MathGroup Archive 1995

[Date Index] [Thread Index] [Author Index]

Search the Archive

NDSolve problem

  • To: mathgroup at christensen.cybernetics.net
  • Subject: [mg900] NDSolve problem
  • From: pitts at mayo.EDU (Todd Pitts)
  • Date: Fri, 28 Apr 1995 00:58:01 -0400
  • Organization: Ultrasound Research Laboratory, Mayo Foundation, Rochester MN

I have been using NDSolve[] to investigate the solutions to some large,
sets of simple, ordinary differential equations.  It is most convenient
to use subscripted variables/functions and parameters to represent the large number
of solution functions and system characteristics.  For certain values
of coefficients the solution proceeds nicely, however, for others I obtain
the error

NDSolve::ndnum: Differential equation does not evaluate to a number at t = 0..

Below I have described the simplest system that I have been able to obtain
an error on.

 (* System equation *)
 eq={4.0*D[x[1][t],{t,2}]==x[1][t]-Sin[t]};

 (* Initial conditions *)
 bc={Evaluate[D[x[1][t],t] /. t->0]==0,x[1][0]==0};

(* Make a list of all system equations *)
 sys=Join[eq,bc];

(* Send the problem to NDSolve[] *)
 NDSolve[sys,x[1],{t,0,10}]

Yields the following output:

NDSolve::ndnum: Differential equation does not evaluate to a number at t = 0..

Out[32]= {{x[1] -> InterpolatingFunction[{0., 0.}, <>]}}
 
Changing the coefficient 4.0 in eq to unity (1 not 1.0) gives no error.  Also
writing the equation as

eq={D[x[1][t],{t,2}]==(x[1][t]-Sin[t])/4};

works.  If I don't use subscripted variables I don't have this problem.  For
example 

sys2 = sys /. x[1] -> x
NDSolve[sys2,x,{t,0,10}]

seems to work fine.  I would greatly appreciate any hints.  Is this a bug or
have I misunderstood the use of subscripted variables?  The version number is
2.2.  I get the same behavior on NeXT and DEC Alfa.

Thanks in Advance,

Todd Pitts





  • Prev by Date: Bivariate Normal Distribution
  • Next by Date: Re: Help using Compile with Solve
  • Previous by thread: Re: Bivariate Normal Distribution
  • Next by thread: Re: NDSolve problem