Abstract
For integers k, n with k, \(n \geqslant 1\), the n-color weak Schur number \(W\hspace{-0.6mm}S_{k}(n)\) is defined as the least integer N, such that for every n-coloring of the integer interval [1, N], there exists a monochromatic solution \(x_{1},\dots , x_{k}, x_{k+1}\) in that interval to the equation:
with \(x_{i} \ne x_{j}\), when \(i\ne j.\) In this paper, we obtain the exact values of \(WS_{6}(2)=166\), \(WS_{7}(2)=253\), \(WS_{3}(3)=94\) and \(WS_{4}(3)=259\) and we show new lower bounds on n-color weak Schur number \(W\hspace{-0.6mm}S_{k}(n)\) for \(n=2,3.\)
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
For integers \(a \leqslant b\), we shall denote [a, b] the integer interval consisting of all \(t \in \mathbb {N}_+=\{1,2,\dots \}\) such that \(a \leqslant t \leqslant b\). A function
where \(c_1,\dots ,c_n\in \mathbb {N}_+\) represent different colors, is an n-coloring of the interval [1, N].
Given an n-coloring \(\Delta \) and the equation \( x_{1}+ \dots +x_{k}=x_{k+1}\) in \(k+1\) variables, we say that a solution \( x_{1},\dots ,x_{k},x_{k+1} \) to the equation is monochromatic if and only if \(\Delta (x_{1})=\Delta (x_{2})=\dots =\Delta (x_{k+1})\).
For integers k, n with k, \(n \geqslant 1\), the n-color weak Schur number \(W\hspace{-0.6mm}S_{k}(n)\) is defined as the least integer N, such that for every n-coloring of the integer interval [1, N], there exists a monochromatic solution \(x_{1},\dots , x_{k}, x_{k+1}\) in that interval to the equation: \( x_{1}+x_{2}+\dots +x_{k} =x_{k+1},\) with \(x_{i} \ne x_{j}\) when \(i\ne j.\) Irwing [14] showed the existence and obtained the following general upper bound:
For \(k=2\), we have \(1+315^{\frac{n-1}{5}}\leqslant W\hspace{-0.6mm}S_{2}(n) \leqslant [n!ne]+1\), the lower bound is due to Sierpinski [20] and the upper bound to Bornsztein [5].
1.1 Schur numbers and weak Schur numbers
A set A of integers is called sum-free if it contains no elements \(x_{1},x_{2},x_{3}\in A\) satisfying \( x_{1}+x_{2}=x_{3}\) where \(x_{1}, x_{2}\) need not be distinct.
Schur [19] in 1916 proved that, given a positive integer n, there exists a greatest positive integer \(S_2(n)=N\) with the property that the integer interval \([1,N-1]\) can be partitioned into n sum-free sets. The numbers \(S_{2}(n)\) are called Schur numbers. The current knowledge on these numbers for \(1 \le n \le 7\) is given in Table 1.
Many generalizations of Schur numbers have appeared since their introduction. Now, a set A of integers is called weakly sum-free if it contains no pairwise distinct elements \(x_{1},x_{2},x_{3}\in A\) satisfying \( x_{1}+x_{2}=x_{3}.\) We denote by \(W\hspace{-0.6mm}S_{2}(n)\), the greatest integer N, for which the integer interval \([1,N-1]\). The exact value of \(S_{2}(4)\) was given by Baumert [2] and recently \(S_{2}(5)\) has been obtained by Heule [13]. Finally, the lower bounds on \(S_{2}(6)\) and \(S_{2}(7)\) were obtained by Fredricksen and Sweet [11] by considering symmetric sum-free partitions. A set A of integers is said to be k-sum-free if it contains no \(k+1\) elements \(x_{1},x_{2}, \cdots ,x_{k+1} \in A\) satisfying \(x_{1}+\cdots +x_{k}=x_{k+1}\), where \(x_{i}\), \(i=1,\cdots , k\) are not necessarily distinct. In 1933, Rado [15] gave the following generalization:
given two positive integers, n and \(k \geqslant 2\), there exists a greatest positive integer, \(S_{k}(n)=N\), such that the integer interval \([1,N-1]\) can be partitioned into n sets which are k-sum-free.
In 1966, Znám [22] established a lower bound on the numbers \(S_{k}(n)\):
In 1982, Beutelspacher and Brestovansky [3] proved the equality for two k-sum-free sets:
In 2010 [18], the last author obtained the exact value of \(S_{3}(3)=43.\) Independently, Ahmed and Schaal [1] in 2016 gave the values of \(S_{k}(3)\) for \(k = 3,4,5.\) In 2019, Boza et al. [6] determined the exact formula of \(S_{k}(3)=k^{3}+2k^{2}-2\) for all \(k\geqslant 3\), finding an upper bound that coincides with the lower bound given by Znám [22].
The numbers \(W\hspace{-0.6mm}S_{2}(n)\) are called the weak Schur numbers for the equation \( x_{1}+x_{2}=x_{3}.\) The known weak Schur numbers are given in Table 2.
