Subsetting method for mmkin objects

Subsetting method for mmkin objects.

# S3 method for mmkin
[(x, i, j, ..., drop = FALSE)

Arguments

x	An mmkin object
i	Row index selecting the fits for specific models
j	Column index selecting the fits to specific datasets
...	Not used, only there to satisfy the generic method definition
drop	If FALSE, the method always returns an mmkin object, otherwise either a list of mkinfit objects or a single mkinfit object.

Value

An object of class mmkin.

Examples

  # Only use one core, to pass R CMD check --as-cran
  fits <- mmkin(c("SFO", "FOMC"), list(B = FOCUS_2006_B, C = FOCUS_2006_C),
                cores = 1, quiet = TRUE)
  fits["FOMC", ]
#>       dataset
#> model  B       C      
#>   FOMC List,45 List,45
#> attr(,"class")
#> [1] "mmkin"
  fits[, "B"]
#>       dataset
#> model  B      
#>   SFO  List,45
#>   FOMC List,45
#> attr(,"class")
#> [1] "mmkin"
  fits["SFO", "B"]
#>      dataset
#> model B      
#>   SFO List,45
#> attr(,"class")
#> [1] "mmkin"

  head(
    # This extracts an mkinfit object with lots of components
    fits[["FOMC", "B"]]
  )
#> $par
#>  parent_0 log_alpha  log_beta 
#> 99.666193  2.549849  5.050586 
#> 
#> $ssr
#> [1] 28.58291
#> 
#> $convergence
#> [1] 0
#> 
#> $iterations
#> [1] 21
#> 
#> $evaluations
#> function gradient 
#>       25       78 
#> 
#> $counts
#> [1] "relative convergence (4)"
#> 

