The article introduces a new way to test hypotheses in science. Instead of using moment (in)equality constraints, the researchers suggest using density function constraints. This method, called the empirical likelihood ratio test, is better at detecting certain types of problems. It works well for both parametric and non-parametric detection tasks. The test is especially effective when the alternative hypothesis has a specific structure. Overall, the density function constraints improve the accuracy of the test compared to other methods like the robust Kolmogorov-Smirnov test.