ModelsEnvironments
Explore with PSE

PSE (Pattern Space Exploration) method is used to explore the output's diversity of a model. Input parameter values are selected to produce new output values, such that as the exploration progresses, the region of the output space that is covered gets bigger. PSE reveals the potential of your model: the variety of the dynamics it is able to produce, even those about which you were not investigating in the first place !
image/svg+xml Output Exploration Input Exploration Sensitivity Optimisation
Method scores:
The PSE method is designed to cover the output space, hence the highest score possible in output exploration. PSe is all about covering output space, hence the low scores in optimization and Input Space exploration. As the methods discovers patterns in the output space, inputs values that lead to these patterns are available, that give litlle insights about the model sensitivity. Contrarily to calibration-based methods, PSE is sensitive to the dimensionality of the output space, as it maintains an archive of the output space locations covered ever since, which is rapidly costly for more than 3-4 dimensions.
PSE handles stochasticity in the sense that the selected pattern are estimated by the median of several model execution output values.


image/svg+xml 62 -6 d=7:1" 2090.56.7 :4;94033,6.496394" y="155705le:evenodd;stecap:butt;stroke-linejoin:mi er;stroke-mi e /StillImage6stro-y8181e6stro-y r22 style="overfl22,0,0tro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81-177.89845)">62 -6 e-op14 20e;s36.881,0,2 26.190 26.190 .496394" y="156732 style="overflow:visible" id="marker4533-0" i e /StillImage6stro-y8181e6stro-y r22 style="overfl22,0,0tro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81-177.89845)">62 -310.p143 20e;s36.881,0,2 26.190 26.190 .496394" y="156732 6le:evenodd;stecap:butt;stroke-linejoin:mi er;stroke-merlimi :4;et:0;stroke- 26;fill-07795 ink;44558 h 25" style="0e="overflrlim,0tro-y81e6stro-e;19109428 0,0 5,-5 -12.5,0 5,5 Z" style="fill:#333333;fill-opacity:1;fill-rule transform="Ailosange 62 -13.58403, .78-13 :4:0px2"m 64:0px2"m 64l 5.145 ari8="m257 5.145 ari8="m258 5.145 ar8="m258 ztroke-wid x="104.91343" height="159.18083" g transform="3156749-6 style="oi erlimi :4;stroke-d160148y:none;160148y142.6460.758772.678269 c -0.056,-0.0347 -0.16stro-y8181e6stro-y r22 style="overfl22,0,0tro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81-177.89845)">62 -6 d=7:1" 2090.56.7 :4;94033,6.496394" y="155705l style="overflow:visible" id="marker4533-0" .056,-0.0347 -0.16stro-y8181e6stro-y r22 style="overfl22,0,0tro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81-177.89845)">62 -6 e-op14 20e;s36.881,0,2 26.190 26.190 .496394" y="156732 9- style="overflow:visible" id="Arrow1Lend-6" .056,-0.0347 -0.16stro-y8181e6stro-y r22 style="overfl22,0,0tro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81-177.89845)">62 -310.p143 20e;s36.881,0,2 26.190 26.190 .496394" y="156732 6l5le:evenodd;stecap:butt;stroke-linejoin:mi er;stroke-merlimi :4;et:0;stroke- 26;fill-07795 ink;44558 h 25" style="0e="overflrlim,0tro-y81e6stro-e;19109428 0,0 5,-5 -12.5,0 5,5 Z" style="fill:#333333;fill-opacity:1;fill-rule transform="Ajlosange 62 359="m274,659=0669 :4:0px2"m 64:0px2"m 64l 5.145 ari8="m258 5.145 ari8="m257 5.145 ar8="m257 ztroke-wid x="104.91343" height="159.18083" g transform="3156749-0-source="hti erlimi :4;stroke-d160148y:none;160148y1426e:no478,68423 88269 c -0.056,-0.0347 -0.16stro-y8181e6stro-y r22 style="overfl22,0,0tro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81-177.89845)">62 -6 d=7:1" 2090.56.7 :4;94033,6.496394" y="155705l 8- style="overflow:visible" id="Arrow1Lend-6" .056,-0.0347 -0.16stro-y8181e6stro-y r22 style="overfl22,0,0tro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81-177.89845)">62 -6 e-op14 20e;s36.881,0,2 26.190 26.190 .496394" y="156732 7le:evenodd;stecap:butt;stroke-linejoin:mi er;stroke-m.056,-0.0347 -0.16stro-y8181e6stro-y r22 style="overfl22,0,0tro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81e6stro-y81-177.89845)">62 -310.p143 20e;s36.881,0,2 26.190 26.190 .496394" y="156732 6l1-5:evenodd;stecap:butt;stroke-linejoin:mi er;stroke-merlimi :4;et:0;stroke- 26;fill-07795 ink;44558 h 25" style="0e="overflrlim,0tro-y81e6stro-e;19109428 0,0 5,-5 -12.5,0 5,5 