LO polarised anomalous dimension by giacomomagni · Pull Request #170 · NNPDF/eko

giacomomagni · 2022-11-26T13:35:53Z

This PR is for the implementation of LO polarised anomalous dimension.

alecandido · 2022-11-26T14:54:39Z

Why did you put it in a new package?

Since they will share almost everything with eko, apart from the anomalous dimensions themselves, I would put together everything else:

eko
|- anomalous_dimensions/
| |- unpolarized/
| | |- ...
| |- polarized/
| | |- ...

We do not want to duplicate the code, and having to maintain two copies...

giacomomagni · 2022-11-26T16:12:23Z

Why did you put it in a new package?

Hi @alecandido, I think this was done by mistake.
This PR is just to keep track of the work done by @adrianneschauss, but for the moment it's just a draft.
I've already warned her about a couple of good practice tips.

alecandido · 2022-11-26T16:24:15Z

Ok, I guess you can take care of it: I know it was draft, but I didn't want the final product to go in that direction (let's say that I've been rather astonished when I saw 87 files changed...).

Sorry if I stepped in to early...
If you wish, I can just review the final product :)

alecandido · 2022-11-29T13:23:18Z

Thank you, @adrianneschauss, for solving the major issue.

About the new modules asp1.py and apem1.py:

naming is currently inconsistent, use:
1. asp and aemp (preferred, by me), or
2. aps and apem
in any case, I would drop the p from the module, and just use the organization proposed above, with a further subdivision in folders between polarized and unpolarized

The idea is that just relying on a p is a bit confusing, and:

Explicit is better than implicit.

However, it is definitely not urgent, please take your time :)

felixhekhorn · 2022-12-05T17:28:42Z

and just use the organization proposed above, with a further subdivision in folders between polarized and unpolarized

definitely 🙃 remember that @t7phy is working in parallel on the time-like unpolarized part - so you should move into polarized/space_like/* and @t7phy into unpolarized/time_like/* (later we need also to move the current content to unpolarized/space_like)

alecandido · 2022-12-22T13:21:30Z

 from ..kernels import non_singlet as ns
 from ..kernels import singlet as s
 from ..member import OpMember
+from scipy import integrate


pre-commit (i.e. isort) should put it back in the former place, please run:

pre-commit install

in the local instance of your repository.

alecandido · 2022-12-22T13:23:45Z

    nf : int
        number of active flavors
+    p : Boolean
+        Polarised (True) or un-Polarised (False)


Unfortunately I believe that introducing a boolean is the correct way at the moment (or equally bad to introducing a parallel dispatcher), but eventually I believe we want to get rid of dispatchers at all, at least not keep using them as they are meant now (as a single entry point for Numba).

This will be addressed by refactoring after the analogue of #123 will be implemnted.

alecandido · 2022-12-22T13:26:02Z


 def test_gamma_ns():
    nf = 3
+    p = False


As already noticed in #184 these tests contains a lot of repetitions. We want to refactor limiting the amount of repetitions as you would for normal code.

E.g. here you can do it by looping over orders and outputs, without repeating very similar calls over and over.

alecandido · 2022-12-22T13:28:31Z

+    -------
+      gamma_qg : complex
+        Leading-order polarised quark-gluon anomalous dimension :math:`\\gamma_{qg}^{(0)}(N)`
+    """


This is more targeted to @felixhekhorn: I propose to drop Parameters and Returns sections, since they are quite verbose and redundant.
We will replace by the specification of a common interface (or a scheme for possible interfaces) in the module/subpackage docstrings.

With "interface" I mean the io = parameters + returns

this is not so trivial, since many of the higher order functions require a non-trivial combination of S-es

But they were getting access to an S-cache, isn't it?

However, in case they are variables, we can agree on a naming scheme for them. Once they adhere to the naming scheme, they do not need to be documented one by one.

alecandido · 2022-12-22T13:30:58Z

+
+
+
+
+
+
+
+
+
+
+
+
+
+
+# #declare parameter values
+# order= (1, 0)
+# n = complex(1.0, 1.0)
+# nf= 4
+# p = False
+# a1= 2
+# a0 = 1
+# #modes used
+# mode0=br.non_singlet_pids_map["ns+"]
+# mode1=0
+
+# #using anomalous dimensions singlet, make that matrix 
+# gamma_s= gamma_singlet(order, n, nf, p)
+# #make the evolution kernel with starting and ending energy 
+# j00= np.log(a1 / a0) / beta.beta_qcd((2, 0), nf)
+# lo_ex = ad.exp_singlet(gamma_s[0] * j00)
+# print (lo_ex)
+# ker_s = lo_ex
+# print(np.shape(ker_s))
+# k = 0 if mode0 == 100 else 1 
+# l = 0 if mode1 == 100 else 1
+# print(k)
+# print(l)
+# select_el= ker_s[k]
+# print (select_el)
+
+# spec = [
+#     ("is_singlet", nb.boolean),
+#     ("is_log", nb.boolean),
+#     ("logx", nb.float64),
+#     ("u", nb.float64),
+# ]


I guess this has been copied from somewhere.

