Get ready Part 32

Calculus Level 2

Is it true that there is a constant C C such that, if p ( x ) p(x) is a polynomial of degree 1999, then p ( 0 ) C 1 1 p ( x ) d x \displaystyle |p(0)| \le C \int_{-1}^1 |p(x)| dx ?

False True

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

2 solutions

Théo Leblanc
May 21, 2020

The key observation :

Consider the complexe plane, η > ε > 0 \eta > \varepsilon > 0 and z 0 C z_0\in\mathbb{C} . Then, inf α z 0 η ( min z z 0 ε z α α ) = C ( z 0 , η , ε ) > 0 \underset{|\alpha - z_0|\geq \eta}{\inf} \left( \underset{|z-z_0|\leq \varepsilon}{\min} \left|\dfrac{z-\alpha}{\alpha}\right|\right) = C(z_0, \ \eta,\ \varepsilon) > 0

Proof the observation

Indeed, if α z 0 1 |\alpha - z_0|\gg 1 , then min z z 0 ε z α α 1 1 / 2 \underset{|z-z_0|\leq \varepsilon}{\min} \left|\dfrac{z-\alpha}{\alpha}\right| \approx 1 \geq 1/2 . Therefore we are bring back to a compact set : η α z 0 R \eta \leq |\alpha-z_0|\leq R for some R > η R>\eta (we call this compact set K K ).

But because α min z z 0 ε z α α \alpha \longmapsto \underset{|z-z_0|\leq \varepsilon}{\min} \left|\dfrac{z-\alpha}{\alpha}\right| is continuous, on the compact set K K , it is bounded and reaches its bounds. Now observes that for every α K \alpha\in K , min z z 0 ε z α α > 0 \underset{|z-z_0|\leq \varepsilon}{\min} \left|\dfrac{z-\alpha}{\alpha}\right| > 0 (In fact η ε R \geq \dfrac{\eta-\varepsilon}{R} )

Therefore, inf α K ( min z z 0 ε z α α ) = min α K ( min z z 0 ε z α α ) > 0 \underset{\alpha \in K }{\inf} \left( \underset{|z-z_0|\leq \varepsilon}{\min} \left|\dfrac{z-\alpha}{\alpha}\right|\right) = \underset{\alpha \in K}{\min} \left( \underset{|z-z_0|\leq \varepsilon}{\min} \left|\dfrac{z-\alpha}{\alpha}\right|\right) > 0

Which ends the proof of the observation.

Remark : the denpence on z 0 z_0 is hidden in R R .

\

Now we are ready to answer our problem. Let P P polynomial of degree 1999.

Without loss of generality (by homogeneity), we can suppose P ( 0 ) = 1 P(0)=1 .

Therefore P = k = 1 1999 X α k α k P=\displaystyle\prod_{k=1}^{1999} \dfrac{X-\alpha_k}{\alpha_k} where the α k \alpha_k 's are the complexe roots of P P .

Consider 2000 following disks : B ( 1 + 2 k 1 2000 , 1 2000 ) B(-1+\frac{2k-1}{2000}, \ \frac{1}{2000}) for k = 1 , 2 , , 2000 k=1, \ 2, \cdots, \ 2000 . Where B ( z , r ) B(z,r) is the disk center at z z with radius r r . The disks are disjoint and there are only 1999 roots so, at least one of them does not contain a root (let's say the m t h m^{th} ball). Call it center z m z_m and apply the key observation with ε = 1 8000 \varepsilon=\frac{1}{8000} and η = 1 4000 \eta=\frac{1}{4000} :

P ( x ) C ( z m , 1 4000 , 1 8000 ) 1999 x [ z m 1 8000 , z m + 1 8000 ] |P(x)| \geq C(z_m, \ \frac{1}{4000},\ \frac{1}{8000})^{1999} \hspace{1cm} \forall x\in [z_m-\frac{1}{8000}, \ z_m + \frac{1}{8000}]

So, P ( x ) min h = 1 , 2 , , 2000 C ( z h , 1 4000 , 1 8000 ) 1999 = Q ( 1 4000 , 1 8000 ) 1999 > 0 x [ z m 1 8000 , z m + 1 8000 ] |P(x)| \geq \underset{ h = 1, \ 2, \cdots, \ 2000}{\min} C(z_h, \ \frac{1}{4000},\ \frac{1}{8000})^{1999} = Q( \frac{1}{4000},\ \frac{1}{8000})^{1999} >0 \hspace{1cm} \forall x\in [z_m-\frac{1}{8000}, \ z_m + \frac{1}{8000}]

Now, Q ( 1 4000 , 1 8000 ) 1999 = : R 1999 Q( \frac{1}{4000},\ \frac{1}{8000})^{1999} =: R_{1999} no more depends on P P at all.

Finally, 1 1 P z m 1 8000 z m + 1 8000 P R 1999 4000 > 0 \displaystyle\int_{-1}^1 |P| \geq \displaystyle\int_{z_m-\frac{1}{8000}}^{z_m + \frac{1}{8000}} |P| \geq \dfrac{R_{1999} }{4000} > 0

So, P ( 0 ) = 1 4000 R 1999 1 1 P |P(0)|=1 \leq \dfrac{4000}{R_{1999}} \displaystyle\int_{-1}^1 |P|

Achal Jain
Jul 16, 2019

I maybe wrong but we just need to set the constant variable to be less than zero since the RHS and LHS are always positive(can be proven easily).

It does not work if C < 0 C<0 because positive >= negative, not the contrary.

Théo Leblanc - 1 year ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...