Home

UQ-PyL User Manual version 1.1

1. bk q e gt UQ PyL User Manual Version 1 1 Chen Wang wangchen O mail bnu edu cn Qingyun Duan qyduan O bnu edu cn Beijing Normal University Beijing China Table of Contents Ni aap ORNE RI RSA TORRE RSA 3 IE E E EE INANE A E PC CO PRO E COR CCP EE TT o 3 1 2 Available UQ PYL Weu E 3 121 Desen Ol s doc EE 3 1 252 Statistical ANa NS ai eoa ls 3 2 5 SOSTUVO SIS ee 3 M24 Surrogate heel ut 3 1 2 AENA At EOD da 4 1 3 Overview about functionality of the UQ PyL package eeeeeeeeeeeeeeeeees 4 GN EE ge E 6 P MB uU EE 6 22 Detailed Installation iba oe bete ince bete iacob eene 6 2 2 WiBdows PLOT ini 6 PANDA AMS A D DLL 14 2 2 5 MacOS Plat Or M EE 20 osi WT EE 25 o ALE TION ee E e AI E RI E RR ERR ERR RR ER 25 S XD OPPIDIS INVITI s ce AI a E IR UR RIR UR RE CR ene ern aene sere 26 Examples M RR O A eee eee 28 SO DOI a NUNC Os MR c atte RD a A a a N 28 TT T Problem Ree EE 28 dE Desremor Experiment nea ee ee is 29 4 LS sta istical ANa YSIS nds 33 SC EVI CO a a OD DOO 38 A LS SUPT OR Model ME aiia liciid 43 4 Eo Parameter Cp EE EE 48 A SAC SMA Modeline 52 2 2 RI Problen Dern ee 52 AD Desen OF EX POr oue eni eite etie a a eh nea hie 58 2 2 S6 DSItEVIES AIDS TS ia ii o Nets irae 61 2 24 Surocate Mod lN E oeronnia c 64 2 2 5 Paramete
2. Choose Model File D UQ PyL UQ test_functions SAC py Choose Model File Stepi Load parameter file and driver file Desizn of Experiment Method Choose DoE method Morris One at Time Morries One At A Time MOAT Configuration Number of total sample points dimension 1 Number of Trajectories Number of Trajectories 20 Generate DoE Script Execute DoE Script Step2 Choose Design of Experiment method and generate results Show Design of Experiment Result Choose Result File Choose Result File Display Result Step 1 Define parameter and model information lt gt Choose Design of Experiment tab lt gt Load parameter file UQ PyL UQ test_functions params SAC txt and model file UQ PyL UQ test_functions SAC py for SAC model it s the model driver file generated before Step 2 Choose DoE method and run the results lt gt Choose DoE method Morris One at A Time and set Number of Trajectories 20 ze Click Generate DoE Script button and Execute DoE Script button to acquire DOE results UQ PyL gives the tabular and graphic results 58 i Figure 1 OO Bv Morris One at Time Sampling LO a C Python27 python exe 59 M Figure 1 Mb T O O gt omm Model Evaluation Results Model Evaluation Results di 50 100 150 200 250 Sample Number This step can also 1mplemented using python script Python script file SAC DoE py
3. lt gt Click Show Results button to show statistical analysis results UQ PyL gives the tabular and graphic results E C Windows system32 cmd exe D NUQ PyL gt python B m UQ analyze m moments p D UQ PyL UQ test_functions para s Sobol G txt I D UQ PuL sample output latin2 2015 85 18 22 12 46 txt Y D U PuL model output latin2 2015 85 18 22 12 46 txt he minimum value of output is 6 60529446436 he maximum value of output is 2 599489 he mean value first moment gt is 8 9562312589 he variance value second moment is 8 350640944697 he standard deviation is 6 592149427676 he skewness value third moment is MB 683569348M56 he kurtosis value fourth moment is 6 169158573535 36 Ww Figure 1 ez 00 GE 7 S Hist Figure of Model Evaluation Results 3 5 3 0 23 2 0 ES Number of Evaluations 1 0 0 5 0 0 0 5 10 15 2 0 2 5 3 0 Model Evaluation Value This step can also implemented using python script Python script file Sobol_G_UA py Optional turn off bytecode pyc files import sys sys dont write bytecode True from UQ DoE import lhs from UQ analyze import from UQ test functions import Sobol G from UQ util import scale samples general read param file discrepancy import numpy as np import random as rd Set random seed does not affect quasi random Sobol sampling seed 1 np random seed seed rd seed seed Read the parameter range file and generat
4. 0 33333333 66666667 1 a 00 from UQ RSmodel import gp SVR DI kiN BayesianRidge Y 9 OrdinaryLeastSquares LAR Lasso Ridge SGD RF pf dict 3 num vars 8 names x1 x2 x3 x4 R I x6 x7 x8 boun 10 from UQ optimizetion import SCE ASMO 11 from UQ test functions import Sobol G SAC s m 1 1 12 from UQ util import scale samples general read param file discrepancy 13 import numpy as np 14 import random as rd 15 16 4 Exa 18 19 20 seed 1 21 np random seed seed 22 rd seed seed 23 24 j i 25 param file UQ test functions params Sobol_G txt WS read parem file param file Spyder 2 2 5 internal shell on Python 2 7 6 32bits Windows scht gt gt gt D UQ PyL UQ optimization SCE py 79 SyntaxWarning name functn is assigned to before E global declaration A E global functn A yl D UQ PyL UQ optimization SCE py 131 SyntaxWarning name functn is assigned to before global B declaration E d global functn Mn i TNS E Parameter Mu Sigma Mu Star Mu Star Conf 34 param values morris oat sample 10 pf num vars num levels 4 grid jump gt 2 x1 0 706156 2 641627 2 640762 0 445780 mti x2 0 127724 1 719118 1 542336 0 482545 S6 4 param values symmetric LH sample S00 pf nus x3 0 039390 0 588605 0 542817 0 148633 A gie x4 0 118547 0 313918 0 295576 0 098900 para xS 0 001200 0 025919 0 024397 0 005641 em x6 0 002367
