Consider any integer word X=x0x1⋯X=x_0x_1\cdots satisfying xn≡gnx_n\equiv g_n (mod MM) for all nn. Adding these congruences from 0 to n−1n-1 gives SX(n)≡F(n)−F(0)S_X(n)\equiv F(n)-F(0) (mod MM), because the intermediate FF-values cancel. If XX has an additive square starting at nn with half-length hh, apply Lemma 1 to the residues nn, n+hn+h, n+2hn+2h modulo qq. The first congruence holds by equal spacing, and the second follows from equality of the adjacent block sums in XX. Hence

n≡n+h≡n+2h(modq),so q∣h.(3)n\equiv n+h\equiv n+2h \pmod q,\qquad\text{so }q\mid h. \tag*{(3)}

Building the auxiliary word ww

Let Q=39Q=3^9, T=(Q−1)/2T=(Q-1)/2. Form the finite set of positive integers

A={gr+Md:0≤r<q, d∈{0,1,Q}}.A=\{g_r+Md:0\le r<q,\ d\in\{0,1,Q\}\}.

To count the possible values of grg_r, add one to the twenty base-8 digits of rr. The trailing 7s change to zeros, and the first digit to their left increases by one. For instance, the base-8 string 2577 becomes 2600. All unchanged digits cancel from F(r+1)−F(r)F(r+1)-F(r). The number of trailing 7s and the old value of the digit to their left therefore determine grg_r. There are 20⋅720\cdot7 choices for these two numbers and one more case, when all twenty digits are 7 and r+1r+1 wraps to zero. Thus ∣A∣≤423|A|\le423. Moreover max⁡A≤M(Q+1)<D\max A\le M(Q+1)<D.

Give each a∈Aa\in A its rank jj in increasing numerical order, starting from 1. Define u(a)u(a) by reading the binary digits of jj as base-3 digits. For example, rank 5=10125=101_2 receives the numerical label 1013=10101_3=10. We call u(a)u(a) the tag of aa. The rank fits in nine binary digits, so 1≤u(a)≤T1\le u(a)\le T. The following implication lets us recognize when three letters coincide.

u(a)+u(c)≡2u(b)(modQ)⟹a=b=c.(4)u(a)+u(c)\equiv2u(b)\pmod Q\quad\Longrightarrow\quad a=b=c. \tag*{(4)}

Indeed, write the three tags in base 3. In the difference u(a)−2u(b)+u(c)u(a)-2u(b)+u(c), the coefficient of each power of 3 lies in [−2,2][-2,2]. Successive division by 3, as in Lemma 1, shows that each coefficient is zero. The binary digits of the three ranks therefore agree at every position, so their letters coincide.

We give the replacement block σ(a)\sigma(a) two jobs. Its letters must have the prescribed residues grg_r modulo MM, and their sums must retain the value of the original letter aa. At offset rr we use gr+Mdr(a)g_r+Md_r(a). The table specifies the integer dr(a)d_r(a) at each of the qq offsets of σ(a)\sigma(a).

Number of positionsValues of dr(a)d_r(a), from left to right
TTu(a)u(a) ones, followed by zeros
q−T−Qq-T-Q3a3a copies of QQ, followed by zeros
QQQ−u(a)Q-u(a) ones, followed by zeros

Table 1.

The middle region has room for all 3a3a copies of QQ because a<Da<D and 3D+T+Q<q3D+T+Q<q. Every letter in σ(a)\sigma(a) belongs to AA. The beginning and end of σ(a)\sigma(a) contribute a total of u(a)+(Q−u(a))=Qu(a) + (Q - u(a)) = Q to the sum of its dd-values, and the middle contributes 3Qa3Qa. Thus

∑r=0q−1dr(a)=3Qa+Q.(5)\sum_{r=0}^{q-1} d_r(a) = 3Qa + Q. \tag*{(5)}

Since u(a)>0u(a) > 0, the first letter of every σ(a)\sigma(a) is a∗=g0+Ma_* = g_0 + M. Start with a∗a_* and apply σ\sigma repeatedly. Write σk(a∗)\sigma^k(a_*) for the result after kk substitutions. Because σ(a∗)\sigma(a_*) begins with a∗a_*, each σk(a∗)\sigma^k(a_*) is a prefix of σk+1(a∗)\sigma^{k+1}(a_*). The lengths qkq^k grow without bound. Define ww to be the unique infinite word with all these prefixes. Another application of the same substitution gives σ(w)=w\sigma(w) = w.

