Finally in v13.1 the function DSolveChangeVariables is introduced, try it out! DChange in the answer below is still a good choice, of course.

Original Question

Maple owns an interesting function called dchange which can change the variables of differential equations, but there seems to be no such function in Mathematica.

Has any one ever tried to write something similar? I found this, this and this post related, but none of them attracted a general enough answer.

"So, what have you tried?" - Well, nothing. I decided to ask this question first to see if someone has already implemented the functionality and waited for a chance to make it public. If this question finally elicits no answer, I'll have a try.

The imaginary syntax for the function is

dChange[DE, relation, var]

where DE is the differential equation(s) to be transformed, and relation is the transformation relation(s) expressed as equation(s) i.e. with head Equal, var is the variable(s) to be changed.

Here are some examples for the imaginary behaviour:

Example 1

Originated from this answer implementing stereographic projection.

dChange[1/η D[η D[f[η], η], η] + (1 - s^2/η^2) f[η] - f[η]^3 == 0, 
        η == Sqrt[(1 + z)/(1 - z)], η]
(1/(1 + z)) ((-(1 + s^2 (-1 + z) + z)) f[z] + (1 + z) f[z]^3 + 
    (-1 + z)^2 (1 + z) (2 z f'[z] + (-1 + z^2) f''[z])) == 0

Example 2

Originated from this answer for Stefan's problem.

dChange[D[u[x, t], t] == D[u[x, t], {x, 2}], x == ξ s[t], x]
Derivative[0, 1][u][ξ, t] - (ξ s'[t] 
      Derivative[1, 0][u][ξ, t])/s[t] == Derivative[2, 0][u][ξ, t]/s[t]^2

Example 3

Originated from this answer. This technique is also used in the reduction of d'Alembert's formula.

dChange[D[y[x, t], t] - 2 D[y[x, t], x] == Exp[-(t - 1)^2 - (x - 5)^2],
        {ξ == t + x/2, η == t}, {x, t}]
Derivative[0, 1][y][ξ, η] == E^(-(-1 + η)^2 - (5 + 2 η - 2 ξ)^2)

I'll add more if I recall other representative examples.

  • 1
    $\begingroup$ possible duplicate of Change variables in differential expressions $\endgroup$
    – m0nhawk
    Apr 18, 2015 at 8:32
  • 4
    $\begingroup$ @m0nhawk Well, as I mentioned above, that's just one of the related questions that are not general enough. $\endgroup$
    – xzczd
    Apr 18, 2015 at 8:33
  • 1
    $\begingroup$ For a quiet long time of using Mathematica the Replace and ReplaceAll are more than enough and, actually, I found them much powerful than Maple's dchange. $\endgroup$
    – m0nhawk
    Apr 18, 2015 at 8:37
  • 3
    $\begingroup$ The link has a few examples, and (re: @m0nhawk) I'm not sure that simply RepkaceAll will provide the same functionality. $\endgroup$ Apr 18, 2015 at 9:12
  • 2
    $\begingroup$ I don't know much about Maple, but it seems from your examples that it's less "careful" when simplifying expressions: Mathematica leaves expressions unevaluated if it can't get a result that's valid generically or consistent with the given assumptions. So probably one would have to allow an additional Assumptions option in the dChange emulation to tell Mathematica which variables are positive, or complex, etc... so it has a better chance of inverting and simplifying the required relations. Anyway, I like the idea... $\endgroup$
    – Jens
    Apr 18, 2015 at 16:32

1 Answer 1


I've put this code on a GitHub but I don't know what features are needed or what problems it may give. I'm just not using it.

But I will incorporate incomming suggestions as soon as I have time.

Feedback in form of tests and suggestions very appreciated!

(If[DirectoryQ[#], DeleteDirectory[#, DeleteContents -> True]];
    "https://raw.githubusercontent.com/" <> 
    FileNameJoin[{#, "MoreCalculus.m"}]
) & @ FileNameJoin[{$UserBaseDirectory, "Applications", "MoreCalculus"}]


So this is a package MoreCalculus` with the function DChange inside.