5. Show Optimization Results z Choose Optimization method and show results Step 2 Choose parameter optimization method and show results lt gt Choose parameter optimization method like Shuffled Complex Evolution lt gt Click Show Results button to show parameter optimization results UQ PyL gives the tabular and graphic results BESTF 6 666600 BESTA 6 49999998 08 53656488 0 34085105 08 56424082 6 49618645 6 61533234 8 57451195 M 76638417 1 WORSTF 6 666614 WORST B 50000747 0 53512092 M 34217853 M 5639931 8 48957986 6 61521699 8 57198135 M 76636876 1 Evolution Loop 24 Trial 1288 BESTF 6 666006 BESTS 6 49999999 M 53686945 M 3412157 56432256 CHE K A M 61545587 8 57439776 8 76657372 1 WORSTF 6 666004 WORST 6 50000185 0 53602233 0 34065989 M 5641151 8 49009352 0 61518117 8 57425417 8 76617901 1 THE POPULATION HAS CONUERGED TO f PRESPECIFIED SMALL PARAMETER SPACE SEARCH WAS STOPPED AT TRIAL NUMBER 1288 NORMALIZED GEOMETRIC RANGE 8 00070 THE BEST POINT HAS IMPROVED IN LAST 18 LOOPS BV 251 225761 Plu At Lie cx MRE Mis 50 OO SH oo SBE This step can also implemented using python script Python script file Sobol G Optimization py Optional turn off bytecode pyc files import sys sys dont write bytecode True import shutil 51 from UQ optimization import SCE ASMO DDS PSO from UQ util import scale samples general read param fil
6. lhs sample 50 pf num vars criterion center res discrepancy evaluate param values print res Samples are given in range 0 1 by default Rescale them to your parameter bounds scale samples general param values pf bounds np savetxt Input SobolX txt param values delimiter 9 Run the model and save the output in a text file This will happen offline for external models Y Sobol G predict param values np savetxt Output Sobol txt Y delimiterz 4 1 3 Statistical Analysis In this section we do statistical analysis using UQ PyL 33 There are also three steps 1 Define parameter and model information 2 Do Design of Experiment or load Design of Experiment results 3 Choose statistical analysis method and show the results Lj UQ PyL Uncertainty Quantification Python Laboratory mim Problem Definition Design of Experiment ncertainty Analysis Y Sensitivity Analysis Surrogate Modelling Optimilah Perform Design of Experiment Load parameter file D UQ PyL UQ test_functions params Sobol_G txt Choose Parameter File Load Model File D UQ PyL UQ test functions Sobol G py Choose Model File Choose DoE method Monte Carlo Define parameter and model information Number of Sample Points 50 T Generate DoE Script Execute DoE Script Choose Analysis Method Load parameter file D UQ PyL VQ test functions params Sobol G txt Choose Parameter File Load data file inp
7. 8 58431708 8 58439443 THE BEST POINT HAS IMPROUED IN LAST 18 LOOPS BV CONUERGENCY HAS ACHIEVED BASED ON OBJECTIVE FUNCTION SEARCH WAS STOPPED AT TRIAL NUMBER NORMALIZED GEOMETRIC RANGE 8 800133 766 THE BEST POINT HAS IMPROVED IN LAST 16 LOOPS BY pim e y 2 gt E q q E m ke m a Kl d DN T E ke o 6 8 Evolution Loop A 486680563 6 48866651 A 48674597 A 48579287 LESS THAN THE 0 075609 8 58179596 8 58128183 8 5815977 8 58188545 THRESHOLD 6 166666 CRITERIA 71 00 86imBdv Trace of model value Model value o Oo N Evolution Loop Also the result is different from run the same algorithm on the original Sobol G function model 4 4 Use Interactive UQ PyL Software 4 4 1 How to run interactive UQ PyL Software In version 1 1 we have an interactive version of UQ PyL software Double click the PL UQ PyL Interactive UQ PyL Interactive icon you can enter the software Also you can run UQ PyL main_interactive pyw file to enter into the interactive version of UQ PyL software Below is the main page of the software 72 x python examplepy UO Pyl Interactive Environment GO RER File Edit UQ PyL About aw p Key Type Size Value import sys sys dont write bytecode True Y flo 90 array 0 92016797 0 93247791 1 17311737 0 65173187 0 64312815 om UQ DoE import monte carlo normal sobol