The current state of knowledge concerning \(W\hspace{-0.6mm}S_{2}(n)\) is a bit confusing. The problem seems to have been first considered in [21], which is Walker’s solution to Problem E985 proposed a year earlier, in 1951, by Moser. Walker considered the cases \(n = 3, 4\) and 5 and claimed the values \(W\hspace{-0.6mm}S_2(3) =24\), \(W\hspace{-0.6mm}S_2(4) = 67\), and \(W\hspace{-0.6mm}S_2(5) = 197\). Unfortunately, the short account written by Moser on Walker’s solution only gives suitable partitions of [1, 23] for \(n = 3\), and no details at all for the cases \(n = 4\) and 5. Walker’s claimed values of \(W\hspace{-0.6mm}S_2(3)\) and \(W\hspace{-0.6mm}S_2(4)\) were later confirmed by Blanchard, Harary, and Reis using computers [4]. In 2012, the two last authors et al. [9] confirmed the lower bound \(W\hspace{-0.6mm}S_{2}(5) \geqslant 197\). In addition, a lower bound on \(W\hspace{-0.6mm}S_{2}(6)\) was obtained in [9] and later improved to \(W\hspace{-0.6mm}S_{2}(6) \geqslant 583\) in [10]. The lower bounds for \(7\leqslant n \leqslant 9\) were obtained [17] in 2015.
In terms of coloring, the \(W\hspace{-0.6mm}S_{k}(n)\) is the least positive integer N such that for every n-coloring of [1, N],
where \(c_1,\dots , c_n\) represent n different colors, there exists a monochromatic solution to the equation \(x_{1}+\dots +x_{k} =x_{k+1} \), such that \( \Delta (x_{1})= \dots =\Delta (x_{k})=\Delta (x_{k+1})\) where \(x_{i} \ne x_{j}\) when \(i\ne j.\)
In addition, for 2-coloring, the known weak Schur numbers \(W\hspace{-0.6mm}S_{k}(2)\) are shown in Table 3.
The exact values of \(WS_{k}(2)\) for \(k=3,4\) and the lower bounds were obtained in [7, 18] and \(WS_{5}(2)\) [8] in 2017.
1.2 Main results
In Section 2, we determine a general lower bound on the 2-color weak Schur numbers for the equation \(x_{1} + \dots + x_{k} = x_{k+1}\), with \(x_{i} \ne x_{j}\) when \(i\ne j\), for \(k\geqslant 5,\) improving the lower bound given in [7].
Lemma 2.1\(W\hspace{-0.6mm}S_{k}(2)\geqslant \frac{1}{2}(k^{3}+4k^{2}-5k+2)\) for any integer \(k\geqslant 5.\)
In Section 3, we determine a general lower bound on \(WS_{k}(3)\) improving the lower bound given in [7].
Lemma 3.1\(W\hspace{-0.6mm}S_{k}(3)\geqslant \frac{1}{2}(k^{4}+5k^{3}-k^{2}-9k+6)\) for any integer \(k\geqslant 5.\)
Lemma3.2\(W\hspace{-0.6mm}S_{k}(3)\geqslant \frac{1}{2}(k^{4}+5k^{3}-8k+4)\) for any integer \(k\geqslant 8.\)
In Section 4, we obtain the exact values of the 2-color weak Schur number \(WS_{6}(2)\) and \(WS_{7}(2)\). In addition, we determinate the exact values of the 3-color weak Schur numbers \(WS_{3}(3)\) and \(WS_{4}(3)\).
Theorem 4.2 \(W\hspace{-0.6mm}S_{6}(2)=166.\)
Theorem 4.6 \(W\hspace{-0.6mm}S_{7}(2)=253.\)
Theorem 4.9 \(W\hspace{-0.6mm}S_{3}(3)=94.\)
Theorem 4.12 \(W\hspace{-0.6mm}S_{4}(3)=259.\)
2 A general lower bound for \(W\hspace{-0.6mm}S_k(2) \)
In terms of coloring, the weak Schur number \(WS_{k}(2)\) is the least positive integer N such that for every 2-coloring of [1, N],
where \(c_1,c_2\) represent 2 different colors, there exists a monochromatic solution to the equation \(x_{1}+x_{2}+\dots +x_{k} =x_{k+1} \), such that \( \Delta (x_{1})= \dots = \Delta (x_{k})=\Delta (x_{k+1})\) where \(x_{i} \ne x_{j}\) when \(i\ne j.\)
In [7], a general lower bound of the weak Schur number \(WS_{k}(2)\) was given, now we show a new general lower bound that improves the previous one.
Lemma 2.1
For any integer \(k\geqslant 5\), we have
Proof
Let \(\Delta \) be a 2-coloring:
where \(c_1,c_2\) represent 2 different colors. Let \(A_{i}=\Delta ^{-1}(c_{i})\) for \(i=1,2,\) such that
where
We show that the above partition of the interval \([1,\frac{1}{2}(k^{3}+4k^{2}-5k)]\) has no monochromatic solution to the equation \(x_{1}+x_{2}+\dots +x_{k} =x_{k+1}\). For that, it is sufficient to prove that for every i, \(1 \leqslant i \leqslant k\), if \(x_1,\dots , x_k\in A_i\) with \(x_{i} < x_{j}\) when \(i< j,\) then \(x_1+\dots +x_k\notin A_i\).
