Recent changes to wiki

InteroperabilityWithOtherSystems modified by KKozlov

KKozlov — Wed, 03 Feb 2021 06:38:44 -0000

--- v1
+++ v2
@@ -23,7 +23,7 @@

 Call DEEP: 

-    dpdeepctl --default-name=model.ini
+    deepmethod --default-name=model.ini


 For each parameter set the following formatted file "/tmp/XXXXXX" will be generated:

ConvertDoubleToInt modified by KKozlov

KKozlov — Wed, 03 Feb 2021 06:35:52 -0000

--- v1
+++ v2
@@ -4,7 +4,7 @@

 DEEP method operates on floating point parameters. On the other hand many real problems are formulated so that the parameters are integer numbers, e.g. indices of some sort. 

-Particularly, this is the case for the problem of constructing the molecule with desired properties. 
+### Conversion by sorting

 The following procedure is used to produce the integer vector of parameters from the floating point vector constructed by DEEP method. 

@@ -34,3 +34,9 @@
     i[2] = 1
     i[3] = 3

+### Conversion by special rounding
+
+Taken from Uher et al., 2016
+```
+x_int = (1+999*x_double)/500
+```

ExternalModelParameters modified by KKozlov

KKozlov — Wed, 03 Feb 2021 06:28:02 -0000

--- v3
+++ v4
@@ -55,7 +55,7 @@

 ###  parmtypes=1; 0; 0;

- * type of parameters: 1 - integer values to be produced by rounding, 2 - integer values to be produced by sorting, 0 - real values; 4 - integer values to be produced by special rounding. Methods a re described in [ConvertDoubleToInt]
+ * type of parameters: 1 - integer values to be produced by rounding, 2 - integer values to be produced by sorting, 0 - real values; 4 - integer values to be produced by special rounding. Methods are described in [ConvertDoubleToInt]

 ###  maskvals=0; 0; 0;

ExternalModelParameters modified by KKozlov

KKozlov — Wed, 03 Feb 2021 06:27:16 -0000

--- v2
+++ v3
@@ -55,7 +55,7 @@

 ###  parmtypes=1; 0; 0;

- * type of parameters: 1 - integer as rounding, 2 - integer as sorting, 0 - real
+ * type of parameters: 1 - integer values to be produced by rounding, 2 - integer values to be produced by sorting, 0 - real values; 4 - integer values to be produced by special rounding. Methods a re described in [ConvertDoubleToInt]

 ###  maskvals=0; 0; 0;

ExternalModelParameters modified by KKozlov

KKozlov — Wed, 21 Oct 2020 10:27:27 -0000

--- v1
+++ v2
@@ -37,7 +37,53 @@
 ### hbound | valhbound

   * **hbound=0;0;...;0;** is the upper bound for each parameter individually. 
-  * **valhbound=0** is the upper bound for all parameters. 
+  * **valhbound=0** is the upper bound for all parameters.
+
+Another form of definition:
+
+###  partnames=x;beta;MB;
+
+ * names of groups of parameters
+
+###  partsizes=60; 55; 1;
+
+ * number of parameters in each group
+
+###  parmvals=200; 5; 5;
+
+ * initial values for each group
+
+###  parmtypes=1; 0; 0;
+
+ * type of parameters: 1 - integer as rounding, 2 - integer as sorting, 0 - real
+
+###  maskvals=0; 0; 0; 
+
+  * values for masks
+
+###  tweakvals=1; 1; 1; 
+
+ * define if the value can be changed. 
+
+###  dparmvals=200; 5.5; 5;
+
+ * initial values for each group
+
+###  lboundvals=0; -10; 0.01;
+
+ * lower bound for each group
+
+###  hboundvals=1924; 10; 9;
+
+ * upper bound for each group
+
+###  limitvals=1; 1; 1;
+
+ * defines if parameters are limited
+
+###  scalevals=1; 1; 1;
+
+ * scale factor for aparameters

 ### command

@@ -48,10 +94,14 @@
 Defines the parameter's file format. The following formats are supported: 