8. Choose Model File Random Latin Hypercube Center Latin Hypercube C Maximin Latin Hypercube C Center Maximin Latin Hypercube Correlation Latin Hypercube 50 Choose DoE method and Execute DoE Script define number of sample points Show Design of Experiment Result Choose Result File Display Result Step 2 Choose DoE method Choose Result File lt gt Choose DoE method like Latin Hypercube choose one specific Latin Hypercube method like Center Latin Hypercube lt Set Number of Sample Points like 50 31 E UQ PyL Uncertainty Quantification Python Laboratory B Problem Definition Design of Experiment Uncertainty Analysis Sensitivity Analysis Surrogate Modelling Optimilah Load Model Information Choose Parameter File D UQ PyL UQ test functions params Sobol G txt Choose Parameter File Choose Model File D UQ PyL UQ test functionz Sobol G py Choose Model File Design of Experiment Method Choose DoE method Latin Hypercube x Latin Hypercube Configuration Choose different Latin Hypercube method C Random Latin Hypercube 8 Center Latin Hypercube C Maximin Latin Hypercube Center Maximin Latin Hypercube O Correlation Latin Hypercube Number of Sample Points 50 2 Generate DoE Script Generate and run the script Execute DoE Script Show Design of Experiment Result Choose Result File Choose Result File Display Result
9. Optional turn off bytecode pyc files import sys sys dont write bytecode True from UQ DoE import morris oat from UQ test functions import SAC from UQ util import scale samples general read param file discrepancy import numpy as np import random as rd Set random seed does not affect quasi random Sobol sampling seed 1 np random seed seed rd seed seed Read the parameter range file and generate samples param file UQ test functions params SAC txt pf read param file param file Generate samples choose method here param values morris oat sample 20 pf num vars num levels 10 60 grid jump 5 Samples are given in range 0 1 by default Rescale them to your parameter bounds scale samples general param values pf bounds np savetxt Input Sobol txt param values delimiterz Run the model and save the output in a text file This will happen offline for external models Y SAC predict param values np savetxt Output Sobol txt Y delimiter 4 2 3 Sensitivity Analysis Then we do sensitivity analysis for 13 parameters of SAC SMA model E UQ PyL Uncertainty Quantification Python Laboratory E ADOUT Definition Design of Experiment Uncertainty Analysis Sensitivity Analysis Surrogate Modelling Optimization L I Perform Design of Experiment Load parameter file D UQ PyL UQ test functions params SAC txt Choose Par
10. analysis Init DY main py confidence py correlations py delta py dgsm py extended fast py hypothesis py H HE HE H GR dB zb ok HE 4 Ensure all needed files are loaded For GUI uses Box behnken design Central composite design FAST sensitivity analysis design Faure design Factorial design DGSM sensitivity analysis design Factorial design Full Factorial design Good Lattic Point design Halton Quasi Monte Carlo design Hammersley Quasi Monte Carlo design Latin Hypercube design Monte Carlo design Morris One at A Time design Plackett Burman design Sobol sensitivity analysis design Sobol Quasi Monte Carlo design Symmetric Latin Hypercube design Ensure all needed files are loaded For GUI uses Confidence Interval Correlation analysis Delta sensitivity analysis DGSM sensitivity analysis FAST sensitivity analysis Hypothesis Test 32 29 34 33 36 dd So 39 40 41 42 43 44 45 46 4 48 49 50 9 52 Do 54 99 56 Sl 58 p 60 61 62 os 64 O5 66 moments py mOPrrisspy sobol analyze py sobol svm py analysis RSmodel LARES Oy main py BayesianRidge py Rer ElasticNet py gP Py kNN py LAR py Lars py Lasso py MARS py OrdinaryLeastSquares py regression RE spy Ridge py col spy SVR py optimization Anit Py main py ASMO py DDS ei MCMC PY POO DY SA py SE py optimization uti
11. choose Parameter Distribution Step 2 Click Add button to save this parameter information to table widget Step 3 Enter every parameter s information click Save to Parameter File button choose the save path UQ PyL UQ test functions params Sobol G txt E UQ PyL Uncertainty Quantification Python Laboratory e B Problem Definition Design of Experiment Uncertainty Analysis Sensitivity Analysis Surrogate Modelling Optimi tad Add Input Variables Parameter Name b um x3 Parameter Lower Bound Choose parameter information 0 00 Input Variables Parameter Upper Bound 1 00 Parameter Distribution Uni form Driver Generator Show input variables Parameter Name Parameter Lower Bound Parameter Upper Bound Parameter Distribution E 1 x1 0 00 1 00 Uniform 2 x2 0 00 1 00 Uniform 3 x3 0 00 1 00 Uniform v Save to Parameter File Click to save parameter file 4 1 2 Design of Experiment After problem definition we do Design of Experiment the experiment has three 29 steps 1 Detine parameter and model information 2 Choose Design of Experiment method 3 Generate script and run the script Lj UQ PyL Uncertainty Quantification Python Laboratory A Problem Definition Design of Experiment Uncertainty Analysis Sensitivity Analysis Surrogate Modelling Optimi Lah Load Model Information Choose Parameter File 0 UQ PyL UQ test_functions params Sobol_G txt Cho
