********  HRAI_PSEARCH.TXT -- 23 Dec 2003 - 3 Jan 2004  ********
**
**        PARAMETER SEARCH FOR A HIERARCHY OF ANCHOR MODELS
**        ABSOLUTE IDENTIFICATION, see HRCR_PSEARCH.TXT for category rating
**

P  -->  PS  -->  PSC  -->  PSCR  -->   PSACR=ANCHOR
            \->  PSA  -->  PSAC  /

Model P:     Perception mechanism only; nearest-anchor responses.
Model PS:    Perception + anchor Selection mechanisms only.
Model PSC:   Perception + Selection + Correction mechanisms only.
Model PSA:   Perception + Selection + Activation, no corrections.
Model PSAC:  All except location learning.
Model PSCR:  All except activation learning.
Model PSACR: Full ANCHOR model.

The supportive data are the following:
a) 24=12+12 empirical absolute-identification S-R sequences.
c) 40=20+20 empirical category-rating S-R sequences.

The general methodology is always the same:
1. Fit the corresponding model to the 24 individual data sets
   from the Absolute Identificaiton Experiment (Expmt 1).
   This produces 24 parameter sets.
   Which parameters are "active", which are shunted off, and
   which are fixed for all subjects depends on the particular model.
2. Run the model with each AI parametrization on each of the
   new AI stimulus sequences, for a total of 9600=24*400 runs.
3. Calculate a battery of model statistics via AI_STATS and
   compare them with the empirical ones.

This file carries out step 1. above.
The results from each parameter search are stored in HIER_OPT_PARAMS.MAT.
See HRAI_MSTATS.TXT for model statistics (steps 2. and 3.).\
The summary statistics for each model are stored in HIER_SUMSTATS.MAT.
See also HRCR_PSEARCH.TXT AND HRCR_MSTATS.TXT for category-rating analogs.


*************************************************************
***                ******************************************
***   Model P      ******************************************
***   Abs. Ident   ******************************************
***                ******************************************
*************************************************************

Active parameters:  perc_k
Shunted parameters: mem_k=0, temper=0, hist=0, cutoff=Inf, alpha=0
Fixed parameters:   perc_k=0.04, decay=0.5, L1=0.275, L9=0.675, avail=[1:9]'

% ANCHOR2MT, with its numerous discretizations and approximations,
% is not appropriate for optimizing the perceptual noise alone.
% The tailor made MODELP_MT is used instead.

 clear all ; cd 'C:\work\anchor\finalsim\' ;
 Sparams=anchor_search_params([1 0 0 0 0 0 0]) ;
 Sparams.model_name = 'modelP_mt' ;    % special-purpose optimizer
Sparams =
    model_name: 'modelP_mt'
    gener_name: 'anchor2'
        params: [1x1 struct]
     p2v_templ: {'VAL = PARAMS.perc_k * 10 ;'}
     v2p_templ: {'PARAMS.perc_k = VAL / 10 ;'}
        bounds: [2x1 double]
     clip_prob: 0.0183
         optns: [1x1 struct]

 Sparams.params.avail = [1:9]' ;   % all anchors available at all times
 Sparams.params.anchors(:,6)=1 ;   % all anchors available at all times
 Sparams.params.mem_k=0 ;          % deterministic anchor magnitudes
 Sparams.params.temper=0 ;         % deterministic anchor selection
 Sparams.params.cutoffs = [-Inf -Inf +Inf +Inf] ;   % no corrections
 Sparams.params.history = 0 ;      % shunting off the activation mechanism
 Sparams.params.alpha = 0 ;        % immutable anchor locations

% Widen up the search bounds for PERC_K
 Sparams.bounds(:,1)=[0.01 0.20]'*10 ;

 Sparams.params
       scale: 'LINEAR'
       N_cat: 9
     SM_conv: 0.0010
      cat_sz: 0.0500
      perc_k: 0.0400
       mem_k: 0
       avail: [9x1 double]
     anchors: [9x6 double]
     cutoffs: [-Inf -Inf Inf Inf]
      temper: 0
     history: 0
       alpha: 0
       decay: 0.5000
         ITI: 4
    M_raster: 7
    A_raster: [5 5 3 3]
    mnfieldp: 1

 Sparams_P_AI = Sparams ; clear Sparams

%%%%%%%%%%%%%%%%%%   Run the optimizer on the 24 AI data sets   %%%%%%%%%%%%%%%%%%
28 Dec 2003

 diary('session_P_AI.txt');disp('Model P, absolute identification');
 load AI24mdata ; Sparams_P_AI.fname='resAI24_P' ;
 tic;resAI24_P=big_psearch(AI24mdata,Sparams_P_AI);toc

% Elapsed_time = 21 sec on a Dell Inspiron 7500.

 resAI24_P
       model_name: 'modelP_mt'
       descriptor: '28-Dec-2003 16:02:06'
    search_params: [1x1 struct]
            opt_X: [24x1 double]
                L: [24x1 double]
         exitflag: [24x1 double]
          N_evals: [24x1 double]
          N_iters: [24x1 double]
             time: [24x1 double]

 resAI24_P.search_params.bounds'
ans = 0.1000    2.0000

 xtab1(resAI24_P.N_iters)
   Value   Count  Percent  Cum_cnt  Cum_pct
-------------------------------------------
       4       4    16.67        4    16.67
       5       5    20.83        9    37.50
       6       5    20.83       14    58.33
       7       8    33.33       22    91.67
       8       2     8.33       24   100.00
-------------------------------------------

 describe(resAI24_P.L)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
 500.912   73.550    351.10  462.30  509.35  559.44  616.69

 describe(resAI24_P.opt_X)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
   0.747    0.113      0.53    0.69    0.74    0.82    0.98  % mem_k * 10

 resAI24_P.opt_X'
[01:07]    0.6626    0.8655    0.7914    0.7428    0.7202    0.8278    0.7151
[08:14]    0.7470    0.8104    0.6087    0.5286    0.8062    0.9827    0.8640
[15:21]    0.5561    0.8348    0.7270    0.7298    0.7301    0.8031    0.9287
[22:24]    0.5777    0.6896    0.6847


*************************************************************
***                ******************************************
***   Model PS     ******************************************
***   Abs. Ident   ******************************************
***                ******************************************
*************************************************************

