Method error when using DifferentialEquations.jl Julia package

Question

I'm trying to solve an ode45 differential equation with the DifferentialEquation.jl package but I'm getting a method error.

using DifferentialEquations

M = 400; m = 35;
C = 3e3; c = 300;
K = 50e3; k = 200e3;
A = 0.05; L = 0.5; vh = 13.9

MM = [M 0;
      0 m] # mass matrix

CC = [C -C;
     -C C+c] # damping matrix

KK = [K -K;
     -K K+k] # stiffness matrix

w(t) = A*sin(2*pi*vh*t/L)
wd(t) = A*2*pi*vh*t/L*cos(2*pi*vh*t/L) # dw/dt

n = 2 # number of (original) equations
Mi = MM^(-1)
AA = [zeros(n,n) eye(n); -Mi*KK -Mi*CC]

f(t) = [0; c*wd(t)+k*w(t)] # force vector
g(t,y) = AA*y+[zeros(n,1); Mi*f(t)]

y0 = zeros(4) # initial conditions
tspan = (0.0, 0.5) # time span

prob = ODEProblem(g, y0, tspan) # defining the problem
solve(prob) 

The code gives an error that says:

MethodError: Cannot convert an object of type Array{Float64,2} to an object of type Array{Float64,1} This may have arisen from a call to the constructor Array{Float64,1}(...), since type constructors fall back to convert methods.

I do not get what I might be doing wrong, although i believe the error might have something to do with y0 (since typeof(y0) = Array{Float64,1}) and the error occures in the line where the solve() function is.

Thanks for any help beforehand!

Answer 1

Untested hunch: change:

g(t,y) = AA*y+[zeros(n,1); Mi*f(t)]

to:

g(t,y) = AA*y+[zeros(n); Mi*f(t)]

The former will create a two-dimensional matrix with one column. The latter will create a one-dimensional vector. The error message you're seeing is that it's getting a two dimensional array in a place where it expects a one-dimensional one.