12. command prompt Learn More No Later on if you want to make Canopy Python the default you can do so from the preferences dialog Warning If you plan to manually specify the full path to Canopy Python you must specify Canopy s User Python rather than the Canopy installation Python Learn Mare Start using Canopy Choose Yes then click Start using Canopy 17 File Edit Tools Window Help ENTHOUGHT Hi welcome to Canopy CANOPY Log in to your Enthought account or create one Package Manager Doc Browser Training on Demand Recent files Restore previous session 7 No recent files Open an existing file Version 1 5 5 3123 Checking for updates In Package Manager section you can check what packages in your Python library now Actually you can check your python installation in your python installation path All files are in YourPythonPath User for me is home quanjp swets software Python User The python executable file is in YourPythonPath User bin and all the packages are installed in YourPythonPath User lib python2 7 site packages Step 3 Test your Python installation If you have multiple python environment please specific one Usually modify 18 your bashrc file can do it Add two sentence into your bashrc file export PY THON home quanjp swets software Python User bin export PATH PATH PYTHON Then enter command source bashrc to make you
13. 3 sys dont write bytecode True Y flo 90 array 0 92016797 0 93247791 1 17311737 0 65173187 0 64312815 4 5 from UQ DoE import monte carlo normal sobol lhs box behnken central composite fast sampler pe str 1 UQ test functions params Sobol G txt 6 ff2n finite diff frac fact full fact morris oat plackett burman saltelli symmetric LH 7 from UQ analyze import p flom 90 array 66666667 O 0 33333333 66666667 1 NE amp from UQ RSmodel import gp SVR DI kNN BayesianRidge OrdinaryLeastSquares LAR Lasso Ridge SGD RF pf dit 3 num vars 8 names x1 x2 x3 x4 x5 x6 x7 x8 boun 10 from UQ optimizetion import SCE ASMO 11 from UQ test functions import Sobol G SAC 12 from UQ util import scale samples general read param file discrepancy 13 import numpy as np 14 import random as rd 15 16 4 Example Ru bol Morri r FAST 2 te 18 19 r 20 seed 1 21 np random seed seed 22 rd seed seed 23 24 t t ge f t E 25 param file UQ test functions params Sobol_G txt 26 pf read param file param file Spyder 2 2 5 internal shell on Python 2 7 6 32bits Windows gt gt gt D UQ PyL UQ optimization SCE py 79 SyntaxWarning name functn is assigned to before global declaration 29 pa alus amp carl ample pf ur ar global functn A d ef Dz UQ PyL UQ oprimization SCE py 131 SyntaxWarning name functn is assi
14. Inc build 5658 LLVM build 2335 6 on darwin Type help copyright credits or license for more information gt gt Step 4 Install UQ PyL software Download UQ PyL MacOS version unzip the source code using command tar xvf UQ PyL_Mac tar gz Then enter into the UQ PyL directory cd UQ PyL Mac Enter command to run UQ PyL main page python main pyw or python2 7 main pyw Or run Interactive UQ PyL Software python main interactive pyw or python2 7 main interactive pyw You can see the main page of UQ PyL software 24 eoo 2 UQ PyL Uncertainty Quantification Python Laboratory Design of Experiment Uncertainty Analysis Sensitivity Analysis Surrogate Modelling Optimization Add Input Variables Parameter Name Parameter Lower Bound 0 00 Parameter Upper Bound 1 00 Parameter Distribution EB Uniform Driver Generator Add Reset Show input variables Parameter Name Parameter Lower Bound Parameter Upper Bound Parameter Distribution Save to Parameter File 3 Using UQ PyL 3 1 UQ PyL Flowchart Fig 1 is the flowchart illustrating how UQ PyL executes an UQ task A typical task 1s carried out in three major steps 1 model configuration preparation 2 uncertainty propagation and 3 UQ analysis In the first step the user specifies the model configuration information 1 e parameter names ranges and distributions and the DoE information 1 e the sampling te
15. Perform the sensitivity analysis uncertainty analysis using the model OUtput Specify which column of the output file to analyze zero indexed morris analyze param Tile Input Sobol txt Output SobolX Utxt column 0 4 1 5 Surrogate Modeling Next we do surrogate modeling using UQ PyL There are three steps 1 Define parameter and model information 2 Do specific Design of Experiment or load Design of Experiment results 3 Choose surrogate modeling method and show the results a UQ PyL Uncertainty Quantification Python Laboratory EXE Problem Definition Design of Experiment Statistical Analysis Sensitivity Analysis C Surrogate Modelling Optimi Lar Perform Design of Experiment Load parameter file D UQ PyL UQ test functions params Sobol G txt Choose Parameter File Load Model File D Ug PyL UQ test functions Sobol G py Choose Model File Choose DoE method QuasiMonte Carlo i Define parameter and model information Number of Sample Points 500 Generate DoE Script Execute DoE Script Choose Analysis Method Load parameter file D UQ PyL Ug test functions params Sobol G txt Choose Parameter File Load data file input file output file Choose Input File Choose Output File Surrogate Model Method SYM v Show Results 43 Step 1 Define parameter and model information lt gt lt gt lt gt R Switch to Surrogate Modeling tab Click Choose Parameter Fi