-
If \(x_1,\dots ,x_k\in A_1\), then
$$\begin{aligned} \sum ^{k}_{i=1} x_i\geqslant & {} 1+\sum ^{k-2}_{i=0}(\frac{1}{2}(k^{2}+3k)+i) =\frac{1}{2}(k^{3}+3k^{2}-6k+4) \\> & {} \frac{1}{2}(k^{3}+3k^{2}-6k+2) \end{aligned}$$Hence \(\sum ^{k}_{i=1} x_i \not \in A_1.\)
-
If \(x_1,\dots ,x_k\in A_2\), then
-
If \(x_{k} \leqslant \frac{1}{2}(k^{2}+3k-2)\), then
$$\begin{aligned} \sum ^{k}_{i=1} x_i \geqslant \sum ^{k-1}_{i=0}(2+i)= \frac{1}{2}(k^{2}+3k)>\frac{1}{2} (k^{2}+3k-2). \end{aligned}$$In addition, for \(k\geqslant 5\),
$$\begin{aligned} \sum ^{k}_{i=1} x_i\leqslant & {} \sum ^{k-1}_{i=0} (\frac{1}{2}(k^{2}+3k-2)-i)\\= & {} \frac{1}{2}(k^{3}+2k^{2}-k)< \frac{1}{2}(k^{3}+3k^{2}-6k+4) \end{aligned}$$Hence \(\sum ^{k}_{i=1} x_i\not \in A_2.\)
-
If \(x_{k} \geqslant \frac{1}{2}(k^{3}+3k^{2}-6k+4)\), then
$$\begin{aligned} \sum ^{k}_{i=1} x_i\geqslant & {} \sum ^{k-2}_{i=0}(2+i)+ \frac{1}{2}(k^{3}+3k^{2}-6k+4) \\= & {} \frac{1}{2}(k^{3}+4k^{2}-5k+2) > \frac{1}{2}(k^{3}+4k^{2}-5k) \end{aligned}$$Hence \(\sum ^{k}_{i=1} x_i\not \in A_2.\)
-
Therefore, we obtain the lower bound. \(\square \)
With this general lower bound, we improve the results shown in Table 3. In addition, in the Section 4, we will prove that these new lower bounds shown in Table 4 for \(k=6\) and \(k=7\), are exact values.
3 A lower bound for \(W\hspace{-0.6mm}S_k(3)\)
Applying the result given in [7], the lower bounds shown in Table 5 were obtained.
In the next result, we improve the general lower bound of \(W\hspace{-0.6mm}S_k(3)\) obtained in [7].
Lemma 3.1
For any integer \(k\geqslant 5\), we have
Proof
We will show that he following partition of the interval
has no monochromatic solution to the equation \(x_{1}+x_{2}+\dots +x_{k} =x_{k+1}\) with \(x_1<x_2<\ldots <x_k\). Consider the following 3-coloring where \(A_1\) and \(A_2\) are the same as used in the construction of the 2-coloring in Lemma 2.1.
Since the above 3-coloring is an extension of 2-coloring given by Lemma 2.1, we just have to try the following cases:
-
Let \(x_1,x_2,\dots ,x_k\in B_1\), with \(x_1<x_2<\dots <x_k\).
-
If \(x_{k} \in A_1\), by Lemma 2.1, \(\sum ^{k}_{i=1} x_i \not \in A_1.\) Therefore,
$$\begin{aligned} \sum ^{k}_{i=1} x_i\leqslant & {} \sum ^{k-1}_{i=0}(\frac{1}{2}(k^{3}+3k^{2}-6k+2)-i) \\= & {} \frac{1}{2}(k^4+3k^{3}-7k^2+3k) \\< & {} \frac{1}{2}(k^4+4k^{3}-3k^2+2k-2). \end{aligned}$$Hence, \(\sum ^{k}_{i=1} x_i\not \in B_1.\)
-
If \(x_{k} \geqslant \frac{1}{2}(k^4+4k^{3}-3k^2+2k-2)\), then
$$\begin{aligned} \sum ^{k}_{i=1} x_i\geqslant & {} 1+ \sum ^{k-3}_{i=0} (\frac{1}{2}(k^{2}+3k)+i)+ \frac{1}{2}(k^4+4k^{3}-3k^2+2k-2) \\= & {} \frac{1}{2}(k^4+5k^{3}-k^{2}-9k+6) \\> & {} \frac{1}{2}(k^4+5k^{3}-k^{2}-9k+4). \end{aligned}$$Hence, \(\sum ^{k}_{i=1} x_i\not \in B_1.\)
-
-
Let \(x_1,\dots ,x_k\in B_2\), with \(x_1<\dots <x_k\).
-
If \(x_{k} \in A_2\), then by Lemma 2.1, \(\sum ^{k}_{i=1} x_i \not \in A_2.\) Therefore,
$$\begin{aligned} \sum ^{k}_{i=1} x_i \leqslant \sum ^{k-1}_{i=0}(\frac{1}{2}(k^{3}+4k^{2}-5k)-i)= & {} \frac{1}{2}(k^4+4k^{3}-6k^2+k) \\< & {} \frac{1}{2}(k^4+4k^{3}-4k^2+k). \end{aligned}$$Hence, \(\sum ^{k}_{i=1} x_i\not \in B_2.\)
-
If \(x_{k} \geqslant \frac{1}{2}(k^4+4k^{3}-4k^2+k)\), then
$$\begin{aligned} \sum ^{k}_{i=1} x_i\geqslant & {} \sum ^{k-2}_{i=0}(2+i)+ \frac{1}{2}(k^4+4k^{3}-4k^2+k) \\= & {} \frac{1}{2}(k^4+4k^{3}-3k^{2}+2k-2)\\> & {} \frac{1}{2}(k^4+4k^{3}-3k^{2}+2k-4). \end{aligned}$$Hence, \(\sum ^{k}_{i=1} x_i\not \in B_2.\)
-
-
Let \(x_1,\dots ,x_k\in B_3\), with \(x_1<\dots <x_k\). Then
$$\begin{aligned} \sum ^{k}_{i=1} x_i\geqslant & {} \sum ^{k-1}_{i=0}(\frac{1}{2}(k^{3}+4k^{2}-5k+2)+i)\\= & {} \frac{1}{2}(k^4+4k^{3}-4k^{2}+k) \\> & {} \frac{1}{2}(k^4+4k^{3}-4k^{2}+k-2). \end{aligned}$$Hence, \(\sum ^{k}_{i=1} x_i \not \in B_3.\)
Therefore, we obtain the desired lower bound.