Why a square in ww would force a shorter square

Write Pi=Sw(i)P_i = S_w(i) for the sum of the first ii letters of ww. The second difference of any sequence ff, starting at ii with step e>0e > 0, is the number

Δe2f(i)=f(i)−2f(i+e)+f(i+2e).\Delta_e^2 f(i) = f(i) - 2f(i + e) + f(i + 2e).

For the sequence of prefix sums PP, this is the right block sum minus the left block sum. Thus a square in ww starting at ii with half-length ee is equivalent to Δe2P(i)=0\Delta_e^2 P(i) = 0.

Claim. The word ww contains no additive square.

Proof. Suppose ww contains a square. Choose one whose half-length hh is as small as possible. Every letter of w=σ(w)w = \sigma(w) is congruent to gng_n modulo MM at its position nn. By (3), h=qeh = qe for a positive integer ee. Write the starting cut as qi+rqi + r, 0≤r<q0 \le r < q. The other two cuts are q(i+e)+rq(i + e) + r and q(i+2e)+rq(i + 2e) + r, each at the same offset rr in its own replacement block.

Fix this rr. Define Zr(a)=∑s=0r−1ds(a)Z_r(a) = \sum_{s=0}^{r-1} d_s(a), the sum of the first rr entries of the table in the block σ(a)\sigma(a). For each block index jj, write zj=Zr(wj)z_j = Z_r(w_j). In function notation j↦Zr(wj)j \mapsto Z_r(w_j) denotes the sequence with value zjz_j at index jj. Thus zi+ez_{i+e}, for example, is the partial sum in the block σ(wi+e)\sigma(w_{i+e}).

Before cut qj+rqj + r there are jj complete replacement blocks and the first rr entries of σ(wj)\sigma(w_j). By (5) their dd-values sum to 3QPj+Qj+zj3QP_j + Qj + z_j. Since gg is periodic with period qq, the two adjacent blocks of length qeqe have identical gg-sums. For the full letter sums to agree, their dd-sums must therefore agree as well. Take the second difference of 3QPj+Qj+zj3QP_j + Qj + z_j at j=i,i+e,i+2ej = i, i + e, i + 2e. The term QjQj cancels, giving

3Q(Pi−2Pi+e+Pi+2e)+zi−2zi+e+zi+2e=0.(6)3Q\left(P_i - 2P_{i+e} + P_{i+2e}\right) + z_i - 2z_{i+e} + z_{i+2e} = 0. \tag*{(6)}

The first parenthesis in (6) compares two complete length-ee blocks of ww. The three zz-terms account for where the common offset rr falls inside the replacement blocks.

First region. If 0≤r≤T0 \le r \le T, all three partial sums zjz_j belong to [0,T][0,T]. Hence ∣zi−2zi+e+zi+2e∣≤2T<3Q\lvert z_i - 2z_{i+e} + z_{i+2e} \rvert\le2T < 3Q. The first term in (6) is an integer multiple of 3Q3Q, so it must be zero. This gives Δe2P(i)=0\Delta_e^2 P(i) = 0.

Middle region. If T≤r≤q−QT \le r \le q-Q, the first region has contributed u(a)u(a) and all later contributions so far are multiples of QQ. Thus Zr(a)≡u(a)(modQ)Z_r(a) \equiv u(a) \pmod{Q}. Reduce (6) modulo QQ and use (4) to get wi=wi+e=wi+2ew_i = w_{i+e} = w_{i+2e}. Their partial sums zi,zi+e,zi+2ez_i, z_{i+e}, z_{i+2e} are equal, so Δe2P(i)=0\Delta_e^2 P(i) = 0 here as well.

Last region. If q−Q≤r<qq-Q \le r < q, let Rr(a)R_r(a) be the sum of the dd-values from offset rr to the end of σ(a)\sigma(a). Only ones and zeros remain, so 0≤Rr(a)≤Q0 \le R_r(a) \le Q. Insert Zr(a)=3Qa+Q−Rr(a)Z_r(a) = 3Qa + Q - R_r(a) into (6). Since Pj+wj=Pj+1P_j + w_j = P_{j+1}, this gives

3QΔe2P(i+1)=Rr(wi)−2Rr(wi+e)+Rr(wi+2e).3Q\Delta_e^2P(i+1)=R_r(w_i)-2R_r(w_{i+e})+R_r(w_{i+2e}).

The right-hand side has magnitude at most 2Q2Q, so Δe2P(i+1)=0\Delta_e^2P(i+1)=0. All cases produce a square of half-length e=h/q<he=h/q<h in ww, contradicting minimality.

Why the six-letter encoding preserves avoidance

Two blocks are abelian equivalent if every symbol appears the same number of times in each block. Peltomäki and Whiteland show in the proof of their Theorem 1.2 that, for every m≥0m \ge0, the periodic repetition of φm(01)\varphi^m(01) has no two adjacent abelian-equivalent blocks of equal length ss with 0<s<2⋅85m0 < s < 2 \cdot85^m [5].

First give the symbols 0,1,2,30,1,2,3 of vv the respective values 0,1,L,L20,1,L,L^2. Repeat this numerical version of vv indefinitely to obtain the periodic sequence bb. Consider two adjacent length-hh blocks of bb, where h=tL+sh=tL+s and 0≤s<L0 \le s < L. Each block sum contains the same contribution from tt full periods. The two remaining pieces of length ss begin ss positions apart modulo LL, so they are adjacent when read in the periodic repetition of vv.

If s>0s>0, let m1,m2,m3m_1,m_2,m_3 count the symbols 1,2,31,2,3 in one remaining piece. Its numerical sum is m1+Lm2+L2m3m_1+Lm_2+L^2m_3. Each count is less than LL, so these counts are exactly its base-LL digits. Equal sums would force equal counts of 1,2,31,2,3, and the common length ss would then force equal counts of 00. The theorem of Peltomäki and Whiteland excludes such a pair. Therefore equality of the two bb-sums implies s=0s=0, or L∣hL\mid h. For any ss, both remaining sums belong to [0,sL2][0,sL^2], so their difference has magnitude at most sL2<L3<UsL^2<L^3<U.

Next define an integer sequence sns_n, the signal, to record which zeros change value in (2). In the copy of vv encoding wiw_i, put sn=H+1s_n=H+1 at the first wiw_i zeros and sn=−1s_n=-1 at zeros numbered D+1,…,D+wiD+1,\ldots,D+w_i, counting the zeros from one. Put sn=0s_n=0 at all other positions. Then

Yn=1+bn+U(1+sn).Y_n=1+b_n+U(1+s_n).

Each complete copy contributes (H+1)wi−wi=Hwi(H+1)w_i-w_i=Hw_i to the signal sum. The sequence bb locates the block boundaries modulo LL, and the signal sums over complete copies recover the letters of ww.

Claim. The word YY of (2) contains no additive square.

Proof. An additive square in YY forces the difference of the two bb-sums to vanish modulo UU. Its magnitude is strictly less than UU, so it vanishes, and h=Leh=Le for a positive integer ee. Equality of the two YY-sums now also forces equality of the corresponding ss-sums. The cuts have a common offset rr in copies ii, i+ei+e, i+2ei+2e of vv. Suppose first that the first rr symbols of vv contain at most DD zeros. At this offset in copy jj, let αj\alpha_j count the positive signals already read. No negative signal has been read. The signal sum up to cut Lj+rL_j+r is therefore HPj+(H+1)αjHP_j+(H+1)\alpha_j. The three signal prefix sums have zero second difference. Since H+1≡1(modH)H+1\equiv1\pmod{H}, this implies αi−2αi+e+αi+2e≡0(modH)\alpha_i-2\alpha_{i+e}+\alpha_{i+2e}\equiv0\pmod{H}. Each αj\alpha_j belongs to [0,D][0,D], so the absolute value of this difference is at most 2D<H2D<H. It must vanish, leaving Δe2P(i)=0\Delta_e^2P(i)=0.

Suppose instead that the first rr symbols of vv contain more than DD zeros. In copy jj, all wjw_j positive signals have been read. Let tjt_j count the negative signals already read. The signal sum up to cut Lj+rL_j+r is now

HPj+(H+1)wj−tj=HPj+1+(wj−tj).HP_j+(H+1)w_j-t_j=HP_{j+1}+(w_j-t_j).

Again the remainder, wj−tjw_j-t_j, belongs to [0,D][0,D]. Its second difference is a multiple of HH with magnitude at most 2D<H2D<H, so it vanishes and Δe2P(i+1)=0\Delta_e^2P(i+1)=0. The two cases give a square in ww starting at ii or at i+1i+1, contrary to § 5. This proves the claim for YY.

