Approximation theory plays a central role in numerical analysis, undergoing continuous evolution through a spectrum of methodologies. Notably, Lebesgue, Weierstrass, Fourier, and Chebyshev approximations stand out among these methods. However, each technique possesses inherent limitations, underscoring the critical importance of selecting an appropriate approximation method tailored to specific problem domains. This article delves into the utilization of Chebyshev polynomials at Chebyshev nodes for approximation. For sufficiently smooth functions, the partial sum of Chebyshev series expansion offers optimal polynomial approximation, rendering it a preferred choice in various applications such as digital signal processing and graph filters due to its computational efficiency. In this article, we focus on functions of bounded variation, which find numerous applications across mathematical physics, hyperbolic conservations, and optimization. We present two optimal error estimations associated with Chebyshev polynomial approximations tailored for such functions. To validate our theoretical assertions, we conduct numerical experiments. Additionally, we delineate promising future avenues aligned with this research, particularly within the realms of machine learning and related fields.

翻译：逼近理论在数值分析中占据核心地位，通过各种方法不断演进。其中，勒贝格、魏尔斯特拉斯、傅里叶和切比雪夫逼近方法尤为突出。然而，每种技术都固有局限性，这凸显了针对具体问题领域选择合适逼近方法的关键重要性。本文深入探讨了在切比雪夫节点上利用切比雪夫多项式进行逼近的方法。对于足够光滑的函数，切比雪夫级数展开的部分和提供了最优多项式逼近，因其计算效率而在数字信号处理和图滤波器等应用中成为优选。本文聚焦于有界变差函数，这类函数在数学物理、双曲守恒律和优化中有广泛应用。我们针对此类函数提出了两种与切比雪夫多项式逼近相关的最优误差估计。为验证理论论断，我们进行了数值实验。此外，我们还展望了与本研究一致的有前景的未来方向，特别是在机器学习及相关领域。