With this general lower bound,we improve the results shown in Table 5. In addition, in Section 4, we will prove that these new lower bounds shown in Table 6 for \(k=3\) and \(k=4\) are exact values.
In the next result, we improve the lower bounds of Lemma 3.1 for any integer \(k\geqslant 8.\)
Lemma 3.2
For any integer \(k\geqslant 8\), we have
Proof
The following partition of the interval
has no monochromatic solution to the equation \(x_{1}+x_{2}+\dots +x_{k}=x_{k+1}\) with \(x_1<x_2<\ldots <x_k\).
This 3-coloring is an extension of 3-coloring given in Lemma 3.1, so we just have to try the following cases:
-
Let \(x_1,x_2,\dots ,x_k\in C_1\), with \(x_1<x_2<\dots <x_k\), by Lemma 3.1, \(\sum ^{k}_{i=1} x_i \notin B_1=C_1.\)
-
Let \(x_1,x_2,\dots ,x_k\in C_2\), with \(x_1<x_2<\dots <x_k\), by Lemma 3.1, \(\sum ^{k}_{i=1} x_i \notin B_2.\) We only need to prove that
$$\begin{aligned} \sum ^{k}_{i=1} x_i \notin [\frac{1}{2}(k^{4}+5k^{3}-k^{2}-9k+6),\frac{1}{2}(k^{4}+5k^{3}-8k+2)]. \end{aligned}$$We consider four cases:
-
If \(x_{k} \leqslant \frac{1}{2}(k^3+4k^2-5k)\), then
$$\begin{aligned} \sum ^{k}_{i=1} x_i\leqslant & {} \sum ^{k-1}_{i=0}\left( \frac{1}{2}(k^{3}+4k^{2}-5k)-i\right) =\frac{1}{2}(k^4+4k^{3}-6k^2+k)\\< & {} \frac{1}{2}(k^{4}+5k^{3}-k^{2}-9k+6). \end{aligned}$$Hence \(\sum ^{k}_{i=1} x_i\not \in C_2.\)
-
If \(x_{k} \in [\frac{1}{2}(k^4+4k^{3}-4k^2+k),\frac{1}{2}(k^4+4k^{3}-3k^2+2k-4)] \) and \(x_{k-1} \leqslant \frac{1}{2}(k^2+3k-2)\), then for \(k\geqslant 8\),
$$\begin{aligned} \sum ^{k}_{i=1} x_i\leqslant & {} \sum ^{k-2}_{i=0}(\frac{1}{2}(k^{3}+3k-2)-i)+\frac{1}{2}(k^4+4k^{3}-3k^2+2k-4)\\= & {} \frac{1}{2}(k^4+5k^{3}-2k^2-4) < \frac{1}{2}(k^4+5k^{3}-k^2-9k+6). \end{aligned}$$Hence \(\sum ^{k}_{i=1} x_i\not \in C_2.\)
-
If \(x_{k} \in [\frac{1}{2}(k^4+4k^{3}-4k^2+k),\frac{1}{2}(k^4+4k^{3}-3k^2+2k-4)] \) and \(x_{k-1}\geqslant \frac{1}{2}(k^{3}+3k^2-6k+4)\), then
$$\begin{aligned} \sum ^{k}_{i=1} x_i\geqslant & {} \sum ^{k-3}_{i=0}(2+i)+ \frac{1}{2}(k^{3}+3k^2-6k+4) +\\{} & {} \frac{1}{2}(k^4+4k^{3}-4k^2+k) \\= & {} \frac{1}{2}(k^4+5k^{3}-6k+2)> \frac{1}{2}(k^4+5k^{3}-8k+2). \end{aligned}$$Hence \(\sum ^{k}_{i=1} x_i\not \in C_2.\)
-
If \(x_{k} \geqslant \frac{1}{2}(k^4+5k^{3}-k^2-9k+6)\), then
$$\begin{aligned} \sum ^{k}_{i=1} x_i\geqslant & {} \sum ^{k-2}_{i=0}(2+i)+ \frac{1}{2}(k^4+5k^{3}-k^2-9k+6)\\= & {} \frac{1}{2}(k^4+5k^{3}-8k+4)> \frac{1}{2}(k^4+5k^{3}-8k+2) \end{aligned}$$Hence, \(\sum ^{k}_{i=1} x_i\not \in C_2.\)
-
-
Let \(x_1,\dots ,x_k\in C_3\), with \(x_1<\dots <x_k\), by Lemma 3.1, \(\sum ^{k}_{i=1} x_i \notin B_3=C_3.\)
Therefore,we obtain the desired improved lower bound.
Applying Lemmas 3.1 and 3.2, the following lower bounds are shown in Table 7.
4 Computer-assisted proofs for the exact values of \(WS_{6}(2)\), \(WS_{7}(2)\), \(WS_{3}(3)\) and \(WS_{4}(3)\)
4.1 The exact value of \(WS_{6}(2)\)
We shall prove that \(WS_{6}(2)=166\). By Lemma 2.1,we have \(W\hspace{-0.6mm}S_{6}(2)\geqslant 166.\)
To prove that the equation \(x_{1}+\dots +x_{6}=x_{7}\) has a monochromatic solution for every 2-coloring of the integer interval [1, 166], it is necessary to show the following result.
