|
| | Abstract | | In nonlinear least squares method, a computational cost of Jacobian using finite difference becomes bigger with increasing number of unknown parameters. A Jacobian-free algorithm for nonlinear least squares method was developed to avoid the large computational cost. In this algorithm implicit Jacobian information are used instead of explicit Jacobian. Two examples of resistivity inversion using Jacobian-free algorithm are shown in this paper. As a result, the Jacobian-free algorithm is found to be effective for large scale inversion. |
|
|
|
| |
|