Active parameters:  mem_k, temper
Shunted parameters: hist=0, cutoff=Inf, alpha=0
Fixed parameters:   perc_k=0.04, decay=0.5,
                    L1=0.275, L9=0.675, avail=[1:9]'

 clear all ; cd 'C:\work\anchor\finalsim\' ;
 Sparams=anchor_search_params([0 1 1 0 0 0 0])
Sparams =
    model_name: 'anchor2mt'
    gener_name: 'anchor2'
        params: [1x1 struct]
     p2v_templ: {'VAL = PARAMS.mem_k * 10 ;'  'VAL = PARAMS.temper * 10 ;'}
     v2p_templ: {'PARAMS.mem_k = VAL / 10 ;'  'PARAMS.temper = VAL / 10 ;'}
        bounds: [2x2 double]
     clip_prob: 0.0183
         optns: [1x1 struct]

 Sparams.params.avail = [1:9]' ;   % all anchors available at all times
 Sparams.params.anchors(:,6)=1 ;   % all anchors available at all times
 Sparams.params.cutoffs = [-Inf -Inf +Inf +Inf] ;   % no corrections
 Sparams.params.history = 0 ;      % shunting off the activation mechanism
 Sparams.params.alpha = 0 ;        % immutable anchor locations

% Widen up the search bounds for MEM_K and TEMPER:
 Sparams.bounds(:,1)=[0.02 0.20]'*10;Sparams.bounds(:,2)=[0.01 0.10]'*10;

 Sparams.params
       scale: 'LINEAR'
       N_cat: 9
     SM_conv: 0.0010
      cat_sz: 0.0500
      perc_k: 0.0400
       mem_k: 0.0700
       avail: [9x1 double]
     anchors: [9x6 double]
     cutoffs: [-Inf -Inf Inf Inf]
      temper: 0.0500
     history: 0
       alpha: 0
       decay: 0.5000
         ITI: 4
    M_raster: 7
    A_raster: [5 5 3 3]
    mnfieldp: 1

 Sparams_PS_AI = Sparams ; clear Sparams

%%%%%%%%%%%%%%%%%%   Run the optimizer on the 24 AI data sets   %%%%%%%%%%%%%%%%%%
27 Dec 2003

 Sparams_PS_AI.fname='resAI24_PS';
 disp('Model PS, absolute identification');
 tic;resAI24_PS=big_psearch(AI24mdata,Sparams_PS_AI);toc

% Elapsed_time = 1.4530e+004 sec (4 hours) on a Dell Optiplex GX400.
% No "Divide by zero" warnings on any of the searches.

 resAI24_PS =
       model_name: 'anchor2mt'
       descriptor: '24-Dec-2003 08:07:08'
    search_params: [1x1 struct]
            opt_X: [24x2 double]
                L: [24x1 double]
         exitflag: [24x1 double]
          N_evals: [24x1 double]
          N_iters: [24x1 double]
             time: [24x1 double]

 resAI24_PS.search_params.v2p_templ'
ans = 'PARAMS.mem_k = VAL / 10 ;'
      'PARAMS.temper = VAL / 10 ;'

 xtab1(resAI24_PS.N_iters)
   Value   Count  Percent  Cum_cnt  Cum_pct
-------------------------------------------
       3       1     4.17        1     4.17
       4       2     8.33        3    12.50
       5       5    20.83        8    33.33
       6       4    16.67       12    50.00
       7       4    16.67       16    66.67
       8       1     4.17       17    70.83
      10       4    16.67       21    87.50
      11       3    12.50       24   100.00
-------------------------------------------

 describe(resAI24_PS.L)        % goodness of fit (-log likelihood)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
 489.541   71.695    344.78  444.15  495.74  550.17  592.02

 L=[resAI24_P.L resAI24_PS.L];find(L(:,1)<=L(:,2))'
ans = 21       % For all but one sbj, model PS fits better than model P.

% Allowing L1 to vary yields inferior fits:
%    Mean  Std.dev       Min     Q25  Median     Q75     Max
%------------------------------------------------------------
% 499.522   52.659    364.43  468.38  502.15  529.91  576.40


 describe(resAI24_PS.opt_X)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
   0.672    0.220      0.20    0.52    0.66    0.83    1.02  % mem_k * 10
   0.174    0.085      0.10    0.10    0.11    0.27    0.32  % temper* 10

% When L1 is allowed to vary it tends to be pushed DOWN,
% probably in an effort to suppress response "1" and thereby
% increase the non-uniformity of the response distribution.
%    Mean  Std.dev       Min     Q25  Median     Q75     Max
%------------------------------------------------------------
%   1.012    0.430      0.20    0.63    1.15    1.32    1.50  % mem_k * 10
%   0.141    0.059      0.10    0.10    0.10    0.19    0.26  % temper* 10
%   0.241    0.009      0.24    0.24    0.24    0.24    0.27  % anchor-1

 resAI24_PS.opt_X'
  Columns 1 through 7
    0.6678    1.0178    0.5959    0.4642    0.8880    0.9943    0.8110
    0.1000    0.1000    0.2679    0.2719    0.1000    0.1000    0.1000
  Columns 8 through 14
    0.8463    0.6500    0.5674    0.2000    0.5887    0.9265    0.5957
    0.1000    0.2869    0.1000    0.1413    0.2669    0.2750    0.2599
  Columns 15 through 21
    0.4698    0.3570    0.4120    0.6824    0.8154    0.6592    1.0100
    0.1000    0.3159    0.2520    0.1000    0.1000    0.2796    0.2232
  Columns 22 through 24
    0.4043    0.7849    0.7231
    0.1280    0.1000    0.1000


*************************************************************
***                ******************************************
***   Model PSC    ******************************************
***   Abs. Ident   ******************************************
***                ******************************************
*************************************************************

Active parameters:  mem_k, temper, cutoff
Shunted parameters: hist=0, alpha=0
Fixed parameters:   perc_k=0.04, decay=0.5,
                    L1=0.275, L9=0.675, avail=[1:9]'
                    cutoff multipliers=[-3 -1 +0.9 +2.7]

 clear all ; cd 'C:\work\anchor\finalsim\' ;
 Sparams=anchor_search_params([0 1 1 0 1 0 0])
Sparams =
    model_name: 'anchor2mt'
    gener_name: 'anchor2'
        params: [1x1 struct]
     p2v_templ: {1x3 cell}
     v2p_templ: {1x3 cell}
        bounds: [2x3 double]
     clip_prob: 0.0183
         optns: [1x1 struct]