16. Step 3 Run for DoE results lt gt Click Generate DoE Script button to generate DoE script which contains information you just choose lt gt Click Execute DoE Script button to run DoE script Then UQ PyL gives the tabular and graphic results of DoE AN Figure 1 gt E AN Figure 1 ui ZOO SRv 00 SRv Latin Hypercube Sampling o random center maxmin centermaxmin correlate Model Evaluation Results 0 8 Model Evaluation Results 0 2 0 0 y E E E 0 10 20 30 40 50 Sample Number x 0 215726 y 0 549895 The result automatically save in text files the name of files including DoE method used and current time 32 model output latin 2015 05 18 22 12 46 bd 2015 5 18 22 12 Lj sample output latin 2015 05 18 22 12 46 txt 2015 5 18 22 12 This step can also 1mplemented using python script Python script file Sobol_G_DoE py Optional turn off bytecode pyc files import sys sys dont write bytecode True from UQ DoE import lhs from UQ test functions import Sobol G from UQ util import scale samples general read param file discrepancy import numpy as np import random as rd Set random seed does not affect quasi random Sobol sampling seed 1 np random seed seed rd seed seed Read the parameter range file and generate samples param file UQ test functions params Sobol G txt pf read param file param file Generate samples choose method here param values
17. bl bu ngs 2 4 3 Run simulation on surrogate model In Surrogate Modeling part we generate a surrogate model from data sets of original model and save the surrogate model in a pickle file Then we can run simulation on the surrogate model we saved For Design of Experiment part we choose the model file as pickle file then it can run DoE on the surrogate model you created Let s take Sobol G function as an example In section 4 1 5 we have created a surrogate model and saved it as SVRmodel pickle file In Design of Experiment tab we load UQ PyL SVRmodel pickle file as model file all the others as same as section 4 1 2 68 File UQ PyL Uncertainty Quantification Python Laboratory CEN About Problem Definition Design of Experiment gt Statistical Analysis surrogate Modelling Optimi ak Load Model Information Choose Parameter File 0 UQ PyL UQ test_functions params Sobol_G txt Choose Model File D Ug PyL SVRmodel pickle Choose Model File Design of Experiment Method Load SVRmodel pickle as model file Choose DoE method Latin Hypercube Configuration Choose different Latin Hypercube method O Random Latin Hypercube 8 Center Latin Hypercube O Maximin Latin Hypercube Center Maximin Latin Hypercube O Correlation Latin Hypercube Number of Sample Points Generate DoE Script Execute DoE Script Show Design of Experiment Result Choose Result Fite Then we do DoE
18. lhs box behnken central composite fast sampler A Pu str 1 UQ test functions params Sobol G txt ff2n finite diff frac fact full fact morris oat plackett burman saltelli symmetric LH from UQ analyze import p fo 90 array 0 66666667 0 9 33333333 0 66666667 1 from UQ RSmodel import gp SVR DI kNN BayesianRidge uares LAR Lasso Ridge SGD RF pf dict 3 num_vers B names xd x2 x3 ai 28 28 27 ert bow e PA zetion import SCE ASMO ions import Sobol G SAC s mt 1 1 scale samples general read param file discrepancy peram file UQ test functions params Sobol G txt 26 pf read param file param file Spyder 2 2 5 internal shell on Python 2 7 6 32bits Windows gt gt gt D 100 PyLiUQioptimizationiSCE py 79 SyntaxWarning name functn is assigned to before global functn D 0Q PyL UQ optimization SCE py 131 SyntaxWarning name functn is assigned to before global declaration Parameter Mu Sigma Mu Star Mu Star Conf x1 0 706156 2 641627 2 640762 0 445780 param values morris oat sample 10 pf num vars num levels 4 grid jump gt 2 E x2 0 127724 1 719118 1 542336 0 482545 0 039390 0 588605 0 542817 390 0 5 x 547 0 31 5 00 0 02 x6 367 0 0 7 75 0 02 x5 195 0 0 21 41 2015 10 11 B st Yd Interactive version of UQ PyL software The interface is very similar to MATLAB GUI we use Spyder package http pythonhosted org