Lemma 4.1
The set \({\mathcal Y}=\{y_{n}\}^{42}_{n=1} =\{1,2,3,4,5,6,7,8,9,10,11,12,13,14\), \(16,18,21,22,23,26,27,28,29,30,31,35,36,\,40,\,41,\,46,48,51,\,56,61,66,\,106,\) \(141, 146,151,156,161,166\}\) satisfies:
-
1.
We have \({\mathcal Y} \subseteq [1,166].\)
-
2.
For every partition of \({\mathcal Y}\) into two subsets \(A_1, A_2\), some \(A_i\) contains a monochromatic solution of \(x_{1}+x_2+x_3+x_4+x_5+x_{6}=x_{7}\), with \(x_i \ne x_j\), if \(i \ne j.\)
Proof
-
1.
This is trivial.
-
2.
We have checked the result transforming the problem into a Boolean satisfiability problem and solving it with a SAT solver [12].
Let \(\Delta \) be a 2-coloring of [1, 166]:
For any \(\{y_{n}\} \in {\mathcal Y}\), we consider a Boolean variable \(\phi \) defined on [1, 42] as follows:
Let \({\mathcal S} = \{(y_a,y_b,y_c,y_d,y_e,y_f,y_g) \in {\mathcal Y}^7 \mid y_a+y_b+y_c+y_d+y_e+y_f=y_g, \text{ with } \, \, a<b<c<d<e<f \}.\)
For any \(s=(y_a,y_b,y_c,y_d,y_e,y_f,y_g)\in {\mathcal S}\), we consider two clauses:
Then, p(s) is satisfiable if and only if \(\Delta (a)\ne c_1\), \(\Delta (b)\ne c_1\), \(\Delta (c)\ne c_1\), \(\Delta (d)\ne c_1\), \(\Delta (e)\ne c_1\), \(\Delta (f)\ne c_1\) or \(\Delta (g)\ne c_1\), i.e., \(\Delta \) does not induce in s a monochromatic solution on \(c_1\) of the equation \(x_{1}+\dots +x_{6}=x_{7}.\) Analogously, q(s) is satisfiable if and only if \(\Delta \) does not induce in s a monochromatic solution of the equation \(x_{1}+\dots +x_{6}=x_{7}\) on \(c_2\).
Clearly \({\mathcal {C}}\) is satisfiable if and only if \(\Delta \) does not induce on \({\mathcal Y}\) a monochromatic solution of the equation \(x_{1}+\dots +x_{6}=x_{7}.\) The SAT-Solver shows that \({\mathcal {C}}\) is not satisfiable, hence for every 2-coloring of the set,\({\mathcal Y}\) has a monochromatic solution to the equation \(x_{1}+\dots +x_{6}=x_{7}.\) \(\square \)
With this result, we have tested the upper bound on \(W\hspace{-0.6mm}S_{6}(2)\).
Therefore, we conclude with the following result:
Theorem 4.2
\(W\hspace{-0.6mm}S_{6}(2)= 166.\)
4.2 The exact value of \(WS_{7}(2)\)
We shall prove that \(WS_{7}(2)=253\). By Lemma 2.1, we have \(W\hspace{-0.6mm}S_{7}(2)\geqslant 253.\)
We have to prove that the equation \(x_{1}+\dots +x_{7}=x_{8}\) has a monochromatic solution for every 2-coloring of the interval [1, 253]. We will suppose the opposite: for every 2-coloring \(\Delta : [1,253] \longrightarrow \{c_1,c_2\}\) without monochromatic solution, we can consider without loss of generality \(\Delta (61)=c_1\). Let \(D_1=\{u_{n}\}^{73}_{n=1}=(6[0,42]+\{1\})\cup (6\{0,1,4,8,16,32\}+\{2, 3, 4, 5, 6\})\subset [1,253]\) and \(F_1=\{43\), 49, 55, 61, 67, 73, 79, 85, 91, 97, 103, 109, 115, 133, 139, 145, 151, 157, 163, 169, \(175\}\subset D_1\).
The following two lemmas can be proved transforming the problem into a Boolean satisfiability problem and solving it with a SAT solver [12].
Lemma 4.3
For every 2-coloring \(\Delta \) of \(D_1\) without monochromatic solution, we have \(\Delta (F_1)=\{c_1\}.\)
Proof
Let \(\Delta \) be a 2-coloring of \(D_1\):
such that \(\Delta (61)=c_1\). For any \(\{u_{n}\} \in D_1\), we consider a Boolean variable \(\phi \) defined on [1, 73] as follows:
Let \({\mathcal S} = \{(u_a,u_b,u_c,u_d,u_e,u_f,u_g,u_h) \in D_1^8 \mid u_a+u_b+u_c+u_d+u_e+u_f+u_g=u_h, \text{ with } \, \, a<b<c<d<e<f<g \}.\)
For any \(s=(u_a,u_b,u_c,u_d,u_e,u_f,u_g,u_h)\in {\mathcal S}\), we consider two clauses:
Then, p(s) is satisfiable if and only if \(\Delta (a)\ne c_1\), \(\Delta (b)\ne c_1\), \(\Delta (c)\ne c_1\), \(\Delta (d)\ne c_1\), \(\Delta (e)\ne c_1\), \(\Delta (f)\ne c_1\), or \(\Delta (g)\ne c_1\) or \(\Delta (h)\ne c_1\), i.e., \(\Delta \) does not induce in s a monochromatic solution on \(c_1\) of the equation \(x_{1}+\dots +x_{7}=x_{8}.\) Analogously, q(s) satisfiable if and only if \(\Delta \) does not induce in s a monochromatic solution of the equation \(x_{1}+\dots +x_{7}=x_{8}\) on \(c_2\).