   * **convert=sdf** for molecules. 
-  * **convert=gemstat** for gemstat program. 
-  * **convert=subset**
-  * **convert=subsubset**
-  * **convert=octave**
+  * **convert=gemstat** for gemstat program: one line in a file, starting with parameter name, parameters are delimited by space
+  * **convert=dgemstat** two lines in a file, starting with parameter name, parameters are delimited by space
+  * **convert=subset** for integer indices
+  * **convert=subsubset** for subset of integer indices
+  * **convert=octave** to use in Octave with source command
+  * **convert=r** to use in R with source command
+  * **convert=ini** ini format to use with Qt framework
+  * if this option is not given parameters are printed to the command line 

 ### delimiters

@@ -65,17 +115,23 @@

   * defines the reordering of keys 

+Example:
+    
     [default_model]
-    parts=hb;2;Kr;2;gt;2;kni;2;bcd;2;cad;2;tll;2;basal_transcription =;1;hb hb;1;Kr Kr;1;gt gt;1;kni kni;1;bcd bcd;1;cad cad;1;tll tll;1;
-    mask=1;0;0;0;0;0;0;0;0;0;0;0;0;0;1;1;0;0;0;0;0;0;
-    parms=0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;
-    dparms=0.964910;1.011700;1.056578;1.099454;0.154759;1.699184;1.699184;1.699184;1.699184;1.699184;1.699184;1.699184;1.699184;1.699184;1.257495;0.705276;1.858218;0.566396;1.420308;0.660813;0.356575;0.014055;
-    tweak=1;1;1;1;1;1;1;1;1;1;1;1;1;1;1;1;1;1;1;1;1;1;
-    lbound=0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.01;0.0001;
-    hbound=100;100;100;100;100;100;100;100;100;100;100;100;100;100;500;500;500;500;500;500;500;0.1;
-    command=./script
-    convert=gemstat
-    delimiters= ,\n\t\s
-    keys=23;
-    mapping=0;
-    
+    partnames=x;beta;MB;
+    partsizes=60; 55; 1;
+    parmvals=200; 5; 5; 
+    parmtypes=1; 0; 0;
+    maskvals=0; 0; 0;
+    tweakvals=1; 1; 1;
+    dparmvals=200; 5.5; 5; 
+    lboundvals=0; -10; 0.01;
+    hboundvals=1924; 10; 9; 
+    limitvals=1; 1; 1;
+    scalevals=1; 1; 1;
+    command=script 
+    delimiters=\n=
+    keys=0;1;2;
+    convert=ini
+    mapping=0;0;1;2;
+    type=command

ControlParameters modified by KKozlov

KKozlov — Wed, 21 Oct 2020 09:59:17 -0000

--- v4
+++ v5
@@ -148,17 +148,25 @@

 ## Control Parameters for the Quality Functional

+### debug=0 | 1
+
+### ignore_cost=0 | 1
+
+### constrain_aggr=sum | max
+
+### penalty_aggr=sum | max
+
 ### <name>=<keyword>;<keyword>;<index>;<rank>;<weight>

 <weight> is the multiplier for the function value in the final value 

 <rank> is the value with wich the random number is compared to decide if the decrease in this function is worth 

-mainfunc=scoreint;0;1;1; - scoring function. 0 means it is the main scoring function 
+mainfunc=target;objfunc;0;1;1; - scoring function. 0 means it is the main scoring function 

-func01=readpenalty;1;0.45;1; - read the value of the penalty from scoring program output 
+func01=penalty;readpenalty;1;0.45;1; - read the value of the penalty from scoring program output 

-func02=readpenalty;2;0.45;1; 
+func02=penalty;readpenalty;2;0.45;1; 

 Example:

ControlParameters modified by KKozlov

KKozlov — Wed, 21 Oct 2020 09:34:14 -0000

--- v3
+++ v4
@@ -116,7 +116,7 @@

 Example: 