19. this computer Click Next to continue amp Install for anyone using this computer 1 Install just for me Pythonis v3 Ehe Python Distribution made by Scientists Far Scientists Click Next to continue Ipython x y dias install ich features of Python x y 2 7 6 0 you want to Check the components you want to install and uncheck the components you don t want to sect the type ft components you wish to art SE 2 Python 2 7 6 E y aal DEST space required 473 5MB Position your mouse over a component to see ibs description Pythonis v3 Ehe Python Distribution made by Scientists For Scientists T Choose Custom type to install 4 Jeython x y Choose which Check the components you want to install and uncheck the components you don t want to Select the type of install install is TEN 4 Base Libraries 1 5 0 10 l hel Base Python 1 9 2 24 v setuptools 3 0 12 v requests 2 2 1 1 Jl htmlslib 0 999 2 P i La aen Dex Space required 473 5MB Position your mouse over a component to see its description Pythonis 4 Ehe Python Distribution made by Scientists Far Scientists E CNN For Python option you must check all the package UQ PyL needed ec PyQt 4 9 6 4 NumPy 1 8 0 5 Scipy 0 13 3 6 Matplotlib 1 3 1 4 Scikit learn 0 14 1 4 Please note this one is not checked by default KK Click Next to continue C Python x y 2 7 6 0 Setup CU Choose Inst
20. tit ae Add eae ae E E aE AEE AEE EEE PE PEE EE FUNCTION CALCULATE DESIRE OUTPUT 56 def getOutput Q E Qo functn 0 0 ignore 92 IT 0 ourtile Open ps testol sac day CEM for jj in range ignore lineln outfile readline while 1 lineln outfile readline if lineln break nCols string split lineIn Qe append eval nCols 4 Qo append eval nCols 5 functn functn Oe I Oo 1 O0e 1 Oo 1 I I 1 Outflle close t funetn functn I functn math sgrt functn return functn tit ae at ae ae eae ea eae aE E E AEE aE EE PE EE PEE EE MAIN PROGRAM def predict values pf read param file controlFileName for n in range pf num vars pf names n UO pf names n Y np empty values shape 0 os chdir D UQ PyL UQ test functions SAC for i row in enumerate values inputData values i genAppInputFile inputData appInputTmplts appInputFiles pf num vars p names runApplication Y i getOutput 57 print Job ID str itl return Y 4 2 2 Design of Experiment We do Design of Experiment for SAC SMA model 2 UQ PyL Uncertainty Quantification Python Laboratory zn Problem Definition Design of Experiment Y Uncertainty Analysis Sensitivity Analysis Surrogate Modelling Optimi dar Load Model Information Choose Parameter File 0D VQ PyL UQ test_functions params SAC txt Choose Parameter File
21. to this location home quan p Canapy Press Enter to accept this location Press CIRL C to abort or specify an alternate location Please ensure that your location contains only ASCII letters numbers and the following punctuation Chars 9 7 Pr fhome quanjp Canopy gt gt gt fhome quanjp swgfs software Canopy Type the path you want to install Canopy then press Enter to continue 15 Installing to home quanjp swgfs software Canapy please wait Must specify the vendor namespace for these files with vendor No directories in update desktop database search path could be processed and updated LE E Updating MIME database in home quan p local share mime Wrote 2 strings at 20 44 Wrote aliases at 44 4E Wrote parents at 48 de Wrote literal globs at dc 5 Wrote suffix globs at 50 108 Wrote full globs at 108 10c Wrote magic at 10c 118 Wrote namespace list at 118 lic LET done You can run the Canopy graphical environment by running the script fhome quanjp swots so0ttware Canopy canopy or by selecting Canopy in your Applications menu On your first run your Canopy User Python environment will be initialized and you will have the opportunity to make Canopy be your default Fython at the command line Details at support enthought com forums Ihank you for installing Canopy Complete to install Canopy Step 2 Setting up Canopy environment Enter into the Canopy directory for me
22. variety of formats including png bmp tiff or pdf formats among others Fig 3 shows the interactive version of UQ PyL software In this page you can write down python script to achieve UQ analysis and run the script to obtain the results You can see the output results and internal variables values through the page 26 Ele About Problem Definition Design of Experiment Statistical Analysis Sensitivity Analysis Surrogate Modelling Optimi Add Input Variables Parameter Name gt am o Parameter Lower Bound 0 00 Input Variables Parameter Upper Bound 1 00 Parameter Distribution ES Uni form Driver Generator Show input variables Parameter Name Parameter Lower Bound Parameter Upper Bound Parameter Distribution Save to Parameter File Fig 2 Graphic User Interface of UQ PyL Main Page File Edit UQ PyL About Dg b pDet3x E eem me P Value 1 4 Opt 5 2 import sys Koy Lad Ls 3 sys dont_write_bytecode True Y flo 90 array 0 92016797 0 93247791 1 17311737 0 65173187 0 64312815 4 5 from UQ DoE import monte carlo normal sobol lhs box behnken central composite fast sampler pe str 1 UQ test functions params Sobol G txt 6 ff2n finite diff frac fact full fact morr s oat plackett burman saltelli symmetric LH 7 from UQ analyze import p fo 90 array 0 66666667 O
