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 have 1D diffusion (u(t,x)) PDE with Dirac Delta initial condition.

Question is regarding it's implementation:

Dirac delta func is formally defined as an encapsulation of 2 conditions:

1st condn: function takes value 1 at x=0,

2nd condition:function takes value 0, if x not equal to 0).

Thus, to solve the above PDE for initial condition u(t=0,x)=Dirac_delta(x), is it necessary to treat this dirac delta function initial condition (for that PDE) as ONE or TWO separate conditions (viz. after substituting for t=0 and x=0 giving u(t=0,x=0)=1 solution giving 1st equation for initial condition & substituting t=0 and x=x(i.e non-zero x) giving u(t=0,x)=0 giving 2nd equation for initial condition)??

This means, 1D diffusion PDE with Dirac Delta function (delta(x)) has 1 OR 2 initial conditions, according to for t=0, (x=0 & x not equal to 0)?


share|cite|improve this question
Your question is quite a mess. First, please tell us typically the diffusion PDE. Second, please use TEX when typing the equations. – doraemonpaul Jan 15 '13 at 11:29
up vote 0 down vote accepted

Your description of the Dirac delta ($\delta_0$) is incorrect. One of possible correct definitions is: $\delta_0$ is the derivative of the Heaviside function $h_0$, which is defined as $h_0(x)=1$ when $x\ge 0$; $h_0(x)=0$ when $x< 0$. The meaning of the derivative is not the classical one; instead, it is a stipulation that the certain integral identities hold: Fundamental theorem of Calculus and integration by parts. In particular, $\int_a^b \delta_0 =h_0(b)-h_0(a)=1$ whenever $a<0<b$.

Now back to counting initial conditions. If we have $u(0,x)=x^2$ for all $x\in\mathbb R$, is this one condition? Or maybe infinitely many, because there's a condition for every $x\in\mathbb R$? I guess we take the position that this is one condition: the distribution of matter represented by $u$ is given at time $t=0$, and that's all we have.

Another example: $u(0,x)=h_0(x)$ for all $x\in\mathbb R$. This is also one initial condition, we just have another function here instead of $x^2$. Since the function $h_0$ is piecewise defined, the condition $u(0,x)=h_0(x)$ amounts to $$u(0,x)=0 \text{ when } x<0 \ \text{ and } \ u(0,x)=1 \text{ when } x\ge 0$$ But this split is just our way of writing $h_0$. It is of no consequence for the mathematical problem.

Similarly, the initial condition $u(0,\cdot) =\delta_0$ is also one condition, which may appear as two (or three) depending on how $\delta_0$ is presented in an application-oriented text.

share|cite|improve this answer
@5PM..Thanks for the answer. Could you then tell me, how one implements such a Dirac Delta initial condition on 1D Diffusion PDE (u_sub{t}=D*u_sub{xx}, D:diffusion constant, _sub{x} refers to partial derivative taken wrt x)? Is it implemented only for x=0 (assuming it is δ(0,x)) or rather implemented as an integral(δ(0,x)) fashion? I would like to hear if it can be implementated in x=0 fashion, rather than in integral fashion. Hope its clear. – ems Jan 15 '13 at 22:53
@ems Are you familiar with the concept of a fundamental solution? – user53153 Jan 15 '13 at 22:55
@5PM...Yes. Fundamental solution helps finding the solution for any arbitrary initial condition, provided I have the solution calculated using Dirac Delta initial condition, right? I am trying to find the solution to 1D PDE not by analytical means (which I can find in any textbook), but rather using Differential Transform method (numerical approach, basis function based) – ems Jan 15 '13 at 22:58
@ems For numerical computations you need a nascent delta function, such as the piecewise linear $\eta_\epsilon(x)=\epsilon^{-1} \max(0,1-|x|/\epsilon)$, for small $\epsilon>0$. How small you can take it to be will depend on the particulars of your methods. – user53153 Jan 15 '13 at 23:11

This answer is incomplete reference to the Brownian motion "stocastic Calculus" and the same initial conditions we get the normal distribution

share|cite|improve this answer
Welcome to MSE! Was this supposed to be a comment. I realize you don't yet have enough reputation, but this area is for answers. Regards – Amzoti May 3 '13 at 5:29

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.