-    [default_settings]    
+    [default_settings]
     max_threads=2
     proportional_stop=1e-04
     stop_count=15
@@ -129,32 +129,30 @@
     es_lambda=25
     es_cutoff=3
     es_kind=0
-   mean_cost=0
-   noglobal_eps=0
-   substeps=0.005
-   substieps=0
-   ca_flag=1
-   logdepth=2
-   transform=tanh
-   gamma_init=1.0
-   precision=16
-   keep_order=1
-   roundoff_error=1e-06
-   seed=62534198
-   run_before=initcancel;optpost;optposteval;
-   run=gcadeep;5;gacdeep;1;gdeep;1;edeep;1;sdeep;24;selde;1;dpupdate;1;checkcancel;1;substitute;2;optpost;1;optposteval;1;
-   run_after=optpost;optposteval;printlog;
+    mean_cost=0
+    noglobal_eps=0
+    substeps=0.005
+    substieps=0
+    ca_flag=1
+    logdepth=2
+    transform=tanh
+    gamma_init=1.0
+    precision=16
+    keep_order=1
+    roundoff_error=1e-06
+    seed=62534198
+    run_before=initcancel;optpost;optposteval;
+    run=gcadeep;5;gacdeep;1;gdeep;1;edeep;1;sdeep;24;selde;1;dpupdate;1;checkcancel;1;substitute;2;optpost;1;optposteval;1;
+    run_after=optpost;optposteval;printlog;


 ## Control Parameters for the Quality Functional

-### <name>=<keyword>;<index>;<rank>;<weight>
+### <name>=<keyword>;<keyword>;<index>;<rank>;<weight>

 <weight> is the multiplier for the function value in the final value 

 <rank> is the value with wich the random number is compared to decide if the decrease in this function is worth 
-
-precond=doubletoint;-1;1;1; - first convert parameters to integers. -1 means "before anything else" 

 mainfunc=scoreint;0;1;1; - scoring function. 0 means it is the main scoring function 

@@ -163,8 +161,17 @@
 func02=readpenalty;2;0.45;1; 

 Example: 
+
+    [default_target]
+    debug=0
+    ignore_cost=0
+    constrain_aggr=sum
+    penalty_aggr=sum
+    mainfunc=target;objfunc;0;1;0;
+    cbd_pen_em=penalty;readpenalty;1;2;1;
+    cbd_pen_r1=penalty;readpenalty;2;2;1;
+    cbd_pen_r3=penalty;readpenalty;3;2;0;
+    cbd_pen_r5=penalty;readpenalty;4;2;0;
+    cbd_pen_r7=penalty;readpenalty;5;2;0;
+    cbd_pen_r8=penalty;readpenalty;6;2;0;

-    [target]
-    mainfunc=scoredouble;0;1;1;
-    func01=readpenalty;1;0.45;1;
-

ControlParameters modified by KKozlov

KKozlov — Wed, 21 Oct 2020 09:21:18 -0000

--- v2
+++ v3
@@ -115,6 +115,7 @@
   * functions to run after main loop

 Example: 
+
     [default_settings]    
     max_threads=2
     proportional_stop=1e-04

ControlParameters modified by KKozlov

KKozlov — Wed, 21 Oct 2020 09:19:58 -0000

--- v1
+++ v2
@@ -18,7 +18,7 @@

   * number of members of the population 

-### recombination_strategy = simple | de_3_bin | de_3_exp | de_3_exp_T | de_3_bin_T
+### recombination_strategy = de_3_exp_rand |  de_3_bin_rand | de_3_exp_rand_T | de_3_bin_rand_T | simple | de_3_bin | de_3_exp | de_3_exp_T | de_3_bin_T

   * type of evolution 

@@ -37,10 +37,38 @@
 ### es_lambda

   * number of individuals to exchange between branches. 
+ 
+### es_cutoff
+   
+   * =3 number of individuals
+
+### es_kind
+  
+* with what to substitute: 0 - best, 1 - less better
+
+### mean_cost
+
+* =0 

 ### noglobal_eps

-  * small or big cube in parameter space, 0 - big 
+  * small or big cube in parameter space, 0, -1 - big and initial guess is ignored.
+
+### substeps
+
+    * 0 - use limits, -1 - do not perturb, 0.1 - use vecinity
+
+### substieps
+
+    * 0 - use limits, -1 - do not perturb, 0.1 - use vecinity
+
+### ca_flag
+
+    * 0 - use space for the best, 1 - distribute equaly
+
+### logdepth
+
+    * log more info

