Redshift IC Check

This checks in the initial conditions to see if we have a redshift present. If we do, then we compare this against the redshift from the parameter file.

Redshift was chosen to prevent conflicts between the use of the Time varibale between SWIFT and Gadget-oid codes.

Let me know if this implementation is utter crap, or if there's anything that I need to change.

This change has been requested from users in the COLIBRE fork.

marked as a Work In Progress

changed the description

added 2 commits

• 3db112cf - Replaced calls to error with message
• e2861073 - Added improved implementation of the graceful attribute read.

Compare with previous version

unmarked as a Work In Progress

added 1 commit

• 685ec980 - Clarified comment slightly in io_read_attribute_graceful

Compare with previous version

Don't you want to compare Header/Time ? That's the only thing defined in Gadget snapshots.

I have stayed away from that because we do not know if it references 'real' Time (as given by SWIFT) or 'Time' as scale-factor

Right but then Redshift is usually not defined in any ICs header generated by any ICs code that I know of.

I don't know where the SmallCosmoVolume ICs are from, but they contain the Header/Redshift parameter

We could check against the current time and scale factor, should we read the Time variable, but I would be worried about edge cases. We would also have to invoke the unit system, should the internal and IC units be different.

Yes, it's tricky. I don't know what was used to generate the SmallCosmoVolume ICs. They have possibly been processed by some version of Gadget-3 to apply the 2LPT correction. As we did for EAGLE.

It's hard to come up with something generic that is not going to backfire given that there is no agreed-upon standard.

How about:

• Check for Cosmology/Scale-Factor - if there then pass
• Check for Header/Redshift - if there then pass
• Check for Code/Code == SWIFT - if there then we know Header/Time is Scale Factor
• Check for Header/Time - if nonzero then it's probably a and we can check against it.

It's pretty nasty, though.

• /Cosmology/ only exists in SWIFT output.
• /Header/Redshift/ will exist in the output of some flavours of G-3
• If Code/Code == SWIFT (or /Header/Code == SWIFT) then it's a SWIFT output and Header/Time actually contains the time since Big Bang.

All we can do I think is:

• Read header/Time and assume it is a as per Gadget convention. OR
• Impose the Redshift / Scale-factor as part of the file format we ask for. And hence break compatibility with the IC codes we currently use.

We could require these things, or we could just add this as a check for people doing something silly; I don't think this requires drastic measures. I think if people are using snapshots without any actual consistent metadata in them and using inconsistent values in their parameter files then it's their bad.

Right, so which check? The Header/Time one?

I would advocate adding the Redshift check but we cannot add a Time one as it is inconsistent between different codes

Ok. Let's do that.

I'd argue we should do this in the read_ic_xxx() functions. We have a from the parameter file there already.

Shall I propose these changes?

I’d like to do it please

added 1 commit

• bc5328fd - Moved assertion of valid initial redshift to io functions

Compare with previous version

Ok I have moved the logic to its own function in common_io.c and called these functions in all of the other *_io.c files.

I'll rebase if you're happy with the changes

added 66 commits

• bc5328fd...1795cab8 - 59 commits from branch master
• d9e4c3e1 - Added graceful function to read attributes. If failing, this doesn't report anything.
• 4e251c47 - Now read the redshift from snapshot if present, and check against the one from the parameter file
• 09f3f468 - Replaced calls to error with message
• bde32815 - Added improved implementation of the graceful attribute read.
• 3a4a7865 - Clarified comment slightly in io_read_attribute_graceful
• fe15a841 - Moved assertion of valid initial redshift to io functions
• e5cc81c2 - Code formatting

Compare with previous version

Looks good to me. So we settle on checking only Header/Redshift?

Do you want to add to the RTD that the ICs must match the param file?

I'm happy to settle on only that if you agree. If we need to change this later, it can easily be added to the function I added in common_io, but this should silence the complaints for now

Ok. Let's go with that.

Documentation added.

added 1 commit

• c1f3913a - Added documentation for redshift in ICs

Compare with previous version

added 10 commits

• c1f3913a...c2d7c869 - 2 commits from branch master
• 7e3f9898 - Added graceful function to read attributes. If failing, this doesn't report anything.
• 951bf467 - Now read the redshift from snapshot if present, and check against the one from the parameter file
• be33f0ce - Replaced calls to error with message
• 0c77332c - Added improved implementation of the graceful attribute read.
• 89297fdc - Clarified comment slightly in io_read_attribute_graceful
• f32be0df - Moved assertion of valid initial redshift to io functions
• c5e51cd0 - Code formatting
• 8d97d615 - Added documentation for redshift in ICs

Compare with previous version

added 1 commit

• 46e7975c - Clarifications in the RTD.

Compare with previous version

merged

mentioned in commit cd24775e