 Sparams.params.avail = [1:9]' ;   % all anchors available at all times
 Sparams.params.anchors(:,6)=1 ;   % all anchors available at all times
 Sparams.params.history = 0 ;      % shunting off the activation mechanism
 Sparams.params.alpha = 0 ;        % immutable anchor locations

 Sparams.params
       scale: 'LINEAR'
       N_cat: 9
     SM_conv: 0.0010
      cat_sz: 0.0500
      perc_k: 0.0400
       mem_k: 0.0700
       avail: [9x1 double]
     anchors: [9x6 double]
     cutoffs: [-2.4000 -0.8000 0.7200 2.1600]
      temper: 0.0500
     history: 0
       alpha: 0
       decay: 0.5000
         ITI: 4
    M_raster: 7
    A_raster: [5 5 3 3]
    mnfieldp: 1

 Sparams_PSC_AI = Sparams ; clear Sparams

%%%%%%%%%%%%%%%%%%   Run the optimizer on the 24 AI data sets   %%%%%%%%%%%%%%%%%%
24 Dec 2003

 Sparams_PSC_AI.fname='resAI24_PSC';
 disp('Model PSC, absolute identification');
 tic;resAI24_PSC=big_psearch(AI24mdata,Sparams_PSC_AI);toc

% Elapsed_time = 1.9176e+004 sec (5 h 25 min) on a Dell Optiplex GX400.
% A few "Divide by zero" warnings on searches 14 and 16.

 resAI24_PSC =
       model_name: 'anchor2mt'
       descriptor: '24-Dec-2003 18:27:48'
    search_params: [1x1 struct]
            opt_X: [24x3 double]
                L: [24x1 double]
         exitflag: [24x1 double]
          N_evals: [24x1 double]
          N_iters: [24x1 double]
             time: [24x1 double]

 resAI24_PSC.search_params.v2p_templ'
ans = 'PARAMS.mem_k = VAL / 10 ;'
      'PARAMS.temper = VAL / 10 ;'
      'PARAMS.cutoffs = VAL*[-3 -1 +0.9 +2.7] ;'
 resAI24_PSC.search_params.bounds
    0.2000    0.2000    0.5000
    1.0000    0.8000    1.5000

 xtab1(resAI24_PSC.N_iters)
   Value   Count  Percent  Cum_cnt  Cum_pct
-------------------------------------------
       3       2     8.33        2     8.33
       4       1     4.17        3    12.50
       5       4    16.67        7    29.17
       6       1     4.17        8    33.33
       7      11    45.83       19    79.17
       8       3    12.50       22    91.67
      12       1     4.17       23    95.83
      13       1     4.17       24   100.00
-------------------------------------------

 describe(resAI24_PSC.L)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
 481.613   68.470    343.32  438.18  489.11  529.63  577.54

 L=[resAI24_PS.L resAI24_PSC.L];find(L(:,1)<=L(:,2))'
ans = 6 21        % For 22 of 24 Ss the addition of correction helps
 L([6 24],:)
ans = 513.6919  516.6148
      457.5583  454.0241

 describe(resAI24_PSC.opt_X)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
   0.588    0.315      0.20    0.27    0.58    0.89    1.00  % mem_k * 10
   0.376    0.195      0.20    0.20    0.25    0.56    0.76  % temper* 10
   0.898    0.232      0.50    0.72    0.86    1.07    1.50  % cutoff

 resAI24_PSC.opt_X'
  Columns 1 through 7
    0.4600    0.9971    0.2000    0.2000    0.9980    1.0000    0.8661
    0.2299    0.2000    0.6097    0.5488    0.2000    0.2000    0.2269
    1.1415    0.7686    1.0556    1.0796    0.7435    1.1215    0.7000
  Columns 8 through 14
    0.8917    0.3198    0.5645    0.4414    0.2000    1.0000    0.5900
    0.2000    0.7629    0.2442    0.2567    0.5532    0.6409    0.5406
    0.6878    1.0013    0.7260    0.5000    1.1133    0.8815    0.9932
  Columns 15 through 21
    0.3954    0.2149    0.2000    0.7598    0.8909    0.2000    1.0000
    0.2485    0.6565    0.5575    0.2000    0.2000    0.6130    0.2000
    0.8318    1.0700    1.0011    0.5532    0.6606    1.1224    1.5000
  Columns 22 through 24
    0.3210    0.6529    0.7580
    0.3353    0.3955    0.2000
    0.8089    0.7611    0.7221


*************************************************************
***                ******************************************
***   Model PSA    ******************************************
***   Abs. Ident   ******************************************
***                ******************************************
*************************************************************

Active parameters:  mem_k, temper, hist
Shunted parameters: cutoff=Inf, alpha=0
Fixed parameters:   perc_k=0.04, decay=0.5, default_anchors, avail=[1:9]'

 clear all ; cd 'C:\work\anchor\finalsim\' ;
 Sparams=anchor_search_params([0 1 1 1 0 0])
 Sparams.p2v_templ'
    'VAL = PARAMS.mem_k * 10 ;'
    'VAL = PARAMS.temper * 10 ;'
    'VAL = PARAMS.history * 5 ;'

 Sparams.params.avail = [1:9]' ;   % all anchors available at all times
 Sparams.params.anchors(:,6)=1 ;   % all anchors available at all times
 Sparams.params.cutoffs = [-Inf -Inf +Inf +Inf] ;   % no corrections
 Sparams.params.alpha = 0 ;        % immutable anchor locations

 Sparams.params
       scale: 'LINEAR'
       N_cat: 9
     SM_conv: 0.0010
      cat_sz: 0.0500
      perc_k: 0.0400
       mem_k: 0.0700
       avail: [9x1 double]
     anchors: [9x6 double]
     cutoffs: [-Inf -Inf Inf Inf]
      temper: 0.0500
     history: 0.1000
       alpha: 0
       decay: 0.5000
         ITI: 4
    M_raster: 7
    A_raster: [5 5 3 3]
    mnfieldp: 1