  head(
  # The same can be achieved by
    fits["SFO", "B", drop = TRUE]
  )
#> [[1]]
#> $par
#>          parent_0 log_k_parent_sink 
#>         99.174072         -2.549028 
#> 
#> $ssr
#> [1] 30.65564
#> 
#> $convergence
#> [1] 0
#> 
#> $iterations
#> [1] 5
#> 
#> $evaluations
#> function gradient 
#>        8       15 
#> 
#> $counts
#> [1] "relative convergence (4)"
#> 
#> $hessian
#>                     parent_0 log_k_parent_sink
#> parent_0            4.163631         -94.09343
#> log_k_parent_sink -94.093431        6311.34610
#> 
#> $residuals
#>      parent      parent      parent      parent      parent      parent 
#>  0.55407218 -2.98452128  4.20445742 -1.68599939 -0.58185357 -0.72033730 
#>      parent      parent 
#> -0.24260405 -0.07020339 
#> 
#> $ms
#> [1] 3.831956
#> 
#> $var_ms
#>   parent 
#> 3.831956 
#> 
#> $var_ms_unscaled
#>   parent 
#> 3.831956 
#> 
#> $var_ms_unweighted
#>   parent 
#> 3.831956 
#> 
#> $rank
#> [1] 2
#> 
#> $df.residual
#> [1] 6
#> 
#> $solution_type
#> [1] "analytical"
#> 
#> $transform_rates
#> [1] TRUE
#> 
#> $transform_fractions
#> [1] TRUE
#> 
#> $method.modFit
#> [1] "Port"
#> 
#> $maxit.modFit
#> [1] "auto"
#> 
#> $calls
#> [1] 29
#> 
#> $time
#>        User      System verstrichen 
#>       0.174       0.000       0.173 
#> 
#> $mkinmod
#> <mkinmod> model generated with
#> Use of formation fractions $use_of_ff: min 
#> Specification $spec:
#> $parent
#> $type: SFO; $sink: TRUE
#> Coefficient matrix $coefmat available
#> Differential equations:
#> d_parent/dt = - k_parent_sink * parent
#> 
#> $observed
#>     name time value
#> 1 parent    0 98.62
#> 2 parent    3 81.43
#> 3 parent    7 53.18
#> 4 parent   14 34.89
#> 5 parent   30 10.09
#> 6 parent   62  1.50
#> 7 parent   90  0.33
#> 8 parent  118  0.08
#> 
#> $obs_vars
#> [1] "parent"
#> 
#> $predicted
#>       name       time       value
#> 1   parent   0.000000 99.17407218
#> 2   parent   1.191919 90.35253561
#> 3   parent   2.383838 82.31567498
#> 4   parent   3.000000 78.44547872
#> 5   parent   3.575758 74.99369333
#> 6   parent   4.767677 68.32300215
#> 7   parent   5.959596 62.24566915
#> 8   parent   7.000000 57.38445742
#> 9   parent   7.151515 56.70891509
#> 10  parent   8.343434 51.66465547
#> 11  parent   9.535354 47.06908288
#> 12  parent  10.727273 42.88228661
#> 13  parent  11.919192 39.06790599
#> 14  parent  13.111111 35.59281463
#> 15  parent  14.000000 33.20400061
#> 16  parent  14.303030 32.42683275
#> 17  parent  15.494949 29.54246504
#> 18  parent  16.686869 26.91466193
#> 19  parent  17.878788 24.52060198
#> 20  parent  19.070707 22.33949373
#> 21  parent  20.262626 20.35239512
#> 22  parent  21.454545 18.54204899
#> 23  parent  22.646465 16.89273320
#> 24  parent  23.838384 15.39012410
#> 25  parent  25.030303 14.02117212
#> 26  parent  26.222222 12.77398846
#> 27  parent  27.414141 11.63774182
#> 28  parent  28.606061 10.60256435
#> 29  parent  29.797980  9.65946594
#> 30  parent  30.000000  9.50814643
#> 31  parent  30.989899  8.80025617
#> 32  parent  32.181818  8.01747313
#> 33  parent  33.373737  7.30431867
#> 34  parent  34.565657  6.65459931
#> 35  parent  35.757576  6.06267251
#> 36  parent  36.949495  5.52339762
#> 37  parent  38.141414  5.03209124
#> 38  parent  39.333333  4.58448658
#> 39  parent  40.525253  4.17669637
#> 40  parent  41.717172  3.80517911
#> 41  parent  42.909091  3.46670832
#> 42  parent  44.101010  3.15834451
#> 43  parent  45.292929  2.87740968
#> 44  parent  46.484848  2.62146400
#> 45  parent  47.676768  2.38828471
#> 46  parent  48.868687  2.17584671
#> 47  parent  50.060606  1.98230508
