Equilibrium optimizer (EO) is a recently proposed physics-based algorithm inspired by the control volume mass balance model. Although it provides satisfactory solutions for several real-world problems, it still suffers from local optimal areas and the imbalance between exploration and exploitation. To alleviate these shortcomings, this paper suggests an efficient EO (SLEO) with application in numerical optimization. We design a new strategy called Laplacian opposition-based Learning to assist the particles escaping from local optimal solutions. Then, the Sinusoidal Map is injected to maintain a good balance between exploration and exploitation throughout the iterative process. To validate the performance of SLEO, thirteen benchmark functions with different dimensions are used for evaluation, in comparison with six other popular methods. The Friedman testing is used to evaluate the significant difference between approaches. The experimental results demonstrate that SLEO is superior to other methods in terms of optimization accuracy and convergence speed.