Z" style="fill:#333333;fill-opacity:1;fill-rule transform="Aklosange 62 -05.15116,750773478,:4:0px2"m 64:0px2"m 64l 5.145 ari8="m258 5.145 ari8="m258 5.145 ar8="m258 ztroke-wid x="104.91343" height="159.18083" pacity:1;stroke:#2f2404079443" )"> -38 ,565.6678,79.4.019610.p6 transform6stro-y8181e6stro-y r22 style="overflow:visible"> et:0;stroke-----ctor2ity:695:evenodd;st232f2f98y:6503:evenodd;st23.5768 ;ss93.8:vis:evenodd;st23 ink78348235:evenodd;st23-y81e6str108- vie:norb"httppa:norPara.0347 -0.16stro-y8style:norbll;font-variant:nor.2 .496394" :norPara4110-so> ie:norParavie:norbll;>083" pacity:1;stroke:#2f2161389-6 53 7l6width:1.000000"96-271.26424nt-w58 8="6666,9.2pe:c i8=653t-sp3e;st70.285.539 z"319.7517f3b;fill-opacity:1;/rdf:RDF> Input Exploration 1024 style="ove z m 099tyle:norblbll;f6721.943yl>M="29path896-3" asht-family:gch:norbll;font-size:4.130082t="8.4875278" x="133.54297" y="37.516411" ry="0.94631761" /> Input Exploration 10448 style="ove 6 -439258yle:norblbll;f6 9349ty>o1path896-3" asht-family:gch:norbll;font-size:4.130082t="8.4875278" x="133.54297" y="37.516411" ry="0.94631761" /> Input Exploration 10448-8 style="ove 1167.2p38yle:norblbll;f94city04ty>o1path896-3" asht-family:gch:norbll;font-size:4.130082t="8.4875278" x="133.54297" y="37.516411" ry="0.94631761" /> 10488 style="ove 6 -4413arker-end:u:url(704071975y>o2path896-3" asht-family:gch:norbll;font-size:4.130082t="8.4875278" x="133.54297" y="37.516411" ry="0.94631761" /> 10488-8 style="ove 645.28699ker-end:u:url(399.1937ty>o2path896-3" asht-family:gch:norbll;font-size:4.130082t="8.4875278" x="133.54297" y="37.516411" ry="0.94631761" /> 10699ker-end:u:u -441.00019ker-end:u:url(#73.32532y>ipath896-3" asht-family:gch:norbll;font-size:4.130082t="8.4875278" x="133.54297" y="37.516411" ry="0.94631761" /> :none;stroke-oity:1"> 10699-9ker-end:u:u -4widt0162yle:norblbll;f660.5132yl>jpath896-3" asht-family:gch:norbll;font-size:4.130082t="8.4875278" x="133.54297" y="37.516411" ry="0.94631761" /> 10699-9-9ker-end:u:u -445.97012yle:norblbll;f759.62256l>kpath896-3" ddasharr!-- c="2 dummy anorg/mm" xmstart343 from ="47m" --sharranorg/stroke-attri="4eTy guCSS"troke-attri="4eN.6068 id="path46"troke-to co!-- P x=ts en bleu ################################################################################################################## apparaissent en 6 secondeve opuis le iRun novelty search) selects in:tro-ythe parentsd hose output values are rare compared2to the -si< oftro-ythe population and to the provious generg/ions. In order to evalug/strokethe -arity of aethe output values, PSE discretisesethe output spl-o,trokedividrng2it8into cells. Each time aesimulation is run and itsdoutputtrokeis known,ya coun-12eis incremented2inethe cor-sipondrng2cell. PSEtrokeproferentially selects the parentsd hose associated2cells havstrokelow coun-12s. By selectrng2parentsd th rare output values, we2tny and increase thetrokechances2to produce new8individuals th proviously unobserved 8282behaviou2s. genome:ethe model2par.60-12s, varyrng2 thinetheir minimumyand maximumybounds,oke-m
  • objectrves:ethe observables3measured2in: each simulation and thinewhich we2search in: diversity,d th a discretization step ,
  • oke-m
  • stochastic:ethe seed2generg/or,ewhich generg/ss su table seeds in: the method. Mandg/ory if you2 model2contains randomness. The generg/sd seed2in: the model2taskeis tnejo,0 /sd through the variable give as an argument of Replication (here myseed).
  • oCalibrg/ion) //seed2declarg/ion in: random numb12egenerg/ion val myseed =Val[Int] val explorg/ion =oke-mPSE (130082rgenome =oke-mmmmmSeq(oke-mmmmm 2par.61y8in (0.0, 1.0),oke-mmmmm 2par.628in (-10.0, 10.0)),oke-mmmobjectrves =oke-mmmmmSeq(oke-mmmmm 2output18in (0.02to 40.02bym5.0),oke-mmmmm 2output28in (0.02to 4000.02bym50.0)),oke-mmmstochastic = Stochastic(seed = myseed)oke-m) val evol"4ion =oke-mSteadyStg/sEvol"4ion(oke-mmmalgor thm = explorg/ion,oke-mmmevalug/ion = modelTask,oke-mmmpar.llelism = 10,oke-mmm-12mination = 100 oke-m) where par.61 and par.62 are inputo of the taskethat runs the model, and output1 and output2 are outputs of that s.60 task. The numb12 of inputo and outputs are il.5,0 ed.oPLOS ONE 10(9), 2015.oke-mmmmm[online version]mm[bibteX] Available methodsCalibrg/ion5/a>Profiles5/a>PSE5/a>Data Processrng5/a>Other DoEs5/a>See alsoLanguage5/a>GUI guide5/a>Advanced Concepts5/a>