% Set the search bounds for HISTORY as for the full model later.
% (Otherwise the optimizer tends to suppress it down to zero, which
% defeats the purpose of this simulation---namely, to demonstrate
% the effects of the base-level activation mechanism on identification.)
 Sparams.bounds(:,3)=[0.05 0.20]'*5;Sparams.bounds
ans = 0.2000    0.2000    0.2500
      1.0000    0.8000    1.0000

 Sparams_PSA_AI = Sparams ; clear Sparams

%%%%%%%%%%%%%%%%%%   Run the optimizer on the 24 AI data sets   %%%%%%%%%%%%%%%%%%
2 Jan 2004

 Sparams_PSA.fname='resAI24_PSA';
 disp('Model PSA, absolute identification');
 tic;resAI24_PSA=big_psearch(AI24mdata,Sparams_PSA_AI);toc

% Elapsed_time = 1.1889e+004 sec (3 h 20 min) on a Dell Optiplex GX400.
% No "Divide by zero" warnings on any of the searches.

 resAI24_PSA =
       model_name: 'anchor2mt'
       descriptor: '02-Jan-2004 17:42:56'
    search_params: [1x1 struct]
            opt_X: [24x3 double]
                L: [24x1 double]
         exitflag: [24x1 double]
          N_evals: [24x1 double]
          N_iters: [24x1 double]
             time: [24x1 double]

 resAI24_PSA.search_params.v2p_templ'
ans = 'PARAMS.mem_k = VAL / 10 ;'
      'PARAMS.temper = VAL / 10 ;'
      'PARAMS.history = VAL / 5 ;'

 xtab1(resAI24_PSA.N_iters)     % From individualized starting points
   Value   Count  Percent  Cum_cnt  Cum_pct
-------------------------------------------
       2       4    16.67        4    16.67
       3       1     4.17        5    20.83
       5       4    16.67        9    37.50
       6       8    33.33       17    70.83
       7       5    20.83       22    91.67
       8       2     8.33       24   100.00
-------------------------------------------

 describe(resAI24_PSA.L)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
 513.249   64.605    379.56  472.78  522.58  566.71  602.97

%% Earlier search in which HISTORY was allowed to vary all the way
%  down to 0.0 yielded better fits, but not dramatically better.
%    Mean  Std.dev       Min     Q25  Median     Q75     Max
%------------------------------------------------------------
% 495.433   68.905    352.75  451.79  506.27  553.00  590.56

 describe(resAI24_PSA.opt_X)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
   0.425    0.235      0.20    0.21    0.39    0.58    1.00  % mem_k * 10
   0.255    0.051      0.20    0.20    0.25    0.31    0.33  % temper* 10
   0.250    0.000      0.25    0.25    0.25    0.25    0.25  % history* 5
% Note that in all cases HISTORY butts against the lower search bound =0.05*5.

%% Parameter summary for the earlier search:
%    Mean  Std.dev       Min     Q25  Median     Q75     Max
%------------------------------------------------------------
%   0.509    0.264      0.20    0.23    0.51    0.69    1.00
%   0.238    0.052      0.20    0.20    0.20    0.27    0.36
%   0.024    0.030      0.00    0.00    0.01    0.05    0.09  % <-- BLA shut down
% Note that the activation mechanism is not completely shut down---there are
% at least a few subjects that show sequential effects large enough to get
% the HISTORY parameter off the ground. Without corrections, things can easily
% get out of hand. Perhaps this explains the anomalously low HISTORY settings.

 resAI24_PSA.opt_X'
    0.2162    0.6718    0.2311    0.2528    0.6418    0.6172    0.5276
    0.2269    0.2587    0.3297    0.3050    0.2000    0.2529    0.2006
    0.2500    0.2500    0.2500    0.2500    0.2500    0.2500    0.2500
  Columns 8 through 14
    0.5539    0.6041    0.2189    0.2000    0.2000    0.9218    0.3288
    0.2530    0.3099    0.2000    0.2000    0.3334    0.3000    0.3129
    0.2500    0.2500    0.2500    0.2500    0.2500    0.2500    0.2500
  Columns 15 through 21
    0.2000    0.2000    0.2000    0.3121    0.4734    0.4959    1.0000
    0.2000    0.3258    0.2906    0.2634    0.2200    0.3164    0.2106
    0.2500    0.2500    0.2500    0.2500    0.2500    0.2500    0.2500
  Columns 22 through 24
    0.2000    0.4829    0.4575
    0.2000    0.2079    0.2000
    0.2500    0.2500    0.2500


*************************************************************
***                ******************************************
***   Model PSAC   ******************************************
***   Abs. Ident   ******************************************
***                ******************************************
*************************************************************

Active parameters:  mem_k, temper, hist, cutoff
Shunted parameters: alpha=0
Fixed parameters:   perc_k=0.04, decay=0.5,
                    L1=0.275, L9=0.675, avail=[1:9]'
                    cutoff multipliers=[-3 -1 +0.9 +2.7]

 clear all ; cd 'C:\work\anchor\finalsim\' ;
 Sparams=anchor_search_params([0 1 1 1 1 0 0])
Sparams =
    model_name: 'anchor2mt'
    gener_name: 'anchor2'
        params: [1x1 struct]
     p2v_templ: {1x4 cell}
     v2p_templ: {1x4 cell}
        bounds: [2x4 double]
     clip_prob: 0.0183
         optns: [1x1 struct]

 Sparams.params.avail = [1:9]' ;   % all anchors available at all times
 Sparams.params.anchors(:,6)=1 ;   % all anchors available at all times
 Sparams.params.alpha = 0 ;        % immutable anchor locations

 Sparams.params
       scale: 'LINEAR'
       N_cat: 9
     SM_conv: 0.0010
      cat_sz: 0.0500
      perc_k: 0.0400
       mem_k: 0.0700
       avail: [9x1 double]
     anchors: [9x6 double]
     cutoffs: [-2.4000 -0.8000 0.7200 2.1600]
      temper: 0.0500
     history: 0.1000
       alpha: 0
       decay: 0.5000
         ITI: 4
    M_raster: 7
    A_raster: [5 5 3 3]
    mnfieldp: 1

