The primary principle of Monte Carlo method in solving the initial-boundary value problem for partial differential equations is introduced. And an improved strategy to solve the slow speed problem in computations with large grid points and random walks is proposed. Through the theoretical analysis and numerical validation

the improved strategy not only takes much less computational time

but also gives more accurate results comparing to the traditional methods under the same conditions.