search

google / tf-quant-finance

High-performance TensorFlow library for quantitative finance.
Apache License 2.0
4.57k stars 580 forks source link

Integration/interpolation methods #39

Open jonas-eschle opened 4 years ago

jonas-eschle commented 4 years ago

Hi all, many thanks for this library! As I've seen, there is an integration method and interpolation. The latter are also available in other packages (tf-graphics, tf-addons) in one way or another while the integration is not (AFAIK?).

However, we implemented multiple integration algorithms and plan to put them up/into an existing repository. TFP seems nice but maybe not fully suited. However, this repository also does not seem directly like the place for this methods (as it is a very general used functionality, especially in science). Or what do you think?

Would it make sense to put this somewhere else? If so, where?

Many thanks for all the work!

saxena-ashish-g commented 4 years ago

Hi Jonas,

Thank you for getting in touch.

Are your integration algorithm implementations in TF? Are they C++ kernels or python TF?

I agree that integration methods are generally useful and it would be good to separate them out of narrow focus libraries like ours and TFP. We have (very vague at this stage) to create a separate maths package containing commonly used tools but haven't formalised any plans yet.

It would be good to understand the timelines you are looking at and the coverage of your code.

Regards, Ashish

On Wed, 18 Nov 2020 at 17:51, Jonas Eschle notifications@github.com wrote:

Hi all, many thanks for this library! As I've seen, there is an integration method and interpolation. The latter are also available in other packages (tf-graphics, tf-addons) in one way or another while the integration is not (AFAIK?).

However, we implemented multiple integration algorithms and plan to put them up/into an existing repository. TFP seems nice but maybe not fully suited. However, this repository also does not seem directly like the place for this methods (as it is a very general used functionality, especially in science). Or what do you think?

Would it make sense to put this somewhere else? If so, where?

Many thanks for all the work!

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/google/tf-quant-finance/issues/39, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFRDBQXIAKZWRF763CYIWJLSQQCTPANCNFSM4T2JOBLQ .

jonas-eschle commented 4 years ago

The integrations are in Python TF, similar to the simpson. In fact, a tensorflow-math package would be great! Similar to scipy of the scope?

In terms of timeframe we would be interested to have such a package rather sooner than later, just to avoid further duplications of others. Especially to converge on an API would be great for e.g. integration methods (1D and ND).

While our implementations are not excellent, it's better to either a) polish them up and contribute or b) use another method provided by whomever in order to channel the efforts.

Any timeframe regarding the tf-math repository? Or any way we can help/contribute/be involved in?

jonas-eschle commented 3 years ago

Hi, is there any update on this? To me, it really seems in the wrong place currently: all this math tools are great, but they are used in general statistics or science extensively (e.g. TensorFlow Probability) and we should try to have it in a common place.

cyrilchim commented 3 years ago

Hi Jonas,

I agree that many of the maths components could be extracted into a separate repo. At the moment we do not have the capacity to move the code. For now please feel free to make a contribution to the current repo.

Please let me know if you still wish to contribute any specific methods.