uniolympiads Ελληνικά

SEEMOUS 2022, Problem 3

Let α∈C∖{0}\alpha \in \mathbb{C} \setminus \{0\} and A∈Mn(C)A \in \mathcal{M}_n(\mathbb{C}), A≠OnA \neq O_n, be such that

A2+(A∗)2=αA⋅A∗,A^2 + (A^*)^2 = \alpha A \cdot A^*,

where A∗=(A‾)TA^* = \left(\overline{A}\right)^T. Prove that α∈R\alpha \in \mathbb{R}, ∣α∣⩽2|\alpha| \leqslant 2, and A⋅A∗=A∗⋅AA \cdot A^* = A^* \cdot A.

Original from the official PDFLet α ∈ C \ {0} and A ∈ M n (C), A 6 = O n , be such that A 2 + (A ∗ ) 2 = αA · A ∗ , where A ∗ = A T . Prove that α ∈ R, |α| ⩽ 2, and A · A ∗ = A ∗ · A.

This transcription was compared symbol by symbol with the original, but not yet by a person.

Time on this problem 0:00

Work for at least 45 minutes before opening the solution.

Official solution

Solution: Let A=(aij)1⩽i,j⩽nA = (a_{ij})_{1\leqslant i,j\leqslant n}. Applying the trace operator in the given identity, it follows that

∑i,j=1naij⋅aji+∑i,j=1naji‾⋅aij‾=α⋅∑i,j=1naij⋅aij‾,\sum_{i,j=1}^n a_{ij}\cdot a_{ji} + \sum_{i,j=1}^n \overline{a_{ji}}\cdot\overline{a_{ij}} = \alpha\cdot\sum_{i,j=1}^n a_{ij}\cdot\overline{a_{ij}},
hence
2Re⁡∑i,j=1naij⋅aji=α∑i,j=1n∣aij∣2⏟∈(0,∞),(1)2\operatorname{Re}\sum_{i,j=1}^n a_{ij}\cdot a_{ji} = \alpha\underbrace{\sum_{i,j=1}^n |a_{ij}|^2}_{\in(0,\infty)}, \tag{1}
which leads to α∈R\alpha\in\mathbb{R}.

Since

∣Re⁡xy∣≤∣x∣⋅∣y∣≤∣x∣2+∣y∣22for all x,y∈C,|\operatorname{Re} xy| \le |x|\cdot|y| \le \frac{|x|^2 + |y|^2}{2} \quad \text{for all } x, y\in\mathbb{C},
using (1) it follows that
∣α∣⋅∑i,j=1n∣aij∣2=2⋅∣Re⁡∑i,j=1naij⋅aji∣≤∑i,j=1n∣aij∣2+∑i,j=1n∣aji∣2=2∑i,j=1n∣aij∣2⏟>0,|\alpha|\cdot\sum_{i,j=1}^n |a_{ij}|^2 = 2\cdot\left|\operatorname{Re}\sum_{i,j=1}^n a_{ij}\cdot a_{ji}\right| \le \sum_{i,j=1}^n |a_{ij}|^2 + \sum_{i,j=1}^n |a_{ji}|^2 = 2\underbrace{\sum_{i,j=1}^n |a_{ij}|^2}_{>0},
hence ∣α∣⩽2|\alpha|\leqslant 2.

Let ε1,ε2\varepsilon_1, \varepsilon_2 be the solutions of z2−αz+1=0z^2 - \alpha z + 1 = 0, hence ε1+ε2=α\varepsilon_1 + \varepsilon_2 = \alpha and ε1ε2=1\varepsilon_1\varepsilon_2 = 1. Let X=A−ε1A∗X = A - \varepsilon_1 A^* and Y=A−ε2A∗Y = A - \varepsilon_2 A^*. Then

XY=A2+ε1ε2⏟=1(A∗)2−ε1A∗A−ε2AA∗=αAA∗−ε1A∗A−ε2AA∗=ε1(AA∗−A∗A)XY = A^2 + \underbrace{\varepsilon_1\varepsilon_2}_{=1}(A^*)^2 - \varepsilon_1 A^*A - \varepsilon_2 AA^* = \alpha AA^* - \varepsilon_1 A^*A - \varepsilon_2 AA^* = \varepsilon_1(AA^* - A^*A)
and, similarly,
YX=ε2(AA∗−A∗A).YX = \varepsilon_2(AA^* - A^*A).
Then XY=ε1ε2YX=ε12YXXY = \dfrac{\varepsilon_1}{\varepsilon_2}YX = \varepsilon_1^2 YX, so (XY)2=ε14(YX)2(XY)^2 = \varepsilon_1^4(YX)^2. Since Tr⁡((XY)2)=Tr⁡((YX)2)\operatorname{Tr}\left((XY)^2\right) = \operatorname{Tr}\left((YX)^2\right), it follows that
(ε14−1)Tr⁡((XY)2)=0,(\varepsilon_1^4 - 1)\operatorname{Tr}\left((XY)^2\right) = 0,
so we distinguish the following cases:

  • ε1∈{−i,i}\varepsilon_1\in\{-i, i\}. Then α=0\alpha = 0, which is a contradiction.
  • ε1∈{−1,1}\varepsilon_1\in\{-1, 1\}. Then α∈{−2,2}\alpha\in\{-2, 2\}, and the equality from the hypothesis becomes (A±A∗)2=±(A∗A−AA∗)(A\pm A^*)^2 = \pm(A^*A - AA^*). The equality of the traces gives Tr⁡((A±A∗)2)=0\operatorname{Tr}\left((A\pm A^*)^2\right) = 0, which leads to A±A∗=OnA\pm A^* = O_n, and the conclusion follows.
  • Tr⁡((XY)2)=0\operatorname{Tr}\left((XY)^2\right) = 0. Then Tr⁡((AA∗−A∗A)2)=0\operatorname{Tr}\left((AA^* - A^*A)^2\right) = 0, which leads to AA∗−A∗A=OnAA^* - A^*A = O_n.

Transcribed from the official solutions, pages 3–4. This transcription has not been checked a second time; if in doubt, look at the original.

Archive problem: the statement is the official text (transcribed, with the original above). Subject classification and difficulty were estimated automatically; hints and a worked solution are added over time.

How did it go?