Re-examine iterations in integral for "exact"

Open
#578 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
5/5
Estimated time
Over a week
Newbie friendliness
25/100
Issue type
Refactor
Clarity
Needs clarification
Activity status
Active
Tech stack
python
Domain
backend

Research direction

Start by locating the iterate-exact evolution integral and its current fixed subdivision in steps of ev_op_iteration over ln(a_s). Compare subdivisions in ln(a_s), a_s, and fixed step sizes, as suggested in the issue, and study how their results converge. The issue gives no file or test names; done means selecting a converged reference method and measuring how fewer steps affect accuracy and numerical stability.

Written by the indexing model from the issue text.

Description

physics

We should study the subdivision of the evolution integral for "iterate-exact". The current algorithm is a fixed subdivision irrespective of the evolution length with ev_op_iteration steps in $\ln(a_s)$. This might not be the best way and Marco Bonvini pretty much confirmed this. He suggested to compare different methods until they converge, which one can then consider as the reference "truth" and then subsequently reduce the number of steps and measure the distance to that reference.

Other viable options are subdivisions in $a_s$ and/or a fixed step size.

The problem is becoming quickly a bit messy with all the different options ... still to be more numerically stable, we should do it at some point. The current implementation is the naive one, which we choose back in the days.

Dominant language
Python
Stars
13
Forks
9
Avg merge
6d 16h
Merged PRs (30d)
3

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from NNPDF/eko

All issues in NNPDF/eko

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.