21 bis. Some problems with complex numbers.

by Virgil Nicula, Apr 22, 2010, 4:20 PM

PP1 (OLM Bucuresti 2006 , Dinu Serbanescu). Let $\{x,y\}\subset\mathbb C$ with the property $\forall\ n\in\mathbb N^*$ we have $|x^n - y^n| = 1$ . Prove that $xy = 0\ .$

Proof. I"ll use twice the identity of the parallelogram $:\ \boxed{|u+v|^2+|u-v|^2=2\left(|u|^2+|v|^2\right)}\ (*)\ .$

$\odot\ \left\{\begin{array}{c}
 |x-y|=1\\\\
 \left|x^2-y^2\right|=1\end{array}\right\|\ \Longrightarrow\ |x+y|=1\ \implies\ |x-y|^2$ $+|x+y|^2=2\left(|x|^2+|y|^2\right)\ \Longrightarrow\ |x|^2+|y|^2=1\ (1)\ .$

$\odot\begin{array}{ccc}
 \nearrow & \left|x^2-y^2\right|=1 & \searrow\\\\
 \searrow & \left|x^4-y^4\right|=1 & \nearrow\end{array}\odot \Longrightarrow\ \left|x^2+y^2\right|=1\ \implies\ \left|x^2-y^2\right|^2+$ $\left|x^2+y^2\right|^2=2\left(|x|^4+|y|^4\right)\ \Longleftrightarrow\ |x|^4+|y|^4=1\ (2)\ .$

$\odot\ \ \left(|x|^2+|y|^2\right)^2=|x|^4+|y|^4+2|xy|^2\ \stackrel{(1)\wedge (2)}{\ \ \Longrightarrow\ \ }\ |xy|=0\ \Longrightarrow\ xy=0\ .$



PP2. Let $ z\in\mathbb C$ be a complex number for which exists $ n\in\mathbb N$ so that $ n\ge 2$ and $ 2^{n - 1}\cdot\left(z^n + 1\right) = (z + 1)^n$ . Prove that $ |z| = 1$ .

Proof. Observe that $ z\ne 0\ (z=0\implies n=1)$ . Suppose w.l.o.g. $ z\ne 1$ . Denote $ |z| = r$ . Thus, $ \overline z = \frac {r^2}{z}$ and $ 2^{n - 1}\cdot\left(z^n + 1\right) = (z + 1)^n$ $ \Longleftrightarrow$ $ 2^{n - 1}\cdot\left(\overline z^n + 1\right) = (\overline z + 1)^n$ $ \Longleftrightarrow$

$ 2^{n - 1}\cdot\left(r^{2n} + z^n\right) = \left(r^2 + z\right)^n$ . Suppose $ \mathrm {ad\ absurdum}$ that $ r\ne 1$ . Therefore, $ \left\{\begin{array}{c} 2^{n - 1}\cdot\left(z^n + 1\right) = (z + 1)^n \\
 \\
2^{n - 1}\cdot\left(r^{2n} + z^n\right) = \left(r^2 + z\right)^n\end{array}\ \right\|\ \boxed - \ \uparrow$ $ \implies$ $ 2^{n - 1}\cdot\left(r^{2n} - 1\right) = \left(r^2 + z\right)^n - (z + 1)^n$

$ \implies$ $ \frac {2^{n - 1}\cdot\left(r^{2n} - 1\right)}{r^2 - 1} = \left|\sum_{k = 1}^n\left(r^2 + z\right)^{n - k}(z + 1)^{k - 1}\right|\le$ $ \sum_{k = 1}^n\left(r^2 + r\right)^{n - k}(r + 1)^{k - 1} =$ $ (r + 1)^{n - 1}\cdot\sum_{k = 1}^nr^{n - k}$ $ \implies$ $ \frac {2^{n - 1}\cdot\left(r^{2n} - 1\right)}{r^2 - 1}\le (r + 1)^{n - 1}\cdot\frac {r^n - 1}{r - 1}$ $ \implies$

$ \boxed {\ 2^{n - 1}\cdot \left(r^n + 1\right)\le (r + 1)^{n}\ \ (*)\ }$ because $ \frac {r^n - 1}{r - 1} > 0$ . From the relation $ (*)$ and the well-known inequality $ 2^{n - 1}\cdot \left(r^n + 1\right)\ge (r + 1)^{n}$ obtain $ 2^{n - 1}\cdot \left(r^n + 1\right) = (r + 1)^{n}$ ,

i.e. $ r = 1$ , what is $ \mathrm {absurd}$ . In concluzion, $ 2^{n - 1}\cdot\left(z^n + 1\right) = (z + 1)^n$ $ \implies$ $ |z| = 1$ . Here is an equivalent enunciation :


An easy extension. Let $ \{u,v\}\subset \mathbb C$ be two complex numbers for which exists $ n\in\mathbb N$ so that $ n\ge 2$ and $ 2^{n - 1}\cdot\left(u^n + v^n\right) = (u+ v)^n$ . Prove that $ |u| = |v|$ .


