Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

upsampfac=1.25 with tol<1e-9 reports warnings for type=1,2 but errors out for type=3 #500

Closed
ahbarnett opened this issue Jul 22, 2024 · 2 comments
Closed

upsampfac=1.25 with tol<1e-9 reports warnings for type=1,2 but errors out for type=3 #500

ahbarnett opened this issue Jul 22, 2024 · 2 comments
Assignees
@ahbarnett

Comments

@ahbarnett
Copy link
Collaborator

The ier handling logic for type 3 in CPU code needs fixing so that if user requests tol<1e-9 then ns=16 is chosen and a warning is given, and the right answer (to about 9.5 digits) is returned. This is fine in type 1,2, but doesn't work in type 3.
@ahbarnett ahbarnett self-assigned this Jul 22, 2024
@ahbarnett
Copy link
Collaborator Author

turns out this was a prob in test/finufft1d_test.cpp.

ahbarnett added a commit that referenced this issue Jul 22, 2024
@ahbarnett
fix 1dtest issue re allowing ier=1 (warning only), #500
79069e5
@ahbarnett
Copy link
Collaborator Author

(will be fixed by PR #499)

DiamonDinoia pushed a commit that referenced this issue Jul 23, 2024
@ahbarnett
regenerate kernel Horner coeffs: scaled to max val 1, and shave off s…
3e231e2 
…ome degrees for upsampfac=1.25 (#499)

* gen_all_horner_cpp_header.m 1st try

* new ES kernel generation

* regen ES kernel with max size 1 not exp(beta)

* fix constexpr in matlab-gen ES kernel code

* done rescaling ES kernel to max value 1, regen all coeffs, fixes #454

* src/ker_horner_allw_loop_constexpr.c

* add 1 to degree for larger-w upsampfac=1.25 ker eval, recovers prior accuracy

* fix 1dtest issue re allowing ier=1 (warning only), #500

* more automated matlb kernel coeff generation

* automated and documented matlab kernel coefficient generation; tweaked the poly degrees for kernel, checked accuracy

* regen ker coeffs .h without nc200() etc func; use .size() instead, much neater

* reverted matlab before clang-format messed it up

* clang-formatted regen with assert logic fixed in .h horner stuff
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@ahbarnett ahbarnett

Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@DiamonDinoia @ahbarnett