You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am re-opening this issue because refactor still needs some adjustments, I believe.
Firstly, there’s an if statement that appears in several Recon1d_ modules:
if ( ( u2 - u1 ) * ( u1 - u0 ) <- 0.0 )
I believe the intended condition was:
if ( ( u2 - u1 ) * ( u1 - u0 ) <= 0.0 )
Additionally, and more critically, we’re seeing NaN values for u2 in the C_PPM_CW case:
do K = 2, n ! K=2 is interface between cells 1 and 2
...
u2 = u(k+1)
when k = n. This is likely due to an off-by-one access error.
Since switching from PPM_CW to C_PPM_CW isn’t critical for the next beta tag, I’ve reverted that change in #298. Once these issues are verified and resolved, we can reintroduce the update safely.
That <- is indeed a typo. @marshallward pointed this out to me a while before the shutdown. I've just checked that changing to "<=" makes no difference to the tests, but I have not yet done a regression test with OM4. I think it will pass.
Activity
alperaltuntas commentedon Oct 21, 2025
I am re-opening this issue because refactor still needs some adjustments, I believe.
Firstly, there’s an if statement that appears in several
Recon1d_modules:if ( ( u2 - u1 ) * ( u1 - u0 ) <- 0.0 )I believe the intended condition was:
if ( ( u2 - u1 ) * ( u1 - u0 ) <= 0.0 )Additionally, and more critically, we’re seeing NaN values for u2 in the
C_PPM_CWcase:when
k = n. This is likely due to an off-by-one access error.Since switching from
PPM_CWtoC_PPM_CWisn’t critical for the next beta tag, I’ve reverted that change in #298. Once these issues are verified and resolved, we can reintroduce the update safely.Mentioning @adcroft
adcroft commentedon Oct 21, 2025
That
<-is indeed a typo. @marshallward pointed this out to me a while before the shutdown. I've just checked that changing to "<=" makes no difference to the tests, but I have not yet done a regression test with OM4. I think it will pass.