First of all, hello and thank you for your time.


I am making a program that solves the differential equation for the time evolution of a system from the equations: $$F[\mathbf{y}]=\int\limits_{\Omega\subset\mathbb{R}^n}f(\mathbf{x},\mathbf{y}, \mathbf{\nabla y})d\mathbf{x}$$ $$\frac{\partial y_i}{\partial t}=k\Delta\frac{\delta F}{\delta y_i}$$ Where $\Delta$ is the Laplacian and $$\frac{\delta F}{\delta y_i}=\sum\limits_k\frac{\partial}{\partial x_k}\frac{\partial f}{\partial (\partial_ky_i)}-\frac{\partial f}{\partial y_i}$$ is the variational derivative ($\partial _k$ is the partial derivative with respect to $x_k$). Also: $\mathbf{x}\in\mathbb{R}^n, \mathbf{y}\in\mathbb{R}^m$

The system should satisfy the global conservation constraints: $$J_i[\mathbf{y}]=\int\limits_{\Omega\subset\mathbb{R}^n}y_id\mathbf{x}=k_i$$ Where the $k_i$ are constants.

I ran the program (without adding the Lagrange multipliers to the integral). And noticed that the $J_i$ increased with time, which was obviously not intended.


I want to add the constraints to the solution. At first I naïvely thought that I could just modify the functional by adding lagrange mutipliers: $$K[\mathbf{y}]=F[\mathbf{y}]-\sum\limits_i\lambda_i J_i[\mathbf{y}]$$

But when checking my reference book I noticed the Theorem said (I modified and omitted parts to take what's most relevant to the current question):

Suppose that $F$ has an extremum at $y\in C^2[x_0, x_1]$ subject to the boundary conditions [...]. Then there exist two numbers $\lambda_0, \lambda_1$ not both zero such that $$\frac{\delta K}{\delta y}=0$$ Where $K=\lambda_0 F-\lambda_1 J$

As the theorem says, this works when one wishes to find the extremum so my naïve assumption is probably wrong since the system I described only reaches the extremum of the functional when $\mathbf{y}$ gets to the steady state (i.e. it approaches it asymptotically).

Is there a way to satisfy the constraint continuously throughout the time evolution of the system despite $F$ not being stationary?

I understand this may be more involved than what an answer in the site may allow so if you know of any good textbook where I could find it I would also be very grateful.


Your Answer

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

Browse other questions tagged or ask your own question.