We can drop here, and in case also in the place this was taken from.

alecandido · 2022-12-22T13:35:41Z

            as0=2,
            as_raw=1,
            nf=3,
+            p=False,


Since the fix is repeated, you see the pattern. If we see the pattern, let's refactor this @felixhekhorn

alecandido · 2022-12-22T13:46:29Z

The style is rather similar to the current one of EKO, and generally good. Most possible improvements are shared with the rest of the package, apart from negotiating the structure of this new part.

Co-authored-by: Alessandro Candido <candido.ale@gmail.com>

giacomomagni · 2023-01-04T13:40:21Z

+# @nb.njit(cache=True)
+# def gamma_pns(N, s1):
+# Computes the leading-order non-singlet anomalous dimension for the polarised case.
+# This is going to be the same expression as the one for the unpolarised case.
+
+# gamma = -(3.0 - 4.0 * s1 + 2.0 / N / (N + 1.0))
+# result = constants.CF * gamma
+# return result


Suggested change

# @nb.njit(cache=True)

# def gamma_pns(N, s1):

# Computes the leading-order non-singlet anomalous dimension for the polarised case.

# This is going to be the same expression as the one for the unpolarised case.

# gamma = -(3.0 - 4.0 * s1 + 2.0 / N / (N + 1.0))

# result = constants.CF * gamma

# return result

Co-authored-by: Giacomo Magni <39065935+giacomomagni@users.noreply.github.com>

…polarised_implementation

giacomomagni · 2023-01-16T13:51:11Z

clesed in favor of #195

Adrianne Schaus and others added 3 commits November 22, 2022 12:35

test

4216e83

Anomalous dimensions changed

38d624a

Merge branch 'develop' into polarised_implementation

e4a6cb5

giacomomagni assigned adrianneschauss Nov 26, 2022

giacomomagni added physics new physics features enhancement New feature or request labels Nov 26, 2022

giacomomagni marked this pull request as draft November 26, 2022 13:37

Adrianne Schaus and others added 2 commits November 29, 2022 13:51

Changed back to just eko, and modified some files further

9c19587

changes made to anomalous dimensions

a813e69

felixhekhorn linked an issue Dec 6, 2022 that may be closed by this pull request

polarized setup #40

Closed

formatting, tests, src changed and 'p' added

ec9a750

alecandido requested changes Dec 22, 2022

View reviewed changes

adrianneschauss and others added 4 commits December 23, 2022 11:11

Update src/eko/evolution_operator/__init__.py

d92b481

Co-authored-by: Alessandro Candido <candido.ale@gmail.com>

Update README.md

6bc0cd3

Co-authored-by: Alessandro Candido <candido.ale@gmail.com>

Update .gitignore

a86b91c

Co-authored-by: Alessandro Candido <candido.ale@gmail.com>

formatting, tests, src changed and 'p' added

7994f7f

giacomomagni commented Jan 4, 2023

View reviewed changes

felixhekhorn mentioned this pull request Jan 6, 2023

Use tempdir during execution #172

Merged

20 tasks

giacomomagni and others added 2 commits January 9, 2023 11:42

Merge branch 'develop' into polarised_implementation

415d27f

Update src/eko/anomalous_dimensions/aem1.py

b6bc093

Co-authored-by: Giacomo Magni <39065935+giacomomagni@users.noreply.github.com>

felixhekhorn mentioned this pull request Jan 11, 2023

Reorganize package content #194

Merged

adrianneschauss and others added 3 commits January 11, 2023 15:02

Update src/eko/anomalous_dimensions/asp1.py

120d024

Co-authored-by: Giacomo Magni <39065935+giacomomagni@users.noreply.github.com>

signs of splitting func, uses theory card input

027b07e

theory card input

4197e7d

Merge branch 'polarised_implementation' of github.com:NNPDF/eko into …

a39b41d

…polarised_implementation

giacomomagni closed this Jan 16, 2023

giacomomagni deleted the polarised_implementation branch January 31, 2023 09:04

Conversation

giacomomagni commented Nov 26, 2022

Uh oh!

alecandido commented Nov 26, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

giacomomagni commented Nov 26, 2022

Uh oh!

alecandido commented Nov 26, 2022

Uh oh!

alecandido commented Nov 29, 2022

Uh oh!

felixhekhorn commented Dec 5, 2022

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

alecandido commented Dec 22, 2022

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

giacomomagni commented Jan 16, 2023

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

alecandido commented Nov 26, 2022 •

edited

Loading