#> 48  parent  51.252525  1.80597899
#> 49  parent  52.444444  1.64533711
#> 50  parent  53.636364  1.49898432
#> 51  parent  54.828283  1.36564963
#> 52  parent  56.020202  1.24417505
#> 53  parent  57.212121  1.13350565
#> 54  parent  58.404040  1.03268029
#> 55  parent  59.595960  0.94082335
#> 56  parent  60.787879  0.85713708
#> 57  parent  61.979798  0.78089471
#> 58  parent  62.000000  0.77966270
#> 59  parent  63.171717  0.71143411
#> 60  parent  64.363636  0.64815202
#> 61  parent  65.555556  0.59049888
#> 62  parent  66.747475  0.53797399
#> 63  parent  67.939394  0.49012119
#> 64  parent  69.131313  0.44652489
#> 65  parent  70.323232  0.40680649
#> 66  parent  71.515152  0.37062104
#> 67  parent  72.707071  0.33765429
#> 68  parent  73.898990  0.30761993
#> 69  parent  75.090909  0.28025713
#> 70  parent  76.282828  0.25532825
#> 71  parent  77.474747  0.23261679
#> 72  parent  78.666667  0.21192552
#> 73  parent  79.858586  0.19307474
#> 74  parent  81.050505  0.17590074
#> 75  parent  82.242424  0.16025436
#> 76  parent  83.434343  0.14599973
#> 77  parent  84.626263  0.13301305
#> 78  parent  85.818182  0.12118154
#> 79  parent  87.010101  0.11040244
#> 80  parent  88.202020  0.10058214
#> 81  parent  89.393939  0.09163535
#> 82  parent  90.000000  0.08739595
#> 83  parent  90.585859  0.08348439
#> 84  parent  91.777778  0.07605845
#> 85  parent  92.969697  0.06929305
#> 86  parent  94.161616  0.06312943
#> 87  parent  95.353535  0.05751406
#> 88  parent  96.545455  0.05239819
#> 89  parent  97.737374  0.04773737
#> 90  parent  98.929293  0.04349113
#> 91  parent 100.121212  0.03962259
#> 92  parent 101.313131  0.03609816
#> 93  parent 102.505051  0.03288723
#> 94  parent 103.696970  0.02996191
#> 95  parent 104.888889  0.02729679
#> 96  parent 106.080808  0.02486874
#> 97  parent 107.272727  0.02265667
#> 98  parent 108.464646  0.02064136
#> 99  parent 109.656566  0.01880531
#> 100 parent 110.848485  0.01713257
#> 101 parent 112.040404  0.01560863
#> 102 parent 113.232323  0.01422024
#> 103 parent 114.424242  0.01295535
#> 104 parent 115.616162  0.01180297
#> 105 parent 116.808081  0.01075310
#> 106 parent 118.000000  0.00979661
#> 
#> $cost
#> function (P) 
#> {
#>     assign("calls", calls + 1, inherits = TRUE)
#>     if (trace_parms) 
#>         cat(P, "\n")
#>     if (length(state.ini.optim) > 0) {
#>         odeini <- c(P[1:length(state.ini.optim)], state.ini.fixed)
#>         names(odeini) <- c(state.ini.optim.boxnames, state.ini.fixed.boxnames)
#>     }
#>     else {
#>         odeini <- state.ini.fixed
#>         names(odeini) <- state.ini.fixed.boxnames
#>     }
#>     odeparms <- c(P[(length(state.ini.optim) + 1):length(P)], 
#>         transparms.fixed)
#>     parms <- backtransform_odeparms(odeparms, mkinmod, transform_rates = transform_rates, 
#>         transform_fractions = transform_fractions)
#>     out <- mkinpredict(mkinmod, parms, odeini, outtimes, solution_type = solution_type, 
#>         use_compiled = use_compiled, method.ode = method.ode, 
#>         atol = atol, rtol = rtol, ...)
#>     assign("out_predicted", out, inherits = TRUE)
#>     mC <- modCost(out, observed, y = "value", err = err, weight = weight, 
#>         scaleVar = scaleVar)
#>     if (mC$model < cost.old) {
#>         if (!quiet) 
#>             cat("Model cost at call ", calls, ": ", mC$model, 
#>                 "\n")
#>         if (plot) {
#>             outtimes_plot = seq(min(observed$time), max(observed$time), 
#>                 length.out = 100)
#>             out_plot <- mkinpredict(mkinmod, parms, odeini, outtimes_plot, 
