# Similarity Solutions

1. Jan 25, 2015

### joshmccraney

Hi PF!

I am confused about solving PDE's using a similarity solution. How do we come up with the similarity variable? Is it scaling or lucky guesses?

I've read several papers on it but haven't found the reasons on why they use the variables they do for the similarity to work. Any advice from you would be great!

Thanks!

2. Jan 25, 2015

### Simon Bridge

Solving PDEs is guesswork ... you get better at guessing as you gain experience.
http://web.iitd.ac.in/~vvksrini/Similar3.pdf [Broken]

Last edited by a moderator: May 7, 2017
3. Jan 25, 2015

### joshmccraney

This is one of the papers I read. I don't see the significance of either $$\frac{z}{s^{a/b}} = \frac{x}{t^{a/b}} \\ vs^{-c/b} = ut^{-c/b}$$ and how the above relations help us decide that $\xi = x / t^{a/b}$ and that $u = t^{c/b} y ( \xi )$.

Could you help me out?

4. Jan 26, 2015

### rumborak

I think that was always my main gripe with them, or at least when I learned about them. I totally understand that they are usually *the* way of describing most physical things. But the solutions one learned about always seemed cherry-picked to work.
In the end, I always had the feeling that if I ever were to encounter one in real life and needed the solution, a computer simulation was the only way.

5. Jan 30, 2015

### Staff: Mentor

If you can cite a specific problem, I think I can help you through this. There was a procedure we learned in school for identifying the similarity variable.

Chet

6. Jan 31, 2015

### joshmccraney

So I'm not looking at one specifically now, but perhaps I can take you up on this offer later on (in a few weeks)?

7. Jan 31, 2015

No problem.

Chet