Erratum to “An improved stability criterion of neural networks with time-varying delays in the form of quadratic function using novel geometry-based conditions” [Appl. Math. Compt. 404 (2021) 126226] (Applied Mathematics and Computation (2021) 404, (S0096300321003167), (10.1016/j.amc.2021.126226))

In [Appl. Math. Compt. 404 (2021) 126226], we have noticed typos in [Formula presented] and [Formula presented] related to Theorem 1 and [Formula presented] in Example 2. Because of the typo in [Formula presented], the numerical results in Tables 1-3 of the original paper lead incorrectly. In this erratum, we would like to correct them and provide corrected numerical results. Also, we would like to clarify typesetting errors for [Formula presented], [Formula presented], and [Formula presented] occurred by the Elsevier ProofCentral System. Here, three typos should be corrected as follows: • [Formula presented] in the fifth line of Page 5• [Formula presented] in Example 2, Page 8• [Formula presented] in the last line of Page 5Also, three typesetting errors which were beyond the page margin should be as follows: • [Formula presented] in the first line of Page 5• [Formula presented] in the third line of Page 5• [Formula presented] in the fourth line of Page 5Hereby, these corrections lead new numerical results as listed in Table 1 of this erratum. As seen in Table 1 of this erratum, our results are still larger (in other words, our methods are less conservative) than those works in the tables of the original paper except [40] of the original paper. In our simulation using MATLAB2020b Simulink with Runge-Kutta method, 0.001 step time, [Formula presented] ([Formula presented]) where a matrix [Formula presented] was given in Example 1 of the original paper, and [Formula presented], the system in Example 1 with constant delay (i.e. [Formula presented], [Formula presented]), however, is not asymptotically stable at [Formula presented], so the result of [40] of the original paper seems to be open for discussion.