 ### transform = tanh | sin | alg | rand

@@ -87,25 +115,34 @@
   * functions to run after main loop 

 Example: 
-    
-    [default_settings]
-    max_threads=5
-    proportional_stop=1e-8
-    stop_count=2
-    population_size=40
-    recombination_strategy=de_3_bin_T
+    [default_settings]    
+    max_threads=2
+    proportional_stop=1e-04
+    stop_count=15
+    absolute_iter=10
+    population_size=150
+    recombination_strategy=de_3_exp_rand
     recombination_weight=0
     recombination_prob=0
     recombination_gamma=0.9
-    es_lambda=2
-    noglobal_eps=0
-    transform=tanh
-    gamma_init=50.0
-    roundoff_error=1e-12
-    seed=206428
-    run_before=initstop;optpost;optposteval;writelog;
-    run=gdeep;1;edeep;1;sdeep;1;dpupdate;1;substitute;5;checkstop;1;optpost;1;optposteval;1;writelog;1;
-    run_after=optpost;optposteval;writelog;
+    es_lambda=25
+    es_cutoff=3
+    es_kind=0
+   mean_cost=0
+   noglobal_eps=0
+   substeps=0.005
+   substieps=0
+   ca_flag=1
+   logdepth=2
+   transform=tanh
+   gamma_init=1.0
+   precision=16
+   keep_order=1
+   roundoff_error=1e-06
+   seed=62534198
+   run_before=initcancel;optpost;optposteval;
+   run=gcadeep;5;gacdeep;1;gdeep;1;edeep;1;sdeep;24;selde;1;dpupdate;1;checkcancel;1;substitute;2;optpost;1;optposteval;1;
+   run_after=optpost;optposteval;printlog;


 ## Control Parameters for the Quality Functional

Home modified by KKozlov

KKozlov — Wed, 21 Oct 2020 08:46:57 -0000

--- v17
+++ v18
@@ -23,18 +23,26 @@
 ## How to cite in publications

        @article{Kozlov2016,
-          title = {A Software for Parameter Optimization with {{Differential Evolution Entirely Parallel}} Method},
-          author = {Kozlov, Konstantin and Samsonov, Alexander M. and Samsonova, Maria},
-          year = {2016},
-          month = aug,
-          volume = {2},
-          pages = {e74},
-          issn = {2376-5992},
-          doi = {10.7717/peerj-cs.74},
-          abstract = {Summary. Differential Evolution Entirely Parallel (DEEP) package is a software for finding unknown real and integer parameters in dynamical models of biological processes by minimizing one or even several objective functions that measure the deviation of model solution from data. Numerical solutions provided by the most efficient global optimization methods are often problem-specific and cannot be easily adapted to other tasks. In contrast, DEEP allows a user to describe both mathematical model and objective function in any programming language, such as R, Octave or Python and others. Being implemented in C, DEEP demonstrates as good performance as the top three methods from CEC-2014 (Competition on evolutionary computation) benchmark and was successfully applied to several biological problems.},
-          journal = {PeerJ Computer Science},
-          keywords = {Bioinformatics,Differential Evolution,Mathematical modeling,Open source software,Parallelization,Parameter optimization},
-          language = {en}
+        title = {A Software for Parameter Optimization with {{Differential Evolution Entirely Parallel}} Method},
+        author = {Kozlov, Konstantin and Samsonov, Alexander M. and Samsonova, Maria},
+        year = {2016},
+        month = aug,
+        volume = {2},
+        pages = {e74},
+        issn = {2376-5992},
+        doi = {10.7717/peerj-cs.74},
+        abstract = {Summary. Differential Evolution Entirely Parallel (DEEP) package is a software for finding unknown real
+        and integer parameters in dynamical models of biological processes by minimizing one or even several objective
+        functions that measure the deviation of model solution from data. Numerical solutions provided by the most
+        efficient global optimization methods are often problem-specific and cannot be easily adapted to other tasks. In
+        contrast, DEEP allows a user to describe both mathematical model and objective function in any programming
+        language, such as R, Octave or Python and others. Being implemented in C, DEEP demonstrates as good performance as
+        the top three methods from CEC-2014 (Competition on evolutionary computation) benchmark and was successfully
+        applied to several biological problems.},
+        journal = {PeerJ Computer Science},
+        keywords = {Bioinformatics, Differential Evolution, Mathematical modeling, 
+        Open source software,Parallelization,Parameter optimization},
+        language = {en}
     }

     @article{Kozlov13,