如果将能谱实验数据的对数拟合成直线以求得粒子温度,则最好对温度误差的平方和极小化。
We combine unitary nonlinear regression with multivariate linear regression, it can reduce the error sum of squares of the model and improve the precision of forecast.即把一元非线性回归和多元线性回归结合起来,构造一个混合回归模型,这样就减小了模型的残差平方和,从而提高预报的准确性。
Fitting a curve for an experimental energy spectrum of plasma particles with the least square method, a square error sum between the fitted curve and experimental spectrum data is usually minimized.用最小二乘法拟合等离子体粒子能谱实验数据时,通常使拟合函数与实验能谱数据之间的误差平方和极小化。