All numerical methods are based on the representation and processing of numerical values. The method of programmable problem solving used may require high accuracy, which cannot be achieved using standard data types of a programming language. The accuracy of the problem solution (the result of the program) depends not only on the accuracy of the method but also on the accuracy of the implementation of values in the computer/program. This is a frequent case in the use of iterative procedures in problem solving. We demonstrate this fact on standard generators of random numbers and numerical sequences, which play a significant role in modeling systems and various random phenomena. We also give an example of obtaining the value of a definite integral - calculating the area in an iterative way. We also suggest the possibility of improving the method used.

Keywords: Numerical methods, Implementation of values, Accuracy of method, Calculation accuracy.