
Economics Economics Forum  Financial Mathematics, Econometrics, Operations Research, Mathematical Finance, Computational Finance 
 LinkBack  Thread Tools  Display Modes 
August 10th, 2010, 10:00 AM  #1 
Newbie Joined: Sep 2008 Posts: 6 Thanks: 0  Equilibrium
I have got a continuous and ntimes differentiable function f(s) defined on the interval (0,1/2) in which f(1/2)=0. f(0) and f’(1/2), which depend on the parameters’ sign, might be either sign, whilst f’’(1/2)=0. Questions: Is there any way to know if the function displays other zeros in the specified interval? If yes, is there any way to know how many those zeros are? Thanks to everybody. Best Regards, Francesco 
August 10th, 2010, 09:36 PM  #2 
Global Moderator Joined: May 2007 Posts: 6,641 Thanks: 625  Re: Equilibrium
Let f(s)=cos(?ns), with n odd. Then the number of zeros depends on n  you can have as many as you want.

August 11th, 2010, 05:47 AM  #3 
Newbie Joined: Sep 2008 Posts: 6 Thanks: 0  Re: Equilibrium
Dear Mathman, Thanks a lot for your answer. Actually I am not looking for a class of function with many zeros, on the contrary I would like to have a sufficient and necessary condition such that there are no many zeros, or at most 1. Thanks Francesco 
August 11th, 2010, 02:26 PM  #4 
Global Moderator Joined: May 2007 Posts: 6,641 Thanks: 625  Re: Equilibrium
As you can see from the example I gave, you need more conditions than can be prescribed on the end points alone.

September 27th, 2010, 03:49 PM  #5 
Member Joined: Sep 2010 Posts: 63 Thanks: 0  Re: Equilibrium
Use the Nth derivative to compute the error term for a Taylor series of order (N1?). If the magnitude of this error term over the subinterval is smaller then your distance from zero at the point which you computed the Taylor series in this subinterval then you know their can't be a root in that subinterval. If you wish to consider the complex plane then use a two dimensional Taylor series (one for the real and imaginary part and rule out the possibility of roots in given sub neighborhoods by bounding the magnate of the Nth derivative over that region. To search quickly for places near where roots might be their is an algorithm MATLAB uses which finds all roots of a polynomial by computing the eigenvalues of a characteristic matrix. Edit: Following considerations for complex roots: Also note that we can compute the magnitude of a polynomial by multiplying it by the complex conjugate. The complex conjugate part is given by replacing z with xiy and expanding. The non conjugate part can be computed by replace z with x+iy. The product of these the product of two resulting 2D polynomial which gives the square of the magnitude. Subtracting the magnitude by the largest value of the error term in this region will give points where their may be enough error for their to be a root at or beyond this point. However, this gives a point and not a region. To construct a region consider the subset of two domains. The first domain is the smallest enclosing polynomial curve where the imaginary part is equal to zero. The second region is the smallest region where the real part is equal to zero. The roots are vertexes where these two curves cross. The region can be drawn by drawing piece wise polynomial curves between these vertexes. 

Tags 
equilibrium 
Thread Tools  
Display Modes  

Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Equilibrium Points  arron1990  Calculus  4  June 5th, 2012 01:44 AM 
Thermal Equilibrium  rnck  Physics  3  June 15th, 2011 07:55 AM 
Equilibrium  Evil Tomato  Physics  2  December 8th, 2010 02:35 PM 
Equilibrium  Francesco Mureddu  Calculus  1  August 24th, 2010 06:09 PM 
equilibrium of triangle  mathsiseverything  Algebra  2  March 5th, 2008 05:03 PM 