Clearly \({\mathcal {C}}\) is satisfiable if and only if \(\Delta \) does not induce on \(D_1\) a monochromatic solution of the equation \(x_{1}+\dots +x_{7}=x_{8}.\) The SAT-Solver shows that \({\mathcal {C}}\) is not satisfiable, hence we have the result. \(\square \)
Trivially we have,
Corollary 4.4
For every 2-coloring \(\Delta \) of [1, 253] without monochromatic solution, we have \(\Delta (F_1)=\{c_1\}.\)
Let \(D_2=\{v_{n}\}^{78}_{n=1}=(6[0,42]+\{1\})\cup (6\{0,1,2,5,6,36,37\}+\{2, 3, 4, 5, 6\})\). We have \([1,8]\cup F_1\cup \{217\}\subset D_2\subset [1,253]\).
Lemma 4.5
For every 2-coloring \(\Delta \) of \(D_2\) without monochromatic solution such that \(\Delta (F_1)=\{c_1\}\), we have \(\Delta (217)=c_1\) and \(\Delta ([1,8])=\{c_2\}.\)
Proof
Let \(\Delta \) be a 2-coloring of \(D_1\):
such that \(\Delta (F_1)=\{c_1\}\). For any \(\{v_{n}\} \in D_2\), we consider a Boolean variable \(\phi \) defined on [1, 78] as follows:
Let \({\mathcal S} = \{(u_a,u_b,u_c,u_d,u_e,u_f,u_g,u_h) \in D_2^8 \mid u_a+u_b+u_c+u_d+u_e+u_f+u_g=u_h, \text{ with } \, \, a<b<c<d<e<f<g \}.\)
For any \(s=(u_a,u_b,u_c,u_d,u_e,u_f,u_g,u_h)\in {\mathcal S}\), we consider two clauses:
Then, p(s) is satisfiable if and only if \(\Delta (a)\ne c_1\), \(\Delta (b)\ne c_1\), \(\Delta (c)\ne c_1\), \(\Delta (d)\ne c_1\), \(\Delta (e)\ne c_1\), \(\Delta (f)\ne c_1\), or \(\Delta (g)\ne c_1\) or \(\Delta (h)\ne c_1\), i.e., \(\Delta \) does not induce in s a monochromatic solution on \(c_1\) of the equation \(x_{1}+\dots +x_{7}=x_{8}.\) Analogously, q(s) satisfiable if and only if \(\Delta \) does not induce in s a monochromatic solution of the equation \(x_{1}+\dots +x_{7}=x_{8}\) on \(c_2\).
Clearly \({\mathcal {C}}\) is satisfiable if and only if \(\Delta \) does not induce on \(D_2\) a monochromatic solution of the equation \(x_{1}+\dots +x_{7}=x_{8}.\) The SAT-Solver shows that \({\mathcal {C}}\) is not satisfiable, hence we have the result. \(\square \)
Now, we can prove:
Theorem 4.6
\(WS_7(2)=253\).
Proof
Let \(\Delta \) be a 2-coloring of [1, 253] without monochromatic solution. Then \(\Delta (217)=c_1\) and \(\Delta ([1,8])=\{c_2\}.\) Therefore, \(\sum ^{8}_{i=1} i =36-n\) with \(i\ne n\), which implies \(\Delta ([28,34])=\{c_1\}\) and \(217=28+29+30+31+32+33+34\). Therefore \(\Delta (217)\ne c_1\), contradicting Lemma 4.5. \(\square \)
4.3 The exact value of \(WS_{3}(3)\)
The weak Schur number \(WS_{3}(3)\) is the least positive integer N such that for every 3-coloring of [1, N],
where \(c_1,c_2,c_3\) represent 3 different colors, there exists a monochromatic solution to the equation \(x_{1}+x_{2}+x_{3} =x_{4} \), such that \( \Delta (x_{1})= \dots =\Delta (x_{3})=\Delta (x_{4})\) where \(x_{i} \ne x_{j}\) when \(i\ne j.\)
We shall prove that \(WS_{3}(3)=94\). Let us first show a lower bound.
Lemma 4.7
\(W\hspace{-0.6mm}S_{3}(3)\geqslant 94.\)
Proof
It is easy to verify that the 3-coloring
defined by
has no monochromatic solution to the equation \(x_{1}+x_{2}+x_{3}=x_{4}\) such that \(x_{i} \ne x_{j}\) when \(i\ne j\). \(\square \)
To prove that the equation \(x_{1}+x_{2}+x_{3}=x_{4}\) has a monochromatic solution for every 3-coloring of the integer interval [1, 94], it is necessary to prove the following result.
Lemma 4.8
The set \({\mathcal Y}=\{y_{n}\}^{51}_{n=1} =\{1,2,3,4,5,6,7,8,9,10,11,12,13,14\), 15, 16, 17, 18, 19, 20, 21, 23, 24, 25, 26, 27, 28, 29, 31, 32, 33, 34, 35, 36, 38, 39, 40, \(42,44,45,52,58,64,65,66,72,75,78,82,91,94\}\) satisfies:
-
1.
We have \({\mathcal Y} \subseteq [1,94].\)
-
2.
