Tinker with README

README.md

@@ -4,18 +4,20 @@
 This package is a Julia complement to [LyoPRONTO](https://github.com/LyoHUB/LyoPronto), an open source Python package.
 It has some overlapping functionality with LyoPRONTO, especially simulation of primary drying for conventional lyophilization.
 LyoPRONTO (the Python version) also has functionality for generating a design space, estimating time to freeze, and picking optimal drying conditions.
-On the other hand, this package has much more advanced utilities for fitting empirical parameters (such as $R_p$ and $K_v$) to experimental data. 
+On the other hand, this package has much more advanced utilities for fitting empirical parameters (such as $R_p$ and $K_v$) to experimental data.
+This package also provides that fitting functionality for a model applicable to microwave-assisted lyophilization.  
 
 ## Installation
 
-From the Julia REPL's Pkg mode (open a REPL and type `]` so that the prompt turns blue), add this package as a Git repo:
+From the Julia REPL's Pkg mode (open a REPL and type `]` so that the prompt turns blue), add this package from the General registry with:
 ```
-add https://github.com/LyoHUB/LyoPronto.jl.git
+add LyoPronto
 ```
 
+
 ## Documentation
 
-The "badge" up above is a link to the documentation.
+The "badge" up above is a link to the documentation, which is [also here](https://lyohub.github.io/LyoPronto.jl/).
 
 ## Versioning
 
@@ -28,7 +30,7 @@ This work was supported in part by funding for NIIMBL project PC4.1-307 .
 
 ## License
 
-None yet. My intentions are to use the MIT license once this has been published in a scientific journal.
+MIT License; see `LICENSE` file.
 
 # Example usage
 

docs/Project.toml

@@ -8,8 +8,6 @@ LineSearches = "d3d80556-e9d4-5f37-9878-2ab0fcc64255"
 Literate = "98b081ad-f1c9-55d3-8b20-4c87d4299306"
 LyoPronto = "86014da0-a5b3-40a4-ad76-3e66dfc3e269"
 NonlinearSolve = "8913a72c-1f9b-4ce2-8d82-65094dcecaec"
-Optim = "429524aa-4258-5aef-a3af-852621145aeb"
-Optimization = "7f7a1694-90dd-40f0-9382-eb1efda571ba"
 OptimizationOptimJL = "36348300-93cb-4f02-beb5-3c3902f8871e"
 Plots = "91a5bcdd-55d7-5caf-9e0b-520d859cae80"
 SavitzkyGolay = "c4bf5708-b6a6-4fbe-bcd0-6850ed671584"
@@ -20,6 +18,7 @@ TypedTables = "9d95f2ec-7b3d-5a63-8d20-e2491e220bb9"
 [compat]
 Plots = "1.41"
 NonlinearSolve = "4.9"
+OptimizationOptimJL = "0.4.6"
 
 [sources]
 LyoPronto = {path = ".."}

docs/example/fitting_mannitol.jl

@@ -10,8 +10,9 @@ using TypedTables, CSV
 # TransformVariables provides tools for mapping optimization parameters to sensible ranges.
 using TransformVariables
 # Optimization provides a common interface to a variety of optimization packages, including Optim.
+# We import it with OptimizationOptimJL to specify Optim as a backend.
 # LineSearches gives a little more granular control over solver algorithms for Optim.
-using Optimization, OptimizationOptimJL
+using OptimizationOptimJL
 using LineSearches
 # Or, instead of using a scalar optimization package, we can use a least-squares solver.
 using NonlinearSolve

src/LyoPronto.jl

@@ -8,6 +8,7 @@ import OrdinaryDiffEqRosenbrock: ODEProblem
 @reexport using Unitful: @u_str, ustrip, uconvert, NoUnits
 import Unitful
 using TransformVariables
+using TransformVariables: logit
 using RecipesBase
 using ColorTypes: RGB
 using CSV
@@ -23,6 +24,7 @@ using ADTypes: AutoForwardDiff
 using DocStringExtensions
 using LinearAlgebra: Diagonal
 
+
 abstract type ParamObj end
 
 const odealg_chunk1 = Rodas4(autodiff=AutoForwardDiff(chunksize=1))
@@ -54,6 +56,7 @@ export ParamObjRF
 # parameter fitting tools
 export gen_sol_pd, obj_pd, gen_nsol_pd, objn_pd
 export KRp_transform_basic, K_transform_basic, Rp_transform_basic, KBB_transform_basic
+export KBB_transform_bounded
 export obj_expT, err_expT, err_expT!, num_errs, nls_pd, nls_pd!
 export ConstWrapTV
 # plotting tools (mostly already done by macros)

src/paramfits.jl

@@ -43,6 +43,21 @@ function KBB_transform_basic(Kvwfg, Bfg, Bvwg)
     return tr
 end
 
+"""
+    $(SIGNATURES)
+Construct a bounded transform for fitting Kvwf, Bf, and Bvw (as for a microwave cycle).
+
+`Kvwf_scalefac`, `Bf_scalefac`, and `Bvw_scalefac` are used to provide upper and lower 
+bounds on the fitted parameter, as `(Kvwfg/Kvwf_scalefac, Kvwfg*Kvwf_scalefac)`, etc.
+This is enforced with a logistic transform scaled and shifted appropriately.
+"""
+function KBB_transform_bounded(Kvwfg, Bfg, Bvwg; Kvwf_scalefac=1e2, Bf_scalefac=1e4, Bvw_scalefac=1e4)
+    tr = as((Kvwf = TVScale(Kvwfg*Kvwf_scalefac) ∘ TVLogistic() ∘ TVShift(logit(inv(Kvwf_scalefac))),
+        Bf = TVScale(Bfg*Bf_scalefac) ∘ TVLogistic() ∘ TVShift(logit(inv(Bf_scalefac))),
+        Bvw = TVScale(Bvwg*Bvw_scalefac) ∘ TVLogistic() ∘ TVShift(logit(inv(Bvw_scalefac))) ))
+    return tr
+end
+
 """
     $(SIGNATURES)
 

test/Project.toml

@@ -4,8 +4,6 @@ JET = "c3a54625-cd67-489e-a8e7-0a5a0ff4e31b"
 LineSearches = "d3d80556-e9d4-5f37-9878-2ab0fcc64255"
 LyoPronto = "86014da0-a5b3-40a4-ad76-3e66dfc3e269"
 NonlinearSolve = "8913a72c-1f9b-4ce2-8d82-65094dcecaec"
-Optim = "429524aa-4258-5aef-a3af-852621145aeb"
-Optimization = "7f7a1694-90dd-40f0-9382-eb1efda571ba"
 OptimizationOptimJL = "36348300-93cb-4f02-beb5-3c3902f8871e"
 SavitzkyGolay = "c4bf5708-b6a6-4fbe-bcd0-6850ed671584"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
@@ -20,8 +18,6 @@ LyoPronto = {path = ".."}
 Aqua = "0.8"
 LineSearches = "7.3"
 NonlinearSolve = "4.9"
-Optim = "1.9"
-Optimization = "4.3"
-OptimizationOptimJL = "0.4"
+OptimizationOptimJL = "0.4.6"
 Test = "1"
 TypedTables = "1"

test/test_KRp_opt.jl

@@ -1,5 +1,5 @@
 using TransformVariables
-using Optimization, OptimizationOptimJL
+using OptimizationOptimJL
 using LineSearches
 using NonlinearSolve
 optalg = LBFGS(linesearch=LineSearches.BackTracking())

test/test_RF_opt.jl

@@ -1,5 +1,5 @@
 using TransformVariables
-using Optimization, OptimizationOptimJL
+using OptimizationOptimJL
 using LineSearches
 optalg = LBFGS(linesearch=LineSearches.BackTracking())