% Set the search bounds as for the full model later.
 Sparams.bounds(:,1)=[0.03 0.10]'*10;  % mem_k * 10
 Sparams.bounds(:,2)=[0.04 0.06]'*10;  % temper* 10
 Sparams.bounds(:,3)=[0.05 0.20]'*5 ;  % history* 5
 Sparams.bounds(:,4)=[0.50 1.10]' ;    % cutoff
 Sparams.bounds
ans = 0.3000    0.4000    0.2500    0.5000
      1.0000    0.6000    1.0000    1.1000

 Sparams_PSAC_AI = Sparams ; clear Sparams

%%%%%%%%%%%%%%%%%%   Run the optimizer on the 24 AI data sets   %%%%%%%%%%%%%%%%%%
3 Jan 2004

% Use the PSC optimal parameters as a starting point
 x0_PSC2PSAC_AI = resAI24_PSC.opt_X ;
 x0_PSC2PSAC_AI(:,4) = x0_PSC2PSAC_AI(:,3) ;    % cutoff is now in column 4
 x0_PSC2PSAC_AI(:,3) = 0.100 * 5 ;              % init = default history
 describe(x0_PSC2PSAC_AI)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
   0.588    0.315      0.20    0.27    0.58    0.89    1.00  % mem_k * 10
   0.376    0.195      0.20    0.20    0.25    0.56    0.76  % temper* 10
   0.500    0.000      0.50    0.50    0.50    0.50    0.50  % history* 5
   0.898    0.232      0.50    0.72    0.86    1.07    1.50  % cutoff

 Sparams_PSAC_AI.fname='resAI24_PSAC';
 disp('Model PSAC, absolute identification');
 tic;resAI24_PSAC=big_psearch(AI24mdata,Sparams_PSAC_AI,[],x0_PSC_AI);toc

% Elapsed_time = 2.5489e+004 sec (7 hours) on a Dell Optiplex GX400.
% Two "Divide by zero" warnings on search 14 and a whole barrage on search 16.

 resAI24_PSAC
       model_name: 'anchor2mt'
       descriptor: '31-Dec-2003 13:37:56'
    search_params: [1x1 struct]
            opt_X: [24x4 double]
                L: [24x1 double]
         exitflag: [24x1 double]
          N_evals: [24x1 double]
          N_iters: [24x1 double]
             time: [24x1 double]
               x0: [24x4 double]

 resAI24_PSAC.search_params.v2p_templ'
ans = 'PARAMS.mem_k = VAL / 10 ;'
      'PARAMS.temper = VAL / 10 ;'
      'PARAMS.history = VAL / 5 ;'
      'PARAMS.cutoffs = VAL*[-3 -1 +0.9 +2.7] ;'

 xtab1(resAI24_PSAC.N_iters)
   Value   Count  Percent  Cum_cnt  Cum_pct
-------------------------------------------
       5       2     8.33        2     8.33
       6       2     8.33        4    16.67
       7       7    29.17       11    45.83
       8       3    12.50       14    58.33
       9       2     8.33       16    66.67
      10       2     8.33       18    75.00
      11       1     4.17       19    79.17
      12       1     4.17       20    83.33
      13       3    12.50       23    95.83
      17       1     4.17       24   100.00
-------------------------------------------

 describe(resAI24_PSAC.L)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
 481.106   67.086    340.66  438.88  493.94  532.26  574.16

 L=[resAI24_PSC.L resAI24_PSAC.L];find(L(:,1)>L(:,2))'
ans = [2 4 5 7 9 11 13 14 16 21 22]     % Improvement for 11 of the 24 Ss.

 describe(resAI24_PSAC.opt_X)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
   0.549    0.258      0.30    0.30    0.50    0.75    1.00  % mem_k * 10
   0.492    0.091      0.40    0.40    0.47    0.60    0.60  % temper* 10
   0.396    0.246      0.25    0.25    0.25    0.47    1.00  % history* 5
   0.816    0.194      0.54    0.66    0.79    0.98    1.10  % cutoffs

 resAI24_PSAC.opt_X'
  Columns 1 through 7
    0.4936    0.3082    0.3000    0.3162    0.7743    1.0000    0.7306
    0.4000    0.5536    0.6000    0.5875    0.4064    0.4560    0.4000
    0.2500    0.2500    0.2500    0.5728    1.0000    0.2500    0.2500
    0.8153    1.1000    1.0531    0.9754    0.5920    0.5355    0.7208
  Columns 8 through 14
    0.5287    1.0000    0.3002    0.3000    0.3000    0.8951    0.5163
    0.5419    0.6000    0.4000    0.4000    0.5626    0.6000    0.5999
    0.2500    0.9297    0.2904    0.2500    0.2500    0.4863    0.3616
    0.9507    0.6644    0.8179    0.6929    1.0771    0.9875    0.9348
  Columns 15 through 21
    0.3104    0.3000    0.3000    0.7191    0.7755    0.3000    0.9996
    0.4000    0.6000    0.6000    0.4000    0.4000    0.6000    0.4242
    0.2500    0.4503    0.2500    0.2500    0.2500    0.2500    0.8239
    0.7406    1.0956    0.8972    0.6131    0.6605    1.1000    0.5444
  Columns 22 through 24
    0.4695    0.5567    0.6825
    0.4000    0.4831    0.4000
    0.8300    0.2500    0.2500
    0.5496    0.7718    0.6994

%%% Apply the same POST-OPTIMIZATION CORRECTION as in model PSACR.
%%  This correction is intended to compensate for the systematic
%%  bias in the optimization machinery (see below).
%%
%%  mem_k  = max(0.030, fmincon_mem_k - 0.011)   % bias = 0.011
%%  temper = fmincon_temper                      % no correction
%%  hist   = max(0.050, fmincon_hist - 0.005)    % bias = 0.005
%%  cutoff = max(0.500, fmincon_cutoff - 0.060)  % bias = 0.060

 x=resAI24_PSAC.opt_X; x(:,1)=max(0.03*10,x(:,1)-(0.011*10));
 x(:,3)=max(0.050*5,x(:,3)-(0.005*5));
 x(:,4)=max(0.500,x(:,4)-0.060);

 resAI24_PSAC.corr_X=x;describe(resAI24_PSAC.corr_X)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
   0.483    0.216      0.30    0.30    0.39    0.64    0.89  % mem_k * 10
   0.492    0.091      0.40    0.40    0.47    0.60    0.60  % temper* 10
   0.386    0.237      0.25    0.25    0.25    0.44    0.97  % history* 5
   0.758    0.191      0.50    0.60    0.73    0.92    1.04  % cutoff