For every partition of \({\mathcal Y}\) into three subsets \(A_1, A_2, A_3\), some \(A_i\) contains a monochromatic solution of \(x_1+x_2+x_3=x_4\), \(x_i \ne x_j\), with \(i \ne j.\)
Proof
-
1.
This is trivial.
-
2.
We have checked the result transforming the problem into a Boolean satisfiability problem and solving it with a SAT solver [12].
Let \(\Delta \) be a 3-coloring of [1, 94]:
For any \(\{y_{n}\} \in {\mathcal Y}\), we consider two Boolean variables \(\phi \) and \(\psi \) defined on [1, 51] as follows:
Thus, for any \(n\in [1,51]\) we have that \(\phi (n)\) is True or \(\psi (n)\) is True. Therefore, \({\mathcal {D}}=\bigwedge _{1\le n\le 51}(\phi (n)\vee \psi (n))\) is satisfiable.
Let \({\mathcal S} = \{(y_a,y_b,y_c,y_d) \in {\mathcal Y}^4 \mid y_a+y_b+y_c=y_d, \text{ with } \, \, a<b<c \}.\)
For any \(s=(y_a,y_b,y_c,y_d)\in {\mathcal S}\), we consider three clauses:
Then, p(s) is satisfiable if and only if \(\Delta (a)\ne c_1\), \(\Delta (b)\ne c_1\), \(\Delta (c)\ne c_1\) or \(\Delta (d)\ne c_1\), i.e., \(\Delta \) does not induce in s a monochromatic solution on \(c_1\) of the equation \(x_1+x_2+x_3=x_4\). Analogously, q(s) or r(s) is satisfiable if and only if \(\Delta \) does not induce in s a monochromatic solution of the equation \(x_1+x_2+x_3=x_4\) on \(c_2\) or \(c_3\), respectively.
Clearly \({\mathcal {D}}\wedge {\mathcal {C}}\) is satisfiable if and only if \(\Delta \) does not induce on \({\mathcal Y}\) a monochromatic solution of the equation \(x_1+x_2+x_3=x_4\). The SAT-Solver shows that \({\mathcal {D}}\wedge {\mathcal {C}}\) is not satisfiable, hence \(WS_3(3)\leqslant 94\).
With this result, we have tested the upper bound on \(W\hspace{-0.6mm}S_{3}(3)\).
Therefore, we conclude with the following result:
Theorem 4.9
\(W\hspace{-0.6mm}S_{3}(3)= 94.\)
4.4 The exact value of \(WS_{4}(3)\)
The weak Schur number \(WS_{4}(3)\) is the least positive integer N such that for every 3-coloring of [1, N],
where \(c_1,c_2,c_3\) represent 3 different colors, there exists a monochromatic solution to the equation \(x_{1}+x_{2}+\dots +x_{4} =x_{5} \), such that \( \Delta (x_{1})= \dots =\Delta (x_{4})=\Delta (x_{5})\) where \(x_{i} \ne x_{j}\) when \(i\ne j.\)
We shall prove that \(WS_{4}(3)=259\). Let us first show a lower bound.
Lemma 4.10
\(W\hspace{-0.6mm}S_{4}(3)\geqslant 259.\)
Proof
It is easy to verify that the 3-coloring
defined by \(\Delta (x)= \left\{ \begin{array}{ll} c_{1} &{} \text{ if } \, x\in [1,9]\cup [46,51]\cup [214,219]\cup [253,258] \\ c_{2} &{} \text{ if }\, x\in [10,45]\cup [220,252] \\ c_{3} &{} \text{ if }\, x\in [52,213] \end{array} \right. \)
has no monochromatic solution to the equation \(x_{1}+x_{2}+x_{3}+x_{4}=x_{5}\) such that \(x_{i} \ne x_{j}\) when \(i\ne j\). \(\square \)
To prove that the equation \(x_{1}+x_{2}+x_{3}+x_{4}=x_{5}\) has a monochromatic solution for every 3-coloring of the integer interval [1, 259], it is necessary to prove the following result.
Lemma 4.11
The set \({\mathcal {Z}}=\{z_{n}\}^{86}_{n=1} =\{1,2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14\), 15, 16, 17, 18, 19, 20, 21, 22, 23, 25, 26, 27, 28, 30, 31, 34, 42, 43, 44, 46, 47, 49, 52, 53, 54, 55, 56, 57, 58, 59, 61, 64, 65, 66, 67, 68, 70, 73, 74, 76, 78, 79, 85, 86, 88, 91, 99, 103, 106, 109, 115, 118, 124, 130, 139, 169, 190, 199, 202, 208, 211, 214, 217, \(220,223,226,229,235,238,250,253,259\}\) satisfies:
-
1.
We have \({\mathcal {Z}} \subseteq [1,259].\)
-
2.
For every partition of \({\mathcal {Z}}\) into three subsets \(A_1, A_2, A_3\), some \(A_i\) contains a monochromatic solution of \(x_{1}+x_{2}+x_{3}+x_{4}=x_{5}\), \(x_i \ne x_j\), with \(i \ne j.\)
Proof
-
1.
This is trivial.
-
2.
We have checked the result transforming the problem into a Boolean satisfiability problem and solving it with a SAT solver [12].
Let \(\Delta \) be a 3-coloring of [1, 259]:
For any \(\{z_{n}\} \in {\mathcal {Z}}\), we consider two Boolean variables \(\phi \) and \(\psi \) defined on [1, 86] as follows:
Thus, for any \(n\in [1,86]\),we have that \(\phi (n)\) is True or \(\psi (n)\) is True. Therefore, \({\mathcal {D}}=\bigwedge _{1\le n\le 86}(\phi (n)\vee \psi (n))\) is satisfiable.