23. 0 039866 0 034427 0 012365 Mamm x7 0 013475 0 027525 0 023902 0 011979 HM x8 0 007195 0 034502 0 029983 0 011464 43 1 k 44 Y para tlus ple 45 finite diff sam 46 re repe a e para slue 4 ar Y lt gt Fig 3 Interactive Version of UQ PyL Software 27 4 Examples 4 1 Sobol g function 4 1 1 Problem Definition The expression of sobol g function is f x gix where 4x 2 aj ita The input parameter x is uniformly distributed within 0 1 aj 10 1 4 5 9 99 99 99 991 The model is implemented using Python and the parameter file is shown below Model file UQ PyL UQ test_functions Sobol_G py from future import division gi xi import numpy as np Non monotonic Sobol G Function 8 parameters First order indices xls D J7169 Tella x21 0 1791 19 9429 xor DUST 2 509 x4 0 0072 0 78 Ron 0 000 Ds HE def predict values a I0 l1 4 5 By 95 99 99 99 Y np empty values shape 0 for i row in enumerate values Y i 1 0 for j in range 8 x row 3 YHJ abs 4 x 2 all 7 G all return Y 28 Parameter file UQ PyL UQ test functions params Sobol G txt xl p0 JD x Dal 22 0 xo DIO 20 x4 0 0 1 0 xo DO 140 xo DD 3 x DU XL x59 0 0 Es 0 Parameter file can also be generated from GUI of UQ PyL Step 1 Enter Parameter Name Parameter Lower Bound and Parameter Upper Bound
24. 11 2H 28 4H txt C Python2 slibssite packages sklearnocross validation py 1137 DeprecationWarni ny Passing function as score func is deprecated and will be removed in 8 15 Either use strings or score ohjects The relevant new parameter is called sco ring scoring scoring The k fold mean square error cross validation scores are H 43H7H918 H 35159512 H 393588897 3H 36591365 68 31136287 H 462HBHBH58 6 36738373 6 36966313 86 486746766 6 48561528 1 The mean value of the scores is 6 386529869176 The standard deviation of the scores is 6 6465546713762 45 400 ET In this new version of UQ PyL the software also save surrogate model as a pickle file automatically For this example is SVRmodel pickle file SVRmodel pickle 2015 10 11 20 44 PICKLE 3244 98 KB This file can be opened by a Text Editor please see the context of this file below 46 SVBmodel pi ckl 1 ccopy reg 2 reconstructor 3 nO 4 zaklearn 2ym classes n MR amp pl 7 c builtin E object 5 pe 10 Ntp3 11 Rp4 12 dps 13 5 impl 14 ne 15 S epailon svr TG p7 17 s5 kernel 18 n8 BENE be 20 D I 21 s5 verbose 22 pl0 23 100 24 s5S probability 25 pil 26 100 27 35 label 2B pl 29 gnumpy core multiarray 30 _ reconstruct al pls 32 cnumpy 34 pl 4 35 IO 36 tpl5 EMA 38 pl BS Ep 7 40 Rp18 It saved the data structure of the surrogate model you built In section 4 3 we w
25. 34427 0 012365 0 013475 0 027525 0 023902 0 011979 x8 0 007195 0 034502 0 029983 0 011464 00 A BET oo Re 74
26. 5201 1 246218 8 361561 6 049704 6 677892 8 583383 6 127671 H H31H89 8 232696 6 174826 6 842758 8 660593 8 831374 8 822963 4 865992 H 6658601 8 639418 6 829168 8 687556 8 666185 8 824323 6 816861 4 684874 8 666736 8 834242 6 625536 8 666663 41 W Figure 1 OO A BET This step can also implemented using python script Python script file Sobol G SA py Optional turn off bytecode pyc files import sys sys dont write bytecode True from UO DoE import morris oat from UQ analyze import from UQ test functions import Sobol G from UQ util import scale samples general read param file import numpy as np import random as rd Set random seed does not affect quasi random Sobol sampling seed 1 np random seed seed rd seed seed Read the parameter range file and generate samples param file UQ test functions params Sobol G txt pf read param file param file Generate samples choose method here param values morris oat sample 50 pf num vars num levels 10 grid jump 5 Samples are given in range 0 1 by default Rescale them to your 42 parameter bounds scale samples general param values pf bounds np savetxt Input SobolX txt param values delimiter 9 Run the model and save the output in a text file This will happen offline for external models Y Sobol G predict param values np savetxt Output SobolX txt Y delimiter
27. 8 42740023 77136507 Evolution Loop 23 Trial 1246 BESTF 6 608000 BESTX 6 49999998 6 50684122 4427293 6 526616 6 44431368 6 6048984 8 43158307 77078896 1 WORSTF 6 680001 WORST 6 49999913 00 5027474 6 44747874 0 51922003 0 44448914 6 660441623 8 42740023 M 77136507 THE POPULATION HAS CONVERGED TO f PRESPECIFIED SMALL PARAMETER SPACE SEARCH WAS STOPPED AT TRI NUMBER 1246 NORMALIZED GE ao Trace of the different parameters Trace of model value 300 Da e 7 e o 025 N ul o Model value o N o Parameters value N o o 0 15 150 100 15 20 25 30 35 5 10 15 20 25 30 35 Evolution Loop Evolution Loop This step can also implemented using python script Python script file SAC_Optimization py Optional turn off bytecode pyc files import sys sys dont write bytecode True import shutil 67 from UQ optimization import SCE from UQ util import scale samples general read param file discrepancy import numpy as np import random as rd Read the parameter range file param file UQ test functions params SAC txt bl np empty 0 bu np empty 0 pf read param file param file for i b in enumerate pf bounds bl np append bl b 0 bu np append bu b 1 dir UQ test functions shutil copy dirt SAC py Gir tunctn py Run SCE UA optimization algorithm SCE sceua
28. SA results Mu star Confidence Interval 12 10 Sigma 0 UZTWMZFWMUZK PCTIMADIMPZPERCREXPLZTWNIZFSM ZFPMLZSK LZPK PFREE Mu star This step can also implemented using python script Python script file SAC SA py Optional turn off bytecode pyc files import sys sys dont write bytecode True from UQ DoE import morris oat from UQ analyze import from UQ test functions import SAC from UQ util import scale samples general read param file import numpy as np import random as rd Set random seed does not affect quasi random Sobol sampling seed 1 np random seed seed 62 rd seed seed Read the parameter range file and generate samples param file UQ test functions params SAC txt pf read param file param file Generate samples choose method here param values morris oat sample 20 pf num vars num levels 10 grid jump 5 Samples are given in range 0 1 by default Rescale them to your parameter bounds scale samples general param values pf bounds np savetxt Input SAC txt param values delimiterz Run the model and save the output in a text file This will happen offline for external models Y 2 SAC predict param values np savetxt Output SAC txt Y delimiterz Perform the sensitivity analysis uncertainty analysis using the model output Specify which column of the output file to analyze zero indexed morris analyze param file Inp
29. ad DoE results lt gt Click Choose Input File button to choose sample file you just generated for example sample output latin2 2015 05 18 22 12 46 txt lt gt Click Choose Output File button to choose model output file you just generated for example model output latin 2015 05 18 22 12 46 txt 35 UQ PyL Uncertainty Quantification Python Laboratory Problem Definition Design of Experiment Uncertainty Analysis y Sensitivity Analysis Surrogate Modelling Optimi tah Perform Design of Experiment Load parameter file D UQ PyL UQ test functions params Sobol G txt Load Model File D UQ PyL UQ test functions Sobol G py Choose DoE method Monte Carlo Number of Sample Points 50 Generate DoE Script Execute DoE Script Choose Analysis Method Load parameter file D UQ PyL UQ test functions params Sobol G txt Load data file input file output file D UQ PyL sample output latin2 2015 05 18 22 12 46 txt Basic Statistical Analysis Methods Statistical Moments Methods Advanced Statistical Analysis Methods Pearson Spearman Correlations Analysis Choose Parameter File Choose Model File Choose Parameter File Choose Input File Choose Output File Show Results Show Results Define uncertainty analysis method and show results SRS Step 3 Choose statistical analysis method and show results lt gt Choose statistical analysis method like Statistical Moments Methods
30. all Location Y J Pythonix y Choose Ehe Folder in which Eo install Pythonis v 2 7 6 0 Setup will install all Pythonis yi components in the Following Folder Installation Folders of included packages may be customized see previous page To install in a different Folder click Browse and select another Folder Click Mexk to continue FyEhan x v Base Installation Directory SU Program Pilesipsthonzy Space required 535 7MB Space available 25 036 Click Next to continue 3 Pythontz y 2 7 6 0 Setup Choose Start Menu Folder 1 Y ythonix y Choose a Start Menu Folder For the Pythonis wi 2 7 6 0 shortcuts Select Ehe Start Menu Folder in which vau would like to create the program s shortcuts You can also enter a name to create a new folder EndNote Foxit Software Google Chrome OU ve Microsoft Office Microsoft Silverlight Py GPL v4 8 1 For Python v2 7 Python 2 7 F Do nat create shortcuts SS Click Install then waiting for the installation process After installation you executable python exe file will be C Python27 python exe All the package will be in the C Python27 Lib site packages directory Step 2 Install UQ PyL software Please download UQ PyL Windows version double click to run the installation file 10 Installing UQ PyL Software Destination folder Extraction progress Extracting files to D folder Extracting from UQ PyL exe Extracting UQ PyