− | 即便参数值与真实值相差很大,有sloppy特性的模型也可以做出精确的预测。在数学中有一个经典的拟合难题<ref>Formulation, application to fitting algorithms: [https://arxiv.org/abs/0909.3884 "Why are nonlinear fits to data so challenging?"], Mark K. Transtrum, Benjamin B. Machta, and James P. Sethna, [https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.104.060201 Phys. Rev. Lett.] '''104''', 060201 (2010).</ref>:用指数衰变和去拟合放射性模型(c列和d列)得到的衰变常数与真实衰变常数截然不同,但短期内模型预测值与真实值却相差不大 。最后,用多项式系数模型<math>\sum_i a_it^i</math>拟合数据是sloppy的(h列)。但用正交多项式基<math>\sum_ib_iH_i</math>(<math>H_i</math>是一组正交多项式基)去拟合时得到的模型却往往是非sloppy的,这是因为从<math>t^i</math>到<math>H_i</math>的变换是高度非正交的<ref>Expanded formulation, geometry of model manifold: [[9. Expanded formulation, geometry of model manifold: "Geometry of nonlinear least squares with applications to sloppy models and optimization", Mark K. Transtrum, Benjamin B. Machta, and James P. Sethna Phys. Rev. E 83, 036701 (2011); |"Geometry of nonlinear least squares with applications to sloppy models and optimization"]], Mark K. Transtrum, Benjamin B. Machta, and James P. Sethna [https://journals.aps.org/pre/abstract/10.1103/PhysRevE.83.036701 Phys. Rev. E '''83''', 036701 (2011)]; </ref>。 | + | 即便参数值与真实值相差很大,有sloppy特性的模型也可以做出精确的预测。在数学中有一个经典的拟合难题<ref>Formulation, application to fitting algorithms: [https://arxiv.org/abs/0909.3884 "Why are nonlinear fits to data so challenging?"], Mark K. Transtrum, Benjamin B. Machta, and James P. Sethna, [https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.104.060201 Phys. Rev. Lett.] '''104''', 060201 (2010).</ref><ref>Expanded formulation, geometry of model manifold:[http://arxiv.org/abs/1010.1449 "Geometry of nonlinear least squares with applications to sloppy models and optimization"], Mark K. Transtrum, Benjamin B. Machta, and James P. Sethna [http://link.aps.org/doi/10.1103/PhysRevE.83.036701 Phys. Rev. E '''83''', 036701 (2011)]; </ref>:用指数衰变和去拟合放射性模型(c列和d列)得到的衰变常数与真实衰变常数截然不同,但短期内模型预测值与真实值却相差不大 。最后,用多项式系数模型<math>\sum_i a_it^i</math>拟合数据是sloppy的(h列)。但用正交多项式基<math>\sum_ib_iH_i</math>(<math>H_i</math>是一组正交多项式基)去拟合时得到的模型却往往是非sloppy的,这是因为从<math>t^i</math>到<math>H_i</math>的变换是高度非正交的<ref>Expanded formulation, geometry of model manifold: [[9. Expanded formulation, geometry of model manifold: "Geometry of nonlinear least squares with applications to sloppy models and optimization", Mark K. Transtrum, Benjamin B. Machta, and James P. Sethna Phys. Rev. E 83, 036701 (2011); |"Geometry of nonlinear least squares with applications to sloppy models and optimization"]], Mark K. Transtrum, Benjamin B. Machta, and James P. Sethna [https://journals.aps.org/pre/abstract/10.1103/PhysRevE.83.036701 Phys. Rev. E '''83''', 036701 (2011)]; </ref>。 |