完全相同的数据，在Eviews和R里面都可以完美的做出自动寻找平滑参数的hwinters平滑，但是仅仅在stata中不行，在stata中做
tssmooth hwinters data2s=data2
的结果就是像下面这样，最终会算16000次，也不能得到平滑结果。

tssmooth hwinters data2s=data2,replace
computing optimal weights
Iteration 0: penalized RSS = -17854.124(not concave)
Iteration 1: penalized RSS = -11325.967
Iteration 2: penalized RSS = -11226.691
Iteration 3: penalized RSS = -11176.521
Iteration 4: penalized RSS = -11160.593
Iteration 5: penalized RSS =-11156.55
Iteration 6: penalized RSS = -11155.481
Iteration 7: penalized RSS = -11155.226
Iteration 8: penalized RSS = -11155.204
Iteration 9: penalized RSS = -11155.196
Iteration 10:penalized RSS = -11155.192
Iteration 11:penalized RSS = -11155.192(backed up)
Iteration 12:penalized RSS = -11155.191(backed up)
Iteration 13:penalized RSS = -11155.182
Iteration 14:penalized RSS = -11155.182(backed up)
Iteration 15:penalized RSS = -11155.182(backed up)

但是当我把Eviews或者R中自动计算出的平滑参数指定给Stata时，Stata就能做出平滑了。
tssmooth hwinters data2s=data2,parms(0.3187079, 0.07837927)
这个是R中成功进行非季节性平滑的截图结果



为什么？ (帖子最下面有数据）