Linear Diophantine equations

[17/82%]
Show
LaTeX

In how many different ways can you change one dollar? That is, in how many different ways can you pay 100 cents using five different kinds of coins, cents, nickels, dimes, quarters and half-dollars (worth 1, 5, 10, 25, and 50 cents, respectively)?

?

Let nn stand for a non-negative integer and let AnA_n denote the number of solutions of the Diophantine equation

x+5y+10z+25u+50v=n x + 5y + 10z + 25u + 50v = n

in non-negative integers. Then the series

A0+A1ζ+A2ζ2+⋯+Anζn+⋯ A_0 + A_1 \zeta + A_2 \zeta ^2 + \cdots + A_n \zeta ^n + \cdots

represents a rational function of ζ\zeta. Find it.

?
Remark.
?

For an intuitive solution see G. Pólya: Amer. Math. Monthly Vol. 63, pp. 689-697 (1956). Cf. MD, Vol. 1, p. 97, ex. 3.84.

Generalize the preceding examples by replacing the particular values of the coins, stamps and weights by a1,a2,…,ala_1, a_2, \ldots , a_l.

?
Remark.
?

Cf. Euler: Introductio in Analysin infinitorum, Chap. 16, De Partitione Numerorum; Opera Omnia, Ser. 1, Vol. 8, pp. 313-338. Leipzig and Berlin: B. G. Teubner 1922; also e.g. W. Ahrens: Mathematische Unterhaltungen und Spiele, 2nd Ed., Vol. 1, pp. 88-98, Vol. 2, p. 329. Leipzig: B. G. Teubner 1910, 1918.

The total number of non-negative integral solutions of the following Diophantine equations is n+1n+1:

x+2y=n,2x+3y=n−1,3x+4y=n−2,…, x+2y=n, \quad 2x+3y=n-1, \quad 3x+4y=n-2, \ldots , nx+(n+1)y=1,(n+1)x+(n+2)y=0. nx+(n+1)y=1, \quad (n+1)x+(n+2)y=0 .
?
Remark.
?

E. Catalan, Problem: Mathésis (Gand) Vol. 2, p. 158 (1882). Solved by E. Cesàro: Mathésis (Gand) Vol. 3, p. 87 (1883).

The total number NN of non-negative integral solutions of the following Diophantine equations

x+2y=n−1,2x+3y=n−3,3x+4y=n−5,… x+2y=n-1, \quad 2x+3y=n-3, \quad 3x+4y=n-5, \ldots

is smaller than n+2n+2; moreover the difference n+2−Nn+2-N is equal to the number of divisors of n+2n+2 (cf. Part VIII, Chapter 1, §5).

?
Remark.
?

E. Catalan: Nouv. Annls Math. Ser. 3, Vol. 1, p. 528 (1882). Solved by E. Cesàro: Nouv. Annls Math. Ser. 3, Vol. 2, p. 380 (1883).

Prove that the total number of non-negative integral solutions of the following Diophantine equations is nn:

x+4y=3n−1,4x+9y=5n−4,9x+16y=7n−9,… x+4y=3n-1, \quad 4x+9y=5n-4, \quad 9x+16y=7n-9, \ldots
?
Remark.
?

E. Cesàro, Problem: Mathésis (Gand) Vol. 2, p. 208 (1882).

The number of non-negative solutions of the Diophantine equation

x+2y+3z=n x+2y+3z=n

is equal to the integer closest to (n+3)212\frac{(n+3)^2}{12}.

?
Remark.
?

Cf. G. H. Hardy: Some Famous Problems of the Theory of Numbers and in Particular Waring's Problem. Oxford 1920, pp. 9, 10.

Let aa, bb and nn be positive integers, aa and bb relatively prime to each other. The number of non-negative integral solutions of the equation

ax+by=n ax+by=n

is equal to [nab]\left[\frac{n}{ab}\right] or [nab]+1\left[\frac{n}{ab}\right]+1.

?
Remark.
?

More may be less: to prove a more general or a more precise theorem may be less trouble. [Proposition by P. Paoli; cf. Intermédiaire des mathématiciens Vol. 1, p. 247-248 (1894). Ch. Hermite, Problem: Nouv. Annls Math. Ser. 1, Vol. 17, p. 32 (1858). Solved by L. Rassicod: Nouv. Annls Math. Ser. 1, Vol. 17, p. 126-130 (1858).]

Let n≥3n \geq 3. The number of positive integral solutions of

x+y+z=n x+y+z=n

that satisfy the additional conditions

x≤y+z,y≤z+x,z≤x+y x \leq y+z, \quad y \leq z+x, \quad z \leq x+y

is given by

(n+8)(n−2)8 for n even,n2−18 for n odd. \frac{(n+8)(n-2)}{8} \text{ for } n \text{ even}, \qquad \frac{n^2-1}{8} \text{ for } n \text{ odd} .
?
Remark.
?

Cf. Ch. Hermite, Problem: Nouv. Annls Math. Ser. 2, Vol. 7, p. 335 (1868). Solved by V. Schlegel: Nouv. Annls Math. Ser. 2, Vol. 8, p. 91 (1869).

Prove that if a square integer system of linear equations has a unique solution modulo every prime number pp, then it also has a unique solution over the ring of integers.

?

Determine whether a square integer system of linear equations that is consistent modulo every prime number pp is necessarily consistent over the ring of integers.

?