Evaluating a specified position

The following procedure evaluates (2) without writing out a length-qq replacement block or the length-LL period.

  1. Compute (i,r)(i,r) by division of nn by LL. To tabulate AA, evaluate gg at q−1q-1 and at (t+1)8j−1(t+1)8^j-1 for 0≤j<200\leq j<20, 0≤t<70\leq t<7, and use the three allowed dd-values. Sort the distinct letters to obtain their ranks. Start at a∗a_* and read the digits of ii in base qq, most significant first. For each digit ℓ\ell, replace the current letter aa by gℓ+Mdℓ(a)g_\ell+M d_\ell(a). The result is wiw_i. This follows from wqj+ℓ=σ(wj)ℓw_{qj+\ell}=\sigma(w_j)_\ell.

  1. Let Cj(c)C_j(c) count the zeros in φj(c)\varphi^j(c). With zero substitutions the word is the single letter cc, so C0(c)=[c=0]C_0(c)=[c=0]. For higher levels use

Cj+1(c)=∑t=084Cj((pt+c) mod 4)(0≤j<11).C_{j+1}(c)=\sum_{t=0}^{84}C_j((p_t+c)\bmod4)\qquad(0\leq j<11).

Begin with the two blocks φ11(0)\varphi^{11}(0), φ11(1)\varphi^{11}(1). Skip the blocks preceding offset rr, subtracting their lengths and accumulating their zero counts. At each remaining level j>0j>0, split the current block φj(c)\varphi^j(c) into its 85 children φj−1((pt+c) mod 4)\varphi^{j-1}((p_t+c)\bmod4) of length 85j−185^{j-1} and repeat. The single symbol reached at level zero and the accumulated count are exactly vrv_r and kk in (2).

As checks on indexing, w0=243+320=8 799 579 806 609w_0=2^{43}+3^{20}=8\,799\,579\,806\,609 and the first occurrence of the integer 1 is at zero-based index 1,152,921,504,466,744,130, the position of the (D+1)(D+1)-st zero of the first copy of vv.

Optimality

Since an abelian square is an additive square, and since it is impossible to avoid abelian squares over an alphabet of size 3, the optimal alphabet size for an infinite word avoiding additive squares is of size at least 4. The construction in this paper uses an alphabet of size 6. Thus the optimal alphabet size for an additive-square-avoiding word is either 4, 5, or 6. It remains to determine which one is correct.

Statement of AI Usage

This paper was almost entirely written by GPT 6.2 Astra. I added some introductory and material, references, and discussion. I take full responsibility for the correctness.

A Lean verification of the construction, also constructed by GPT 6.2 Astra, is available at https://cs.uwaterloo.ca/~shallit/Papers/additive_square_lean-2.zip.

The main theorem is AddSquare.Six.six_letter_counterexample in AdditiveSquare.lean.

To reproduce the verification, unzip the project and run sh check.sh from the extracted directory.

References

  1. [1]T. C. Brown and A. R. Freedman, Arithmetic progressions in lacunary sets, Rocky Mountain Journal of Mathematics 17 (1987), 587–596. The conjecture appears on pp. 595–596. doi:10.1216/RMJ-1987-17-3-587.DOI
  2. [2]L. Halbeisen and N. Hungerbühler. An application of Van der Waerden’s theorem in additive number theory. INTEGERS: Elect. Journ. Comb. Number Theory 0 (2000), #A7 (electronic), http://www.integers-ejcnt.org/vol0.htmlDOI
  3. [3]J. Justin, Généralisation du théorème de van der Waerden sur les semi-groupes répétitifs, Journal of Combinatorial Theory, Series A 12 (1972), 357–367. doi:10.1016/0097-3165(72)90101-X.DOI
  4. [4]G. Pirillo and S. Varricchio, On uniformly repetitive semigroups, Semigroup Forum 49 (1994), 125–129. doi:10.1007/BF02573477.DOI
  5. [5]J. Peltomäki and M. A. Whiteland, Avoiding abelian powers cyclically, Advances in Applied Mathematics 121 (2020), 102095. doi:10.1016/j.aam.2020.102095. arXiv:2006.06307.DOI

Paper details

vlnc:W_rmcxqa3f
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents