Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I learned the proof of the fact that 3-D laplacian is invariant under all rigid motions in space in Strauss's Partial Differential Equations:

Any rotation in three dimensions is given by $${\bf x'}=B{\bf x},$$ where $B$ is an orthogonal matrix. The laplacian is $$\Delta u=\sum_{i=1}^3 u_{ii}=\sum_{i,j=1}^3\delta_{ij}u_{ij}$$ where the subscripts on $u$ denote partial derivatives. Therefore, $$ \Delta u=\sum_{k,l}\Big(\sum_{i,j}b_{ki}\delta_{ij}b_{lj}\Big)u_{k'l'} =\sum_{k,l}\delta_{kl}u_{k'l'} =\sum_{k}u_{k'k'}. $$

Added: I believe the chain rule is needs here, but I don't see how it is applied here.

Here is my question:

How can I get the first equality, i.e. $$\Delta u=\sum_{k,l}\Big(\sum_{i,j}b_{ki}\delta_{ij}b_{lj}\Big)u_{k'l'}?$$

share|cite|improve this question
The factor $\sum_{i,j} b_{ki} \delta_{ij} b_{lj}$ is just the (k,l)th entry of the matrix $BIB^{-1}$ (because $B^{-1} = B^T$ when B is orthogonal), i.e. $I$, and so is $\delta_{kl}$, though I don't quite understand what k' and l' mean that's different to k and l, so maybe that doesn't answer your question... – Billy Aug 1 '11 at 17:14
The first equality is the chain rule. (If this is not obvious, perhaps the notation needs to be improved.) – Zhen Lin Aug 1 '11 at 17:15
I would think the whole thing could be written in matrix notation, without these sums and indices. – Michael Hardy Aug 1 '11 at 17:33
It looks to me that you are trying to show something like $B^{-1} \triangle ~ B = \triangle $. – Alice Aug 1 '11 at 17:35
up vote 3 down vote accepted

I strongly dislike the use of undecorated subscripts for partial derivatives, so I will write $\partial_j$ for the partial differentiation operator $\partial / \partial x_j$. For convenience I will use the summation convention for repeated indices.

Firstly, we have the change-of-coordinates equation $\tilde{x}_{i'} = B_{i' j} x_j$. The chain rule states $$\frac{\partial}{\partial x_j} = \frac{\partial \tilde{x}_{i'}}{\partial x_j} \frac{\partial}{\partial \tilde{x}_{i'}}$$ so, writing $\tilde{\partial}_{i'}$ for the operator $\partial / \partial \tilde{x}_{i'}$, putting these together, we have $\partial_j = B_{i' j} \tilde{\partial}_{i'}$. The Laplacian is defined by $\nabla^2 = \delta_{i j} \partial_i \partial_j$, so substituting the previous equation in we obtain $\nabla^2 = \delta_{i j} B_{i k'} B_{j l'} \tilde{\partial}_{k'} \tilde{\partial}_{l'}$. By orthogonality, $\delta_{i j} B_{i k'} B_{j l'} = B_{j k'} B_{j l'} = \delta_{k' l'}$, so we are done.

share|cite|improve this answer
+1. Thanks for your proof. It's much clear than the one in the book for me. I still curious about my original question:how can one get the first equality? – Jack Aug 2 '11 at 4:50
@Jack: It's the bit starting at ‘substituting the previous equation in’. It is literally just the chain rule. – Zhen Lin Aug 2 '11 at 6:11

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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