Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Complex Problem

  1. May 18, 2005 #1
    Hi all,,
    I have a problems on complex Analysis:
    Show that the equation
    z^4 + z + 5 = 0 has no solution in the set { z is a subset of C: modulus of z is less than 1}

    i tried doing it using Triangle inequality although i got it but i am looking for a better solution...Pls help
  2. jcsd
  3. May 18, 2005 #2


    User Avatar
    Science Advisor
    Homework Helper

    You could use Rouché's Theorem:

    If f,g are analytic functions on a simply connected region G and C is a contour in G such that |g(z)|<|f(z)| for all points z on C, then f+g has as many roots inside C as f.

    Using this, we note that for |z|=1 we have:|z^4+z|<=2<5.
    So the z^4+z+5 has as many roots inside the circle {z:|z|<1 } as 5. So that's clearly zero.
    Last edited: May 18, 2005
  4. May 18, 2005 #3


    User Avatar
    Homework Helper

    I don't know Rouche's theorem, my knowledge of complex analysis is not great. But here's how I would do it.

    Let [tex]z = r(\cos\theta + i\sin\theta)[/tex]

    where r is the real magnitude and theta the argument.

    Assume a solution for z with r < 1 exists.

    Then using De Moivre's theorem and splitting off the real part of the equation gives (I'm skipping a few steps because I'm a bit rushed, you can fill in the gaps) :

    [tex]r^4\cos{(4\theta)} + r\cos\theta = -5[/tex]

    Taking the magnitude of the LHS,

    [tex]|LHS| \leq |r^4\cos{(4\theta)}| + |r\cos\theta| < |r\cos{(4\theta)}| + |r\cos\theta|[/tex]

    since for nonnegative r < 1, [itex]r^4 < r[/itex]

    [tex]|r\cos{(4\theta)}| + |r\cos\theta| = r(|\cos{(4\theta)}| + |\cos\theta|) < |\cos{(4\theta)}| + |\cos\theta| < 2[/tex]

    using the boundedness of the trig ratios.

    So |LHS| < 2, but for the orig. eqn to be satisfied, |LHS| = 5, this is a contradiction (RAA, QED).
    Last edited: May 18, 2005
  5. May 18, 2005 #4
    Curious solution is quite simple
  6. May 18, 2005 #5
    Thx Curious ..
    Galileo can u illustrate more about urs solution...May be i can learn a new method to do such preoblems
  7. May 19, 2005 #6


    User Avatar
    Science Advisor
    Homework Helper

    I admit Rouché's theorem is overkill for this problem, but since you asked for different methods I posted it anyway.
    As Curious also showed, it can easily be done with the triangle inequality, simply because for |z|<1 we have |z^4+z|<2.

    I don't know any elementary way to prove Rouché's theorem. Only one using the principle of the argument involving contour integration. But I can show some nice illustrations:

    - Take [itex]P(z)=z^n+a_{n-1}z^{n-1}+...+a_1z+a_0[/itex]. Intuitively we would expect |z^n| to grow faster than [itex]|a_{n-1}z^{n-1}+...+a_1z+a_0|[/itex] (like the real counterpart). So let [itex]f(z)=z^n[/itex] and [itex]g(z)=a_{n-1}z^{n-1}+...+a_1z+a_0[/itex], so that P(z)=f(z)+g(z). If we take the circle |z|=R, then:


    [tex]|g(z)| \leq |a_{n-1}|R^{n-1}+|a_{n-2}|R^{n-2}+...+|a_1|R+|a_0|[/tex]


    [tex]\frac{|g(z)|}{|f(z)|}\leq \frac{|a_{n-1}|}{R}+\frac{|a_{n-2}|}{R^2}+\frac{|a_{n-3}|}{R^3}+...+\frac{|a_1|}{R^{n-1}}+\frac{|a_0|}{R^n}[/tex]

    The right lid goes to zero if [itex]R\to \infty[/itex], so for R large enough we have |g(z)|<|f(z)| on the circle |z|=R.
    Now Rouché's theorem says that P(z) has as many roots inside this circle as f(z)=z^n, which is n. This gives a nonstandard proof of the fundamental theorem of algebra.

    -You can approximate the annular region where roots are located. Take [itex]P(z)=z^4+3z+1[/itex]. Then for the circle |z|=1 we have [itex]|z^4+1|\leq 2<3=|3z|[/itex], so P(z) has as many roots inside |z|=1 as 3z. That is, it has 1 root inside this circle.
    For |z|=2 we have [itex]|3z+1|\leq 7 <16 =|z^4|[/itex], so there are 4 roots of P(z) (all of them) inside the circle |z|=2.
    Last edited: May 19, 2005
  8. May 19, 2005 #7
    Thx very much Galileo....i understood it well...it is very well explained.
  9. May 24, 2005 #8
    here is another question in understanding the proof of which i am completely Lost:
    Pls help me

    Let f(z)=Sigma(n=0 to infinity)a(base-n)z^n be a power series with radius of convergence R>0.Show that
    f is differentiable on U with deivative f'(z)=Sigma(n=1 to infinity)n a(base-n)z^n-1 and f' has radius of convergence of R.

    I will be highly thankful to you..
  10. May 24, 2005 #9

    matt grime

    User Avatar
    Science Advisor
    Homework Helper

    There is a famous result stating when you can interchange summation and differentiation. Try googling for it.
  11. May 24, 2005 #10
    by summation u mean here integration....actually i did not get the proper heading for googling...
    It all presented vague results
    Does interchange of limits take place here
  12. May 24, 2005 #11

    matt grime

    User Avatar
    Science Advisor
    Homework Helper

  13. May 24, 2005 #12
    thx for link Matt,,
    That i know that term by term differentiation gives the derivative but My teacher said that its natural to accept this fact but not obvious...There is a trivial Proof::
    The proof she did i could not grasp
  14. May 24, 2005 #13


    User Avatar
    Science Advisor
    Homework Helper

    path integration is obviously a basic tool. Hence basic properties of path integration are also fundamental.

    The first question is: over which paths does an integrand have the same value? this is basic for computing integrals by simplifying the path.

    The fundamental result is that if the differential form you are integrating is "closed", i.e. satisfies the mixed partials condition, [i.e. the "curl" is zero], then the the integral is the same over any two "homotopic" paths, i.e. paths that are deformable continuously into each other.

    all holomorphic functions and hence all polynomials f(z), define closed one forms f(z)dz.

    The hypothesis of rouche's "theorem" [more of a remark] guarantees existence of an obvious homotopy along straight lines between the two paths.

    More generally, the integral of a closed form is the same along any two 'homologous" paths, which is the commutative version of homotopy. e.g. any 1 form, closed or not, has integral zero over the path A*B *A^-1 *B^-1, even when this path is not contractible.

    Geometrically, and intuitively, two paths are homologous when their "difference" forms the boundary of some surface

    verbum sapienti: Learning this "homotopy" technique for computing path integrals will do you much more good in the long run, and short run, than solving the trivial problem at hand.

    remark: the correspondence between integrals of closed forms and boundary properties of loops, gives rise to "de rham" cohomology. I.e. a plane region has "holes" in it, if and only if there exist closed differential one forms with non zero integrals around some closed path in the region.

    thus the analytic invariant: {closed diff one forms}/{forms of type df}, measure the geometry - topology of the region.
    Last edited: May 24, 2005
  15. May 26, 2005 #14
    mathwonk thats cool
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook