Scipy Optimize (L-BFGS-B) Output : What does variable M represent?

Asked Jul 01 '22 at 08:50

Active Jul 01 '22 at 10:54

Viewed 149 times

RUNNING THE L-BFGS-B CODE

Machine precision = 2.220D-16

N = 425 M = 10

At X0 0 variables are exactly at the bounds

What is the meaning of variable M? I know N stands for the number of trainable parameters.

edited Jul 01 '22 at 10:54

asked Jul 01 '22 at 08:50

tonygrey

1

It seems to be the *the maximum number of variable metric corrections used to define the limited memory matrix* see [here](https://github.com/scipy/scipy/blob/cdcbb40062aa4fe2e324e590f159dd1f51ea6375/scipy/optimize/lbfgsb_src/lbfgsb.f#L67) and [here](https://github.com/scipy/scipy/blob/cdcbb40062aa4fe2e324e590f159dd1f51ea6375/scipy/optimize/lbfgsb_src/lbfgsb.f#L2698) – joni Jul 01 '22 at 09:05

0 Answers0