*************************************************************
***                ******************************************
***   Model PSCR   ******************************************
***   Abs. Ident   ******************************************
***                ******************************************
*************************************************************

Active parameters:  mem_k, temper, cutoff
Shunted parameters: hist=0
Fixed parameters:   perc_k=0.04, alpha=0.3, decay=0.5,
                    L1=0.275, L9=0.675, avail=[1 ; 9] only
                    cutoff multipliers=[-3 -1 +0.9 +2.7]

 clear all ; cd 'C:\work\anchor\finalsim\' ;
 Sparams=anchor_search_params([0 1 1 0 1 0 0])
Sparams =
    model_name: 'anchor2mt'
    gener_name: 'anchor2'
        params: [1x1 struct]
     p2v_templ: {1x3 cell}
     v2p_templ: {1x3 cell}
        bounds: [2x3 double]
     clip_prob: 0.0183
         optns: [1x1 struct]

 Sparams.params.history = 0 ;      % shunting off the activation mechanism

 Sparams.params
       scale: 'LINEAR'
       N_cat: 9
     SM_conv: 0.0010
      cat_sz: 0.0500
      perc_k: 0.0400
       mem_k: 0.0700
       avail: [2x1 double]
     anchors: [9x6 double]
     cutoffs: [-2.4000 -0.8000 0.7200 2.1600]
      temper: 0.0500
     history: 0
       alpha: 0.3000
       decay: 0.5000
         ITI: 4
    M_raster: 7
    A_raster: [5 5 3 3]
    mnfieldp: 1

 Sparams.params.anchors
    0.2750    1.0000  -10.0000  -10.0000    1.0000    1.0000
    0.3250    2.0000  -10.0000  -10.0000    1.0000         0
    0.3750    3.0000  -10.0000  -10.0000    1.0000         0
    0.4250    4.0000  -10.0000  -10.0000    1.0000         0
    0.4750    5.0000  -10.0000  -10.0000    1.0000         0
    0.5250    6.0000  -10.0000  -10.0000    1.0000         0
    0.5750    7.0000  -10.0000  -10.0000    1.0000         0
    0.6250    8.0000  -10.0000  -10.0000    1.0000         0
    0.6750    9.0000  -10.0000  -10.0000    1.0000    1.0000

% Set the search bounds as for the full model later.
 Sparams.bounds(:,1)=[0.03 0.10]'*10;  % mem_k * 10
 Sparams.bounds(:,2)=[0.04 0.06]'*10;  % temper* 10
 Sparams.bounds(:,3)=[0.50 1.10]' ;    % cutoff
 Sparams.bounds
ans = 0.3000    0.4000    0.5000
      1.0000    0.6000    1.1000

 Sparams_PSCR_AI = Sparams ; clear Sparams

%%%%%%%%%%%%%%%%%%   Run the optimizer on the 24 AI data sets   %%%%%%%%%%%%%%%%%%
3 Jan 2004

% Use the PSC optimal parameters as a starting point
 x0_PSC2PSCR_AI = resAI24_PSC.opt_X ;    % the same fields, no modifications required
 describe(x0_PSC2PSCR_AI)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
   0.588    0.315      0.20    0.27    0.58    0.89    1.00  % mem_k * 10
   0.376    0.195      0.20    0.20    0.25    0.56    0.76  % temper* 10
   0.898    0.232      0.50    0.72    0.86    1.07    1.50  % cutoff

 Sparams_PSCR_AI.fname='resAI24_PSCR';
 disp(['Model PSCR, absolute identification' datestr(now)]);
 tic;resAI24_PSCR=big_psearch(AI24mdata,Sparams_PSCR_AI,[],x0_PSC2PSCR_AI);toc

% Elapsed_time = 3.0752e+004 sec (8 h 30 min) on a Dell Optiplex GX400.
% Two "Divide by zero" warnings on search 14 and a whole barrage on search 16.

 resAI24_PSCR
       model_name: 'anchor2mt'
       descriptor: '03-Jan-2004 23:44:07'
    search_params: [1x1 struct]
            opt_X: [24x3 double]
                L: [24x1 double]
         exitflag: [24x1 double]
          N_evals: [24x1 double]
          N_iters: [24x1 double]
             time: [24x1 double]
               x0: [24x3 double]

 resAI24_PSCR.search_params.v2p_templ'
ans = 'PARAMS.mem_k = VAL / 10 ;'
      'PARAMS.temper = VAL / 10 ;'
      'PARAMS.cutoffs = VAL*[-3 -1 +0.9 +2.7] ;'

 xtab1(resAI24_PSCR.N_iters)
   Value   Count  Percent  Cum_cnt  Cum_pct
-------------------------------------------
       4       2     8.33        2     8.33
       5       5    20.83        7    29.17
       6       6    25.00       13    54.17
       7       6    25.00       19    79.17
       8       2     8.33       21    87.50
      12       2     8.33       23    95.83
      15       1     4.17       24   100.00
-------------------------------------------

 describe(resAI24_PSCR.L)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
 495.028   65.298    364.48  456.88  505.24  547.81  586.96

%% Earlier search with wider search bounds produced marginally better fits:
%    Mean  Std.dev       Min     Q25  Median     Q75     Max
%------------------------------------------------------------
% 494.518   64.992    364.48  456.92  505.35  546.62  586.42

 L=[resAI24_PSC.L resAI24_PSCR.L];find(L(:,1)>L(:,2))'
ans = 21         % Improvement for only 1 of the 24 Ss.

