about some numbers

[dyt0414]

hi i have read the paper and the code, and i think it is very cool! But i have some questions about the code.
What are the meanings of the numbers , such as ''zeros(2mn + 5n + 11mm + 8m, float64)'', in the hoag.py around line 74?

[fabianp]

these are taken from the implementation of lbfgs in scipy, see for example https://github.com/scipy/scipy/blob/v1.5.4/scipy/optimize/lbfgsb.py#L337

…

On Sat, Dec 5, 2020 at 11:47 PM dyt0414 ***@***.***> wrote: hi i have read the paper and the code, and i think it is very cool! But i have some questions about the code. What are the meanings of the numbers , such as ''zeros(2*m*n + 5*n + 11*m*m + 8*m, float64)'', in the hoag.py around line 74? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AACDZBZ7HFD4GPTCINNRKATSTMEHHANCNFSM4UPBRCKA> .

[dyt0414]

Thank you so much!
I access the lbfgs example in scipy, but there is no explanation too, so I track the paper which the example cites. I found the paper is hard to understand, so I try to ignore the explicit meaning of the numbers, and wonder whether these numbers are fixed in most situation the lbfgs used?

[fabianp]

Yes, that's right. They're fixed

…

On Tue, Dec 8, 2020, 03:23 dyt0414 ***@***.***> wrote: Thank you so much! I access the lbfgs example in scipy, but there is no explanation too, so I track the paper which the example cites. I found the paper is hard to understand, so I try to ignore the explicit meaning of the numbers, and wonder whether these numbers are fixed in most situation the lbfgs used? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#1 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AACDZBZASYHAX3NMNCE7D4LSTXPA3ANCNFSM4UPBRCKA> .