Do inequalities in boundary conditions imply inequalities in solutions to a PDE?

66 Views Asked by At

I'm working on an unassessed course problem to show the solution $p$ to a PDE satisfies an inequality. I think I can show that of the 2 boundary conditions of the PDE, $p$ is equal to another solution $f$ in one of the BCs and greater-than-or-equal-to $f$ in the other BC. I understand that if both BCs matched then $p=f$. Given one equality and one inequality, does this mean $p\geq f$?


For those familiar with the context topic (finance), here's the problem.

Consider a European put option with strike price $X$ and maturity time $T>0$. For any time $0\leq t\leq T$, let $S$ denote the spot price of the underlying stock and let $p(S_t,t)$ be the price of a European put option at time $t$. Assume that all spot interest rates are constant and equal to $r$. Assume also that the underlying stock price follows the Ito process $$\text{d}S = rS\text{ d}t + \sigma S \text{ d}B.$$ Verify that $f(S_t,t)=Xe^{-r(T-t)}-S$ is a solution to the Black-Scholes equation, $$\frac{\partial f}{\partial t}+rS\frac{\partial f}{\partial S}+\frac{1}{2}\sigma^2S^2\frac{\partial^2f}{\partial S^2}=rf.$$ Use this to deduce that $p\geq Xe^{-rT}-S.$

I get \begin{align} & \lim_{S\downarrow 0}p(S_t,t)=Xe^{-r(T-t)}=\lim_{S\downarrow 0}f(S_t,t), \\ & \lim_{t\uparrow T}p(S_t,t)=\max(\{X-S_T,0\})\geq X-S_T=\lim_{t\uparrow T}f(S_t,t) \end{align} and then find myself asking the question in the title.

1

There are 1 best solutions below

0
On BEST ANSWER

IIRC this does not work in general unless your equation satisfies the maximum principle. If your equation is linear and for boundary conditions $a \leq b$ solutions are $f$ and $g$ respectively, you get that $h = f-g$ is also a solution (due to linearity) and $h = (a - b) \leq 0$ on the boundary. Hence $h$ must be non-positive everywhere, otherwise it would have a maximum not on the boundary. That is $a\leq b$ implies $f\leq g$ everywhere. For heat equation (which you can get from BS by change of variables) you have maximum principles.

Besides that, BS equation has explicitly known fundamental solution $\phi$ (solution for boundary condition given by Dirac delta), so that for any terminal (boundary) condition (payoff) $F(x, T)$ you get that the solution is $$ f(x,t) = \int_\Bbb R F(x, T)\phi(x)\mathrm dx. \tag{1} $$ From that you can also get your inequalities.


Note that $(1)$ is the formula which states that $f(x,t) = \mathrm e^{-r(T - t)}\Bbb E_\Bbb Q[F(x, T)]$ i.e. the value of the contigent claim now is a discounted expected value of its payoff w.r.t. the martingale measure.