Replace ProSpectSEDlike manual parameter plumbing with ParmOff

[Copilot]

ProSpectSEDlike was still manually handling argument dispatch, bounds (lower/upper), and log-space transforms (logged), which diverged from current ParmOff capabilities (including logical-vector .logged). This updates the likelihood path to use ParmOff as the single parameter/dispatch mechanism with .check = FALSE.

• Parameter transform/bounds delegation

  • Removed in-function manual interval clipping and log10 expansion logic.
  • Added a ParmOff::ParmOff(..., .return = "args", .check = FALSE) preprocessing step to produce the final scalar parameter vector used by the model call.
  • Mapped Data$logged to name-based selection so both scalar and logical-vector cases are handled consistently (including z in photo-z mode).

• ProSpectSED invocation delegation

  • Replaced both do.call("ProSpectSED", ...) call sites (returnall = TRUE/FALSE) with ParmOff::ParmOff(.func = ProSpectSED, .return = "func", .check = FALSE).
  • Preserved existing call payload structure (SFH, speclib, Dale, AGN, filtout, returnall, Data$arglist), while centralizing argument matching and filtering through ParmOff.

• Behavior-preserving guardrails

  • Kept Data$constraints application order before bounds/log transforms.
  • Kept downstream scat_scale, likelihood, monitor, and return-shape behavior intact.
  • Added explicit scalar-parameter guard after ParmOff arg extraction.

parmlist <- ParmOff::ParmOff(
  .func   = ProSpectSED,
  .args   = as.list(parm),
  .lower  = parm_lower,
  .upper  = parm_upper,
  .logged = logged_parm_names,
  .return = "args",
  .check  = FALSE
)$current_args

[asgr]

All good with the switch to ParmOff

[Copilot]

Pull request overview

This PR refactors ProSpectSEDlike() to delegate parameter bounds and log-space transforms, plus ProSpectSED argument dispatch, to ParmOff, aiming to remove manual parameter “plumbing” and better support logical-vector .logged.

Changes:

• Added a ParmOff-based preprocessing step to generate the final scalar parameter vector passed into ProSpectSED.
• Replaced both do.call("ProSpectSED", ...) sites with ParmOff::ParmOff(..., .return = "func") to centralize argument matching/filtering.
• Updated photo-z “z is logged” handling to use name-based logged_parm_names.

Comments suppressed due to low confidence (2)

R/ProSpect.R:644

• scat_scale extraction currently uses parmlist before parmlist is created by the ParmOff::ParmOff(..., .return='args') call, which will error on the first evaluation when scat_scale is in Data$parm.names. Extract scat_scale from the (possibly constrained/bounded/logged) parameter vector after it’s been processed, or process scat_scale separately before building the ProSpectSED argument list.

This issue also appears on line 649 of the same file.

  if('scat_scale' %in% Data$parm.names){
    sel = which('scat_scale' == Data$parm.names)
    scat_scale = parmlist[sel]
    parmlist = parmlist[-sel]
    Data$parm.names = Data$parm.names[-sel]

R/ProSpect.R:656

• After removing scat_scale from Data$parm.names, parm_lower/parm_upper are still created from Data$intervals$lo/hi (which likely still include scat_scale). Assigning names(parm_lower) = Data$parm.names will fail if the lengths differ. Build parm_lower/upper from the same set of parameters you intend to pass to ParmOff (or drop scat_scale from the intervals before converting/naming).

  parm_lower = NULL
  parm_upper = NULL
  if (!is.null(Data$intervals)) {
    parm_lower = as.list(Data$intervals$lo)
    parm_upper = as.list(Data$intervals$hi)
    names(parm_lower) = Data$parm.names
    names(parm_upper) = Data$parm.names
  }

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.