Let \({\mathcal S} = \{(z_a,z_b,z_c,z_d,z_e) \in {\mathcal {Z}}^5 \mid z_a+z_b+z_c+z_d=z_e, \text{ with } \, \, a<b<c<d \}.\)
For any \(s=(z_a,z_b,z_c,z_d,z_e)\in {\mathcal S}\), we consider three clauses:
Then, p(s) is satisfiable if and only if \(\Delta (a)\ne c_1\), \(\Delta (b)\ne c_1\), \(\Delta (c)\ne c_1\), \(\Delta (d)\ne c_1\) or \(\Delta (e)\ne c_1\), i.e., \(\Delta \) does not induce in s a monochromatic solution on \(c_1\) of the equation \(x_1+x_2+x_3+x_4=x_5\). Analogously, q(s) or r(s) is satisfiable if and only if \(\Delta \) does not induce in s a monochromatic solution of the equation \(x_1+x_2+x_3+x_4=x_5\) on \(c_2\) or \(c_3\), respectively.
Clearly \({\mathcal {D}}\wedge {\mathcal {C}}\) is satisfiable if and only if \(\Delta \) does not induce on \({\mathcal {Z}}\) a monochromatic solution of the equation \(x_1+x_2+x_3+x_4=x_5\). The SAT-Solver shows that \({\mathcal {D}}\wedge {\mathcal {C}}\) is not satisfiable, hence \(WS_4(3)\leqslant 259\). \(\square \)
With this result, we have verified the upper bound for \(W\hspace{-0.6mm}S_{4}(3)\).
Therefore, we obtain the following result:
Theorem 4.12
\(W\hspace{-0.6mm}S_{4}(3)= 259.\)
References
Ahmed, T., Schaal, D.J.: On generalized Schur numbers. Exp. Math. 25, 213–218 (2015)
Baumert, L.D.: Sum-free sets. J.P.L. Res. Summary 36–10, 16–18 (1961)
Beutelspacher, A., Brestovansky, W.: Generalized Schur Number. Lecture Note Math, pp. 30–38. Springer, New York (1982)
Blanchard, P.F., Harary, F., Reis, R.: Partitions into sum-free sets. Electron. J. Combin. Numbers Theory 6, 1–10 (2006)
Bornsztein, P.: On an extension of a theorem of Schur. Acta Arith. 101(4), 395–399 (2002)
Boza, L., Marín, J.M., Revuelta, M.P., Sanz, M.I.: \(3\)- color Schur number. Discrete Appl. Math. 263, 59–68 (2019)
Boza, L., Revuelta, M.P., Sanz, M.I.: A general lower bound on the weak Schur number. Electron. Notes Discrete Math. 68, 137–142 (2018)
Boza, L., Marín, J.M., Revuelta, M.P., Sanz, M.I.: On the \(n\)-color weak Rado numbers for the equation \(x_{1}+x_{2}+\dots + x_{k}+ c=x_{k+1}\). Exp. Math. 28(2), 194–208 (2019)
Eliahou, S., Marín, J.M., Revuelta, M.P., Sanz, M.I.: Weak Schur numbers and the search for G.W. Walker’s lost partitions. Comput. Math. Appl. 63, 175–182 (2012)
Eliahou, S., Fonlupt, C., Fromentin, J., Marion-Poty, V., Robilliard, D., Teytaud, F.: Investigating Monte-Carlo methods on the weak Schur problem. Lect. Notes Comput. Sci. 7832, 191–201 (2013)
Fredricksen, H., Sweet, M.: Symmetric sum-free partitions and lower bounds for Schur numbers. Electron. J. Comb. 7, #R32 (2000)
Heule, M.: http://www.st.ewi.tudelft.nl/sat/, March RW. SAT Competition (2011)
Heule, M.J.H.: Schur Number Five. In: The Thirty-Second AAAI Conference on Artificial Intelligence (AAAI-18) (2018)
Irwing, R.W.: An extension of Schur’ s theorem on sum-free partitions. Acta Arith. XXV, 55–64 (1973)
Rado, R.: Studien zur Kombinatorik. Math. Z. 36, 424–480 (1933)
Rado, R.: Some solved and unsolved problems in the theory of numbers. Math. Gaz. 25, 72–77 (1941)
Rafilipojaona, F.A.: Nombres de Schur classiques et faibles. Ph. Thesis, Université Lille Nord-de-France (2015)
Sanz, M.I.: Números de Schur y Rado. Ph. Thesis, Universidad de Sevilla (2010)
Schur, I.: Uber die Kongruenz \(x^{m}+y^{m}\equiv z^{m}(\)mod \(p)\). Jber. Deutsch. Math.- Verein. 25, 114–117 (1916)
Sierpinski, W.: Theory of Numbers, part 2. PWN Warszawa, 440 (1959)
Walker, G.W.: A problem in partitioning. Am. Math. Mon. 59, 253 (1952)
Znám, S.: Generalization of a number—theoretical result. Mat-Fyz.Casopis. Sloven Akad Vied 17, 297–307 (1966)
Funding
Funding for open access publishing: Universidad de Sevilla/CBUA
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Ahmed, T., Boza, L., Revuelta, M.P. et al. Exact values and lower bounds on the n-color weak Schur numbers for \(n=2,3\). Ramanujan J 62, 347–363 (2023). https://doi.org/10.1007/s11139-023-00760-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11139-023-00760-y