Two sets of unknowns are called integrally equivalent if they are related by

[y1⋮yn]=U[x1⋮xn], \left[\begin{smallmatrix} y_{1} \\ \vdots \\ y_{n} \end{smallmatrix}\right]=U\left[\begin{smallmatrix} x_{1} \\ \vdots \\ x_{n} \end{smallmatrix}\right],

where UU is an integer matrix whose determinant is equal, in absolute value, to 1. Prove that the system of equations

∑j=1naijxj=bi,i=1,2,…,m \sum _{j=1}^{n} a_{i j} x_{j}=b_{i}, \quad i=1,2, \ldots , m

where aij,bia_{i j}, b_{i} are integers, is equivalent, over the ring of integers, to a system of equations of the form

diyi=ci,i=1,2,…,m d_{i} y_{i}=c_{i}, \quad i=1,2, \ldots , m

where the set of unknowns (y1,…,yn)\left(y_{1}, \ldots , y_{n}\right) is integrally equivalent to the set (x1,…,xn)\left(x_{1}, \ldots , x_{n}\right).

?

Prove that an integer system of equations has an integer solution if and only if, for every natural number kk, the greatest common divisors of all minors of order kk in the matrix of the system and in its augmented matrix coincide.

?

Prove that an integer system of equations has an integer solution if and only if it has a solution modulo every prime pp.

?

Justify the following practical method for finding all integer solutions of a system of equations

∑j=1naijxj=bi,i=1,2,…,m \sum _{j=1}^{n} a_{i j} x_{j}=b_{i}, \quad i=1,2, \ldots , m

with integer coefficients. Construct the matrix (AbEn0)\left(\begin{array}{cc}A & b \\ E_{n} & 0\end{array}\right) of size (n+m)×(n+1)(n+m) \times (n+1). Then, using only integer elementary transformations of the first mm rows and nn columns, reduce this matrix to the form (DCU0)\left(\begin{array}{cc}D & C \\ U & 0\end{array}\right), where C=(c1⋮cm),∣det⁡U∣=1,D=(d10…00…00d2…00…0…………………00…dr0…000…00…0…………………00…00…0)C=\left(\begin{array}{c}c_{1} \\ \vdots \\ c_{m}\end{array}\right), \quad \left|\operatorname {det} U\right|=1, \quad D=\left(\begin{array}{ccccccc}d_{1} & 0 & \ldots & 0 & 0 & \ldots & 0 \\ 0 & d_{2} & \ldots & 0 & 0 & \ldots & 0 \\ \ldots & \ldots & \ldots & \ldots & \ldots & \ldots & \ldots \\ 0 & 0 & \ldots & d_{r} & 0 & \ldots & 0 \\ 0 & 0 & \ldots & 0 & 0 & \ldots & 0 \\ \ldots & \ldots & \ldots & \ldots & \ldots & \ldots & \ldots \\ 0 & 0 & \ldots & 0 & 0 & \ldots & 0\end{array}\right), di∣di+1,d1≠0,…,dr≠0d_{i} \mid d_{i+1}, \quad d_{1} \neq 0, \ldots , d_{r} \neq 0. Then the system is consistent if

di∣ci for i=1,…,r,ck=0 for k>r, \begin{gathered} d_{i} \mid c_{i} \quad \text{ for } \quad i=1, \ldots , r, \\ c_{k}=0 \quad \text{ for } \quad k>r, \end{gathered}

and the general solution is given by the formula

[x1⋮xn]=U[c1/d1⋮cr/dryr+1⋮yn] \left[\begin{smallmatrix} x_{1} \\ \vdots \\ x_{n} \end{smallmatrix}\right]=U\left[\begin{smallmatrix} c_{1} / d_{1} \\ \vdots \\ c_{r} / d_{r} \\ y_{r+1} \\ \vdots \\ y_{n} \end{smallmatrix}\right]

where yr+1,yr+2,…,yny_{r+1}, y_{r+2}, \ldots , y_{n} are arbitrary integers.

?

Find all integer solutions of the following systems of equations:

?
(a)

2x1+3x2+4x3=52 x_{1}+3 x_{2}+4 x_{3}=5;

(b)

{2x1+3x2−11x3−15x4=1,4x1−6x2+2x3+3x4=2,2x1−3x2+5x3+7x4=1.\begin{cases} 2 x_{1}+3 x_{2}-11 x_{3}-15 x_{4}=1, \\ 4 x_{1}-6 x_{2}+2 x_{3}+3 x_{4}=2, \\ 2 x_{1}-3 x_{2}+5 x_{3}+7 x_{4}=1. \end{cases}

?
(i)

Let x1,x2,x3∈Zx_{1}, x_{2}, x_{3} \in \mathbb {Z}. Find all solutions of the system of linear equations

7x1+5x2−5x3=8,17x1+10x2−15x3=−42. 7 x_{1}+5 x_{2}-5 x_{3} = 8, \qquad 17 x_{1}+10 x_{2}-15 x_{3} = -42.
(ii)

Find all positive solutions.

References

[PóSz98] Pólya G., Szegő G. Problems and Theorems in Analysis. Springer, 1998.
[Кос15] Кострикин А.И. Сборник задач по алгебре. МЦНМО, 2015.
[StHa17] Steeb W., Hardy Y. Problems and Solutions in Introductory and Advanced Matrix Calculus. 2nd ed., World Scientific Publishing, 2017.