0
$\begingroup$

I have integro-differential equations like

$X'(t)=\operatorname{diag}(\int_0^t X(t) dt)AX(t)-X(t)$,

where $A$ is a known $N\times N$ matrix and $X$ is a $N\times 1$ vector, and $\operatorname{diag}$ means constructing a diagonal matrix from a vector.

I have no idea about how to represent a vector variable in Mathematica and how to program to solve the equations.

$\endgroup$
  • $\begingroup$ As Jim Baldwin said this question needs more clarity and examples. For example, knowing -- or imposing -- the shape of the matrix $X$ and function forms in the matrix $X$ can make the equation easy to program. What is the definition of "$diag$" -- it gives a diagonal matrix or a vector of the diagonal? What is the definition of the "$*$" operator -- standard matrix (and matrix-vector) multiplication, matrix element-wise multiplication? $\endgroup$ – Anton Antonov May 28 '17 at 16:19
1
$\begingroup$

This answer does not produce solutions of OP's equation, but it does show how to deal with a similar matrix equation.

NDSolve can handle equations given with matrices, but in this answer I am showing the generation of a list of equations corresponding to the original matrix one.

First we introduce the functions to be found:

n = 4;
funcs = Table[ToExpression["f" <> ToString[i]], {i, n}];
X = Through[funcs[t]]

(* {f1[t], f2[t], f3[t], f4[t]} *)

dX = D[X, t]

(* {Derivative[1][f1][t], Derivative[1][f2][t], 
 Derivative[1][f3][t], Derivative[1][f4][t]} *)

Here we define the vector of integral functions -- F[_,_] will be replaced below.

iX = F[#, t] & /@ funcs

(* {F[f1, t], F[f2, t], F[f3, t], F[f4, t]} *)

Obtain a random matrix for A:

SeedRandom[1326]
A = RandomChoice[Exp[-Range[0, 6]] -> Range[0, 6], {n, n}];
A = Rescale[A + DiagonalMatrix[ConstantArray[1, n]]];
MatrixForm[A]

enter image description here

Form the equations:

eqs = Thread[dX == DiagonalMatrix[iX].A.X - X];

Next ideally we will do this:

eqs = eqs /. F[f_, t_] :> Integrate[f[x], {x, 0, t}];

or this:

eqs = eqs/.F[f_,t_] :> (t (f[t]+f[0])/2);

(Approximation of the integral using Trapezoidal formula.)

But since NDSolve can deal with constant delays only, we are going to use this replacement rule:

eqs = eqs/.F[f_,t_] :> (t (f[t]-f[t-1])/2);

ColumnForm[eqs]

enter image description here

Define initial conditions:

initConds = Thread[Through[funcs[t /; t <= 0]] == 1]

(* {f1[t /; t <= 0] == 1, f2[t /; t <= 0] == 1, f3[t /; t <= 0] == 1, f4[t /; t <= 0] == 1} *)

Find the solution:

NDSolve[Join[eqs, initConds], funcs, {t, 0, 10}]

enter image description here

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.