Abstract:We introduce a high-order numerical scheme for fractional ordinary differential equations with the Caputo derivative. The method is developed by dividing the domain into a number of subintervals, and applying the quadratic interpolation on each subinterval. The method is shown to be unconditionally stable, and for general nonlinear equations, the uniform sharp numerical order $3-\nu$ can be rigorously proven for sufficiently smooth solutions at all time steps. The proof provides a general guide for proving the sharp order for higher-order schemes in the nonlinear case. Some numerical examples are given to validate our theoretical results.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to develop a new high - order numerical method for solving nonlinear fractional - order differential equations with Caputo derivatives and ensure uniform accuracy at all time steps. Specifically, the main contributions of the paper include:
1. **Proposing a new finite - difference approximation**: This method can provide a high - order numerical method for fractional - order differential equations with uniform accuracy at all time steps.
2. **Proving unconditional stability**: For the eigenvalue problem, the paper strictly proves the unconditional stability of the method.
3. **Proposing a general convergence proof for the nonlinear right - hand side**: The paper proposes a new technique to prove the convergence order of the general nonlinear right - hand side, thus achieving a convergence order of 3 - ν.
### Background of the Paper
Fractional - order differential equations are more accurate in describing certain physical phenomena and chemical processes than integer - order differential equations. However, the exact solutions of fractional - order differential equations are usually difficult to obtain. Even if there are analytical solutions, they often appear in the form of series and are difficult to calculate. Therefore, the numerical study of fractional - order differential equations becomes very important.
### Content of the Paper
The paper mainly focuses on the following initial - value problem:
\[ {}_0^C D_x^\nu y(x) = f(x, y(x)), \quad 0 < x \leq T, \]
with the initial condition \( y(0) = y_0 \), where \( {}_0^C D_x^\nu \) is the Caputo derivative, defined as:
\[ {}_0^C D_x^\nu y(x) = \int_0^x \omega_{1-\nu}(x - s) y'(s) \, ds, \]
where \( \omega_{1-\nu}(x)=\frac{x^{-\nu}}{\Gamma(1 - \nu)} \).
### Numerical Method
The paper proposes a high - order numerical method based on piecewise quadratic interpolation. The specific steps are as follows:
1. **Piecewise quadratic interpolation**: Divide the interval \([0, T]\) into several sub - intervals and apply quadratic interpolation on each sub - interval.
2. **Discrete Caputo derivative**: Obtain the discrete Caputo derivative through integration and interpolation.
3. **Numerical scheme**: The final numerical scheme is an implicit scheme, which needs to solve the values of the first two time steps simultaneously to maintain the accuracy at the first time step.
### Stability and Convergence Analysis
1. **Stability analysis**: The paper proves the unconditional stability of the method, that is, for any time step, the modulus of the numerical solution is less than the modulus of the initial value.
2. **Convergence analysis**: By introducing a new technique, the paper proves that for a sufficiently smooth solution and a general nonlinear right - hand side, the convergence order of the method can reach 3 - ν.
### Conclusion
The paper successfully develops a new high - order numerical method for solving nonlinear fractional - order differential equations with Caputo derivatives and has uniform accuracy at all time steps. This method is not only strictly proven theoretically but also verified its effectiveness and accuracy through numerical experiments.