PP3 (Stan Fulger). Construct outside of $\triangle ABC$ the equilateral $\left\{\begin{array}{ccc}
\triangle BCA'\ ; & \triangle CAB'\ ; & \triangle ABC'\\\\
\triangle B'C'A''\ ; & \triangle C'A'B''\ ; & \triangle A'B'C''\end{array}\right\|$

so that $BC$ separates $\{A,A'\}$ a.s.o. and $B'C'$ separates $\{A',A''\}$ a.s.o. Prove that $\left\{\begin{array}{c}
B\in B'B''\ ;\ BB'=BB''\\\\
O\in O'O''\ ;\ OO'=OO''\end{array}\right\|$


Proof. Let $X(x)$ be the point $X$ from the complex plane and which has the affix $x\in\mathbb C$ and $w=\cos\frac {\pi}3+i\sin\frac {\pi}3$ for which $\left\{\begin{array}{c}
w^3+1=0\ ;\ w^2-w+1=0\\\\
\overline w=\frac 1w=-w^2=1-w\end{array}\right\|$ .

Thus, $\left\{\begin{array}{ccc}
a'-b=w(c-b) & \implies & a'=(1-w)b+wc\\\\
b'-c=w(a-c) & \implies & b'=(1-w)c+wa\\\\
c'-a=w(b-a) & \implies & c'=(1-w)a+wb\end{array}\right\|$ $\implies$ $\left\{\begin{array}{ccccc}
a''-b'=w(c'-b') & \implies & a''=b'+w(c'-b') & \implies & a''=2a+w^2b-wc\\\\
b''-c'=w(a'-c') & \implies & b''=c'+w(a'-c') & \implies & b''=2b+w^2c-wa\\\\
c''-a'=w(b'-a') & \implies & c''=a'+w(b'-a') & \implies & c''=2c+w^2a-wb\end{array}\right\|$

and $\left\{\begin{array}{ccc}
3o=b+c+a'=b+c+[(1-w)b+wc] & \implies & 3o=(2-w)b+(1+w)c\\\\
3o'=a'+c'+b''=[(1-w)b+wc]+[(1-w)a+wb]+[2b+w^2c-wa] & \implies & 3o'=(1-2w)a+3b+(2w-1)c\\\\
3o''=a'+b'+c''=[(1-w)b+wc]+[(1-w)c+wa]+[2c+w^2a-wb] & \implies & 3o''=(2w-1)a+(1-2w)b+3c\end{array}\right\|$

Observe easily that $\left\{\begin{array}{ccccc}
b'+b''=[(1-w)c+wa]+[2b+w^2c-wa] & \implies & b'+b''=2c & \implies & B\in B'B''\ ;\ BB'=BB''\\\\
3(o'+o'')=2(2-w)b+2(1+w)c=3o & \implies & o'+o''=2o & \implies & O\in O'O''\ ;\ OO'=OO''\end{array}\right\|$ .

Remark. Prove easily that $w^2-w+1=0\implies$ for any $\{x,y\}\subset\mathbb C\ ,\ \boxed{[(1-w)x+wy][(1-w)u+wv]=-wxu+w^2yv+(xv+yu)}$ . For $u:=x\ ,\ v:=y$ get

$\boxed{[(1-w)x+wy]^2=-wx^2+w^2y^2+2xy}$ . With the well-known property $"\ ABC$ is equilateral $\iff \sum a^2=\sum bc\ "$ can prove easily that $A'B'C'$ is equilateral $\implies$

$ABC$ is equilateral. Suppose that $\boxed{\sum a'^2=\sum b'c'}\ (*)$ . Thus, $\sum a'^2=\sum \left[(1-w)b+wc\right]^2=$ $\sum \left[-wb^2+w^2c^2+2bc\right]=$ $\sum\left(w^2-w\right)a^2+2\sum bc\implies$

$\boxed{\sum a'^2=\sum \left(-a^2+2bc\right)}\ (1)$ . On other hand, $\sum b'c'=\sum [(1-w)c+wa][(1-w)a+wb]=$ $\sum \left[-wca+w^2ab+\left(a^2+bc\right)\right]=$ $\sum\left[\left(w^2-w+1\right)bc+a^2\right]\implies$

$\boxed{\sum b'c'=\sum a^2}\ (2)$ . In conclusion, the relations $(*)\wedge (1)\wedge (2)\implies$ $\sum a^2=\sum \left(-a^2+2bc\right)\implies$ $\sum a^2=\sum bc\implies$ the triangle $ABC$ is equilateral.