 describe(resAI24_PSCR.opt_X)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
   0.601    0.263      0.30    0.38    0.51    0.89    1.00  % mem_k * 10
   0.533    0.080      0.40    0.47    0.58    0.60    0.60  % temper* 10
   0.855    0.204      0.50    0.66    0.86    1.07    1.10  % cutoff

 resAI24_PSCR.opt_X'
  Columns 1 through 7
    0.4790    0.3889    0.3043    0.3001    0.9062    1.0000    0.8795
    0.4000    0.5875    0.6000    0.6000    0.5386    0.5731    0.4001
    0.8667    1.1000    1.0624    1.0414    0.7027    0.5885    0.6726
  Columns 8 through 14
    0.9852    0.5726    0.6797    0.4790    0.3693    1.0000    0.5420
    0.4370    0.6000    0.4729    0.4001    0.6000    0.6000    0.5790
    0.5839    1.0868    0.6373    0.5000    1.1000    0.9307    1.0086
  Columns 15 through 21
    0.4215    0.3000    0.3000    0.5845    0.8913    0.3893    1.0000
    0.4000    0.6000    0.6000    0.5812    0.4650    0.6000    0.4759
    0.7291    1.1000    0.9350    0.7948    0.6526    1.1000    1.1000
  Columns 22 through 24
    0.3533    0.4636    0.8280
    0.4757    0.6000    0.6000
    0.7328    0.8535    0.6509

%%% Apply the same POST-OPTIMIZATION CORRECTION as in model PSACR.
%%  This correction is intended to compensate for the systematic
%%  bias in the optimization machinery (see below).
%%
%%  mem_k  = max(0.030, fmincon_mem_k - 0.011)   % bias = 0.011
%%  hist   = max(0.050, fmincon_hist - 0.005)    % bias = 0.005
%%  cutoff = max(0.500, fmincon_cutoff - 0.060)  % bias = 0.060

 x=resAI24_PSCR.opt_X; x(:,1)=max(0.03*10,x(:,1)-(0.011*10));
 x(:,3)=max(0.500,x(:,3)-0.060);

 resAI24_PSCR.corr_X=x;describe(resAI24_PSCR.corr_X)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
   0.515    0.238      0.30    0.30    0.40    0.78    0.89  % mem_k * 10
   0.533    0.080      0.40    0.47    0.58    0.60    0.60  % temper* 10
   0.798    0.200      0.50    0.60    0.80    1.01    1.04  % cutoff


*************************************************************
***                ******************************************
***   Model PSACR  = full ANCHOR model **********************
***   Abs. Ident   ******************************************
***                ******************************************
*************************************************************

Active parameters:  mem_k, temper, cutoff, hist
Fixed parameters:   perc_k=0.04, decay=0.5, alpha=0.3
                    cutoff multipliers=[-3 -1 +0.9 +2.7]
                    only two anchors initially: L1=.275, L9=.675

% This search has already been performed many times.
% The "official" results are available in \ANCHOR\SIMUL\MAIN_OPTIM.MAT.
% See also OPT_PARAMS.TXT, AIRESULTS.TXT, and DEFAULT_RESULTS.TXT.

21 Oct, 2002, see SIMUL\OPT_PARAMS.TXT

INITIAL PARAMS, SEARCH PARAMS, and FMINCON OPTIONS:
 M_raster: 7           A_raster: [5 5 3 3]     mnfield: 'posterior'
   perc_k: 0.0400         mem_k: 0.0700        optns.TofFun: 1e-1
  cutoffs: [-2.40 -0.80 0.72 2.16]             optns.TolCon: 1e-4
   temper: 0.050        history: 0.100           optns.TolX: 1e-2
    alpha: 0.300          decay: 0.500          MaxFunEvals: 400
    avail: [1 , 9]'     anchors: [0.275:0.050:0.675]'

 Sparams_PSACR_AI
    model_name: 'anchor2mt'
        params: [1x1 struct]
     p2v_templ: {1x4 cell}
     v2p_templ: {1x4 cell}
        bounds: [2x4 double]
     clip_prob: 0.0183
         optns: [1x1 struct]
         fname: 'AI24_poster'

 Sparams_PSACR_AI.v2p_templ'
ans = 'PARAMS.mem_k = VAL / 10 ;'
      'PARAMS.temper = VAL / 10 ;'
      'PARAMS.history = VAL / 5 ;'
      'PARAMS.cutoffs = VAL*[-3 -1 +0.9 +2.7] ;'

 Sparams_PSACR_AI.bounds
ans = 0.3000    0.4000    0.2500    0.5000
      1.0000    0.6000    1.0000    1.1000

 Sparams_PSACR_AI.params
       scale: 'LINEAR'
       N_cat: 9
     SM_conv: 0.0010
      cat_sz: 0.0500
      perc_k: 0.0400
       mem_k: 0.0700
       avail: [2x1 double]
     anchors: [9x6 double]
     cutoffs: [-2.4000 -0.8000 0.7200 2.1600]
      temper: 0.0500
     history: 0.1000
       alpha: 0.3000
       decay: 0.5000
         ITI: 4
    M_raster: 7
    A_raster: [5 5 3 3]
    mnfieldp: 1

%%%%%%%%%%%%%%%%%%   Run the optimizer on the 24 AI data sets
21 Oct, 2002

 tic;resAI24_PSACR=big_psearch(AI24mdata,Sparams_PSACR);toc

 resAI24_PSACR =
       model_name: 'anchor2mt'
       descriptor: '21-Oct-2002 10:07:30'
    search_params: [1x1 struct]
            opt_X: [24x4 double]
                L: [24x1 double]
         exitflag: [24x1 double]
          N_evals: [24x1 double]
          N_iters: [24x1 double]
             time: [24x1 double]

 xtab1(resAI24_PSACR.N_iters)
   Value   Count  Percent  Cum_cnt  Cum_pct
-------------------------------------------
       4       2     8.33        2     8.33
       5       4    16.67        6    25.00
       6       6    25.00       12    50.00
       7       6    25.00       18    75.00
       8       1     4.17       19    79.17
       9       2     8.33       21    87.50
      10       3    12.50       24   100.00
-------------------------------------------

 describe(resAI24_PSACR.L)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
 491.977   65.742    355.49  453.90  499.49  545.52  580.62

 L=[resAI24_PSCR.L resAI24_PSACR.L];find(L(:,1)<=L(:,2))'
ans = [3 6 12 15 17 19 20]         % Improvement for only 7 of the 24 Ss.