What's new:

DChange automatically takes under consideration range assumptions for built-in transformations: (not heavily tested)

  D[f[x, y], x, x] + D[f[x, y], y, y] == 0, 
  "Cartesian" -> "Polar", {x, y}, {r, θ}, f[x, y]

enter image description here


DChange[expresion, {transformations}, {oldVars}, {newVars}, {functions}]

DChange[expresion, "Coordinates1"->"Coordinates2", ...]   

DChange[expresion, {functionsSubstitutions}] 

You can also skip {} if a list has only one element.


Change of coordinates

  • rules accepted by CoordinateTransform are now incorporated, as well as coordinates ranges assumptions associated with them

      D[f[x, y], x, x] + D[f[x, y], y, y] == 0, 
      "Cartesian" -> "Polar", {x, y}, {r, θ}, f[x, y]

    enter image description here

    The transformation is returned too, to check if the canonical (in MMA) order of variables was used.

  • wave equation in retarded/advanced coordinates

     D[u[x, t], {t, 2}] == c^2 D[u[x, t], {x, 2}]
     {a == x + c t, r == x - c t}, {x, t},  {a, r},  {u[x, t]}  ]
    c Derivative[1, 1][u][a, r] == 0

  • stereographic projection

     D[η*D[f[η], η], η]/η + (1 - s^2/η^2)*f[η] - f[η]^3 == 0
     η == Sqrt[(1+z)/(1-z)],  η,  z,   f[η]   ]
    ((z-1)^2 (z+1)((z^2-1) f''[z]+2 z f'[z])-f[z] (s^2 (z-1)+z+1)+(z+1)     f[z]^3)/(z+1)==0

Example from @Takoda

$$ \begin{pmatrix}\dot{x}\\ \dot{y} \end{pmatrix}=\begin{pmatrix}-y\sqrt{x^{2}+y^{2}}\\ x\sqrt{x^{2}+y^{2}} \end{pmatrix} $$

out = DChange[
  Dt[{x, y}, t] == {-y r^2, x r^2}, "Cartesian" -> "Polar", 
  {x, y}, {r, θ}, {}

Solve[out[[1]], {Dt[r, t], Dt[θ, t]}]
{{Dt[r, t] -> 0, Dt[θ, t] -> r^2}}

Functions replacement

  • example on special case separation of Fokker-Planck equation

      -D[u[x, t], {x, 2}] + D[u[x, t], {t}] - D[x u[x, t], {x}]
      u[x, t] == Exp[-1/2 x^2] f[x] T[t]
    ] // Simplify
    % / Exp[-x^2/2] / f[x] / T[t] // Expand

    enter image description here

Code: (latest version is on GitHub)


DChange[expr_, transformations_List, oldVars_List, newVars_List, functions_List] := 
  Module[ {pos, functionsReplacements, variablesReplacements, arguments,
           heads, newVarsSolved}
    pos = Flatten[
            Outer[Position, functions, oldVars], 
            {{1}, {2}, {3, 4}}

    heads = functions[[All, 0]];
    arguments = List @@@ functions;
    newVarsSolved = newVars /. Solve[transformations, newVars][[1]];

    functionsReplacements = Map[
        heads[[i]] -> (
          Function[#, #2] &[
            ReplacePart[functions[[i]], Thread[pos[[i]] -> newVarsSolved]]
          ] )
      Range @ Length @ functions

   variablesReplacements = Solve[transformations, oldVars][[1]];

   expr /. functionsReplacements /. variablesReplacements // Simplify // Normal

DChange[expr_, functions : {(_[___] == _) ..}] := expr /. Replace[
  functions, (f_[vars__] == body_) :> (f -> Function[{vars}, body]), {1}]

DChange[expr_, x___] := DChange[expr, ##] & @@ Replace[{x}, 
   var : Except[_List] :> {var}, {1}];

DChange[expr_, coordinates:Verbatim[Rule][__String], oldVars_List,
        newVars_List, functions_    ]:=Module[{mapping, transformation},
        mapping = Check[
            CoordinateTransformData[coordinates, "Mapping", oldVars],
        transformation = Thread[newVars == mapping ];
            DChange[expr, transformation, oldVars, newVars, functions],


  • add some user friendly DownValues for simple cases
  • heavy testing needed, feedback appreciated
  • exceptions/errors handling. it is only as powerful as Solve so may brake for more convoluted implicit relations
  • it is not designed as a scoping construct

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