Version 7 (modified by pkienzle, 7 years ago) (diff)

Autogenerated parts of the model

  • [PAK-9 #332] fix cos(theta) issue on angular dispersion
  • [PAK-9] mixture models/product models
  • [PAK-8] tied parameters
  • [PAK-7 #363] linearize for loop to avoid OpenCL problems when model runs too long
  • [PAK-5] vector parameters
  • [PAK-0] magnetism on each of the sld parameters
  • [WP-?] autogenerate figures for docs from demo parameters (or maybe default)
  • [PAK?-7] need an easy way to reparameterize an existing model
  • [PAK?-?] can't currently reference other models from within one model
  • #492 reparameterize orientation

Existing models

  • #364, #377, #439, #410, #288, #347, #472, #484 fix problems with specific models (see tickets)
  • [various] clean up model code (e.g., use J1c, sinc, etc.)
  • WP move from NR J1 (higher precision—current is 1e-9;licensing issues) l
  • [steve, richard?] check docs #19

Integration

  • slowly evolve the sasview model api
  • #505 Thoroughly test sasview with new sasmodels (e.g., is sas.models.SubCompenent? needed/supported)
  • #348 preserve parameter order using ordered dict, with removing sort in GUI
  • #506 introspect to find available models
  • redo interface to resolution within sasview GUI so that it doesn't recalc q every time
  • show min max in gui

GUI aspects

  • show components separately for product and sum models
  • weighted sum of several models (mixture models) to avoid e.g., p1_p2_radius
  • can we do mixture models "on the fly"
  • #504 generate new model.py file from GUI
  • #411 no stop button for constrained fits
  • #473

Bumps improvements

  • #270 check bumps error bars
  • #456 better handling of bumps plots

Parameter enhancements

  • display ER and VR when available
  • access to effective radius in constraints
  • don't use automatic constraint to set effective radius in product model
  • allow model to define the derived parameters (possibly polydisperse)
  • allow parameters not part of existing models for use in constraints (e.g., by having a zero model with k parameters)

Fitting Enhancements

  • integer parameters not fit properly
  • deweighting SAXS fits
  • downsampling SAXS data
  • assign cost to structure in the residuals
  • mask of middle part of data (e. g., spurions)