 describe(resAI24_PSACR.opt_X)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
   0.670    0.272      0.30    0.40    0.71    0.93    1.00  % mem_k * 10
   0.495    0.086      0.40    0.40    0.46    0.59    0.60  % temper* 10
   0.366    0.189      0.25    0.25    0.25    0.40    0.86  % history* 5
   0.779    0.190      0.54    0.65    0.71    0.93    1.10  % cutoff

%%%%%%   POST-OPTIMIZATION CORRECTION    %%%%%%%%%%%%%%%%%%%%%%%%%
%%
%%  Monte Carlo experiments with the optimizer revealed that there
%%  is a systematic bias in the parameter search process. It may be
%%  due in part to discretizations and approximations in the model
%%  tracer ANCHOR2MT, in part to the model-tracing nature of the
%%  log-likelihood calculation scheme, and in part to inherent
%%  statistical instability in the gradient search method under
%%  rather volatile tolerances and termination criteria.
%%  This corrections below are intended to compensate for
%%  the systematic biases in the optimization machinery.
%%
%%  mem_k  = max(0.030, fmincon_mem_k - 0.011)   % bias = 0.011
%%  temper = fmincon_temper                      % no correction
%%  hist   = max(0.050, fmincon_hist - 0.005)    % bias = 0.005
%%  cutoff = max(0.500, fmincon_cutoff - 0.060)  % bias = 0.060

 x=resAI24_PSACR.opt_X; x(:,1)=max(0.03*10,x(:,1)-(0.011*10));
 x(:,3)=max(0.050*5,x(:,3)-(0.005*5));
 x(:,4)=max(0.500,x(:,4)-0.060);

 resAI24_PSACR.corr_X=x;describe(resAI24_PSACR.corr_X)
    Mean  Std.dev       Min     Q25  Median     Q75     Max
------------------------------------------------------------
   0.582    0.244      0.30    0.30    0.60    0.82    0.89  % mem_k * 10
   0.495    0.086      0.40    0.40    0.46    0.59    0.60  % temper* 10
   0.356    0.179      0.25    0.25    0.25    0.37    0.83  % history* 5
   0.720    0.188      0.50    0.59    0.65    0.87    1.04  % cutoff


===========  Final optimal parameters, full model  ======================

ANCHOR2MT, ABSOLUTE-IDENTIFICATION DATA, 24 Ss   21 Oct, 2002
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
POSTERIOR-BASED model tracing.
Fixed parameters: PERC_K=0.040, ALPHA=0.30, DECAY=0.50.
Initial anchors:  L1=0.275, L9=0.675.

          logL     mem_k    temper      hist    cutoff
------------------------------------------------------
fmincon lo_bound  0.0300    0.0400    0.0500    0.5000
fmincon init_app  0.0700    0.0500    0.1000    0.8000
fmincon hi_bound  0.1000    0.0600    0.2000    1.1000
------------------------------------------------------
mean   491.976    0.0582    0.0495    0.0711    0.7200
std     65.742    0.0244    0.0086    0.0359    0.1882

min    355.491    0.0300    0.0400    0.0500    0.5000
P10    385.734    0.0300    0.0400    0.0500    0.5000
P25    453.900    0.0303    0.0403    0.0500    0.5941
med    499.486    0.0596    0.0460    0.0500    0.6517
P75    545.515    0.0818    0.0591    0.0745    0.8702
P90    565.438    0.0890    0.0600    0.1336    1.0400
max    580.618    0.0890    0.0600    0.1668    1.0400
------------------------------------------------------
 1     433.496    0.0407    0.0400    0.0500    0.7570
 2     530.396    0.0312    0.0577    0.0500    1.0400
 3     530.167    0.0300    0.0600    0.0500    0.9843
 4     504.475    0.0673    0.0497    0.1668    0.6973
 5     473.448    0.0801    0.0449    0.0939    0.5666
 6     532.644    0.0890    0.0566    0.0500    0.5000
 7     455.813    0.0723    0.0400    0.0500    0.6227
 8     514.244    0.0834    0.0400    0.0500    0.5000

 9     565.205    0.0890    0.0600    0.1560    0.6217
10     399.251    0.0499    0.0405    0.0623    0.6448
11     355.491    0.0300    0.0400    0.0500    0.6855
12     531.534    0.0300    0.0580    0.0500    1.0400
13     580.618    0.0889    0.0600    0.0844    0.8740
14     560.668    0.0874    0.0581    0.0629    0.6254
15     371.190    0.0307    0.0400    0.0500    0.6587
16     558.386    0.0300    0.0600    0.0642    1.0400

17     494.497    0.0300    0.0600    0.0500    0.8664
18     490.305    0.0519    0.0425    0.0646    0.6664
19     480.035    0.0767    0.0400    0.0500    0.5510
20     567.538    0.0300    0.0600    0.0500    1.0400
21     560.943    0.0890    0.0459    0.1209    0.6356
22     387.350    0.0430    0.0427    0.1311    0.5000
23     451.987    0.0717    0.0462    0.0500    0.6244
24     477.754    0.0734    0.0447    0.0500    0.5376
------------------------------------------------------
sbj       logL     mem_k    temper      hist    cutoff


INITIAL PARAMS, SEARCH PARAMS, and FMINCON OPTIONS:
 M_raster: 7           A_raster: [5 5 3 3]     mnfield: 'posterior'
   perc_k: 0.0400         mem_k: 0.0700        optns.TofFun: 1e-1
  cutoffs: [-2.40 -0.80 0.72 2.16]             optns.TolCon: 1e-4
   temper: 0.050        history: 0.100           optns.TolX: 1e-2
    alpha: 0.300          decay: 0.500          MaxFunEvals: 400
    avail: [1 , 9]'     anchors: [0.275:0.050:0.675]'

POST-OPTIMIZATION CORRECTION (based on Monte-Carlo estimation
                  of the biases of the optimization machinery):
  mem_k  = max(0.030, fmincon_mem_k - 0.011)   % bias = 0.011
  temper = fmincon_temper                      % no correction
  hist   = max(0.050, fmincon_hist - 0.005)    % bias = 0.005
  cutoff = max(0.500, fmincon_cutoff - 0.060)  % bias = 0.060

%%%
%%%%%%%%%%  End of file HRAI_PSEARCH.TXT