PP4. Let $ABCD$ , $EBFG$ be two squares for which denote the midpoints $M$ , $N$ , $L$ , $P$ of the segments $[BE]$ , $[CB]$ , $[DF]$ , $[AG]$ respectively. Prove that $MNLP$ is a square.

Proof. Prove easily that for $A(a)\ ,\ B(-ia)\ ,\ C(-a)\ ;\ D(ia)\ ($ can verify $A+C=B+D$ , i.e. $ABCD$ is a parallelogram and $C-B=i(A-B)$ , i.e. $BA\perp BC\ )$

obtain that $E(z)\ ,\ F[iz-(1+i)a]\ ,\ G[-a+(1+i)z]\ ($ can verify $E+F=B+G$ , i.e. $ABCD$ is a parallelogram and $F-B=i(E-B)$ , i.e. $BF\perp BE\ )$ .

Therefore, $M\left(\frac{z-ia}2\right)\ ,\ N\left[\frac{-(1+i)a}2\right]\ ,\ L\left(\frac{-a+iz}2\right)\ ,\ P\left[\frac{(1+i)z}2\right]$ . Observe that $ML=NP=(1+i)(z-a)$ , i.e. $MNLP$ is a parallelogram

and $N-M=i(P-M)=\frac {-(a+z)}2$ , i.e. $MN\perp MA$ . In conclusion, $MNLP$ is a square.



PP5. Let $ABC$ , $EBD$ be two equilateral triangles for which denote the midpoints $M$ , $N$ , $P$ of the segments $[BE]$ , $[CB]$ , $[AD]$ respectively and

the centroids $R$ , $S$ , $T$ of the triangles $EBD$ , $MNP$ , $ABC$ respectively. Prove that $MNP$ is an equilateral triangle and $S$ is the midpoint of $[RT]$ .


Proof. Prove easily that for $A(a)\ ,\ B(0)\ ,\ C(wa)\ ,\ E(z)$ obtain that $D(wz)\ ,\ M\left(\frac z2\right)\ ,\ N\left(\frac{wa}2\right)\ ,\ P\left(\frac{a+wz}2\right)$ , where $w^3+1=0=w^2-w+1$ . Prove easily that

$m^2+n^2+p^2=mn+np+pm$ , i.e. $\triangle MNP$ is equilaterally and $R\left[\frac {(1+w)z}{3}\right]$ , $S\left[\frac {(1+w)(a+z)}{6}\right]$ , $T\left[\frac {(1+w)a}{3}\right]$ , where $2S=R+T$ , i.e. $S$ is the midpoint of $[RT]$ .
This post has been edited 48 times. Last edited by Virgil Nicula, Nov 23, 2015, 6:57 AM

Comment

1 Comment

The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Nice ones!!

by Fermat_Theorem, Jan 12, 2019, 6:51 AM

Own problems or extensions/generalizations of some problems which was posted here.

avatar

Virgil Nicula
Archives
+ October 2017
+ September 2017
+ December 2016
+ October 2016
+ February 2016
+ September 2013
+ October 2010
+ September 2010
Shouts
Submit
  • orzzzzzzzzz

    by mathMagicOPS, Jan 9, 2025, 3:40 AM

  • this css is sus

    by ihatemath123, Aug 14, 2024, 1:53 AM

  • 391345 views moment

    by ryanbear, May 9, 2023, 6:10 AM

  • We need virgil nicula to return to aops, this blog is top 10 all time.

    by OlympusHero, Sep 14, 2022, 4:44 AM

  • :omighty: blog

    by tigerzhang, Aug 1, 2021, 12:02 AM

  • Amazing blog.

    by OlympusHero, May 13, 2021, 10:23 PM

  • the visits tho

    by GoogleNebula, Apr 14, 2021, 5:25 AM

  • Bro this blog is ripped

    by samrocksnature, Apr 14, 2021, 5:16 AM

  • Holy- Darn this is good. shame it's inactive now

    by the_mathmagician, Jan 17, 2021, 7:43 PM

  • godly blog. opopop

    by OlympusHero, Dec 30, 2020, 6:08 PM

  • long blog

    by MrMustache, Nov 11, 2020, 4:52 PM

  • 372554 views!

    by mrmath0720, Sep 28, 2020, 1:11 AM

  • wow... i am lost.

    369302 views!

    -piphi

    by piphi, Jun 10, 2020, 11:44 PM

  • That was a lot! But, really good solutions and format! Nice blog!!!! :)

    by CSPAL, May 27, 2020, 4:17 PM

  • impressive :D
    awesome. 358,000 visits?????

    by OlympusHero, May 14, 2020, 8:43 PM

72 shouts
Tags
About Owner
  • Posts: 7054
  • Joined: Jun 22, 2005
Blog Stats
  • Blog created: Apr 20, 2010
  • Total entries: 456
  • Total visits: 404396
  • Total comments: 37
Search Blog
a