A New Hybrid Algorithm of Trust Region Methods
De Qing Zhang, Peipei Zhou, Qing Hua Zhou
Abstract
De Qing Zhang, Peipei Zhou, Qing Hua Zhou
Abstract
In the solution of trust region subproblem within the classical trust region algorithms, the centre of sphere is the current interaction point and one step-size is the upper bound. Considering that only with the negative gradient direction to acute angle may reduce the function value, we introduce the parameter to control of the centre of sphere and the radius. Based on the numerical experiments, obtains the value range of the parameter. The numerical evaluation demonstrates the validity of the new trust region algorithms.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
In the solution of trust region subproblem within the classical trust region algorithms, the centre of sphere is the current interaction point and one step-size is the upper bound. Considering that only with the negative gradient direction to acute angle may reduce the function value, we introduce the parameter to control of the centre of sphere and the radius. Based on the numerical experiments, obtains the value range of the parameter. The numerical evaluation demonstrates the validity of the new trust region algorithms.
Key concepts: RADIUS, Trust region, Range (aeronautics), Function (biology), Upper and lower bounds, Value (mathematics), Point (geometry), Mathematical optimization