#>                 solution_type = solution_type, use_compiled = use_compiled, 
#>                 method.ode = method.ode, atol = atol, rtol = rtol, 
#>                 ...)
#>             plot(0, type = "n", xlim = range(observed$time), 
#>                 ylim = c(0, max(observed$value, na.rm = TRUE)), 
#>                 xlab = "Time", ylab = "Observed")
#>             col_obs <- pch_obs <- 1:length(obs_vars)
#>             lty_obs <- rep(1, length(obs_vars))
#>             names(col_obs) <- names(pch_obs) <- names(lty_obs) <- obs_vars
#>             for (obs_var in obs_vars) {
#>                 points(subset(observed, name == obs_var, c(time, 
#>                   value)), pch = pch_obs[obs_var], col = col_obs[obs_var])
#>             }
#>             matlines(out_plot$time, out_plot[-1], col = col_obs, 
#>                 lty = lty_obs)
#>             legend("topright", inset = c(0.05, 0.05), legend = obs_vars, 
#>                 col = col_obs, pch = pch_obs, lty = 1:length(pch_obs))
#>         }
#>         assign("cost.old", mC$model, inherits = TRUE)
#>     }
#>     return(mC)
#> }
#> <bytecode: 0x55555bd78ed0>
#> <environment: 0x55555bc7eda8>
#> 
#> $cost_notrans
#> function (P) 
#> {
#>     if (length(state.ini.optim) > 0) {
#>         odeini <- c(P[1:length(state.ini.optim)], state.ini.fixed)
#>         names(odeini) <- c(state.ini.optim.boxnames, state.ini.fixed.boxnames)
#>     }
#>     else {
#>         odeini <- state.ini.fixed
#>         names(odeini) <- state.ini.fixed.boxnames
#>     }
#>     odeparms <- c(P[(length(state.ini.optim) + 1):length(P)], 
#>         parms.fixed)
#>     out <- mkinpredict(mkinmod, odeparms, odeini, outtimes, solution_type = solution_type, 
#>         use_compiled = use_compiled, method.ode = method.ode, 
#>         atol = atol, rtol = rtol, ...)
#>     mC <- modCost(out, observed, y = "value", err = err, weight = weight, 
#>         scaleVar = scaleVar)
#>     return(mC)
#> }
#> <bytecode: 0x55555ca99e50>
#> <environment: 0x55555bc7eda8>
#> 
#> $hessian_notrans
#>                   parent_0 k_parent_sink
#> parent_0          4.163631     -1203.894
#> k_parent_sink -1203.893702   1033188.753
#> 
#> $start
#>               value   type
#> parent_0      98.62  state
#> k_parent_sink  0.10 deparm
#> 
#> $start_transformed
#>                       value lower upper
#> parent_0          98.620000  -Inf   Inf
#> log_k_parent_sink -2.302585  -Inf   Inf
#> 
#> $fixed
#> [1] value type 
#> <0 Zeilen> (oder row.names mit Länge 0)
#> 
#> $data
#>   time variable observed   predicted    residual
#> 1    0   parent    98.62 99.17407218 -0.55407218
#> 2    3   parent    81.43 78.44547872  2.98452128
#> 3    7   parent    53.18 57.38445742 -4.20445742
#> 4   14   parent    34.89 33.20400061  1.68599939
#> 5   30   parent    10.09  9.50814643  0.58185357
#> 6   62   parent     1.50  0.77966270  0.72033730
#> 7   90   parent     0.33  0.08739595  0.24260405
#> 8  118   parent     0.08  0.00979661  0.07020339
#> 
#> $atol
#> [1] 1e-08
#> 
#> $rtol
#> [1] 1e-10
#> 
#> $weight.ini
#> [1] "none"
#> 
#> $tc.ini
#> sigma_low  rsd_high 
#>      0.50      0.07 
#> 
#> $reweight.tol
#> [1] 1e-08
#> 
#> $reweight.max.iter
#> [1] 10
#> 
#> $bparms.optim
#>      parent_0 k_parent_sink 
#>   99.17407218    0.07815759 
#> 
#> $bparms.fixed
#> numeric(0)
#> 
#> $bparms.ode
#> k_parent_sink 
#>    0.07815759 
#> 
#> $bparms.state
#>   parent 
#> 99.17407 
#> 
#> $date
#> [1] "Fri Sep 14 17:41:26 2018"
#> 
#> $version
#> [1] "0.9.47.5"
#> 
#> $Rversion
#> [1] "3.5.1"
#> 
#> attr(,"class")
#> [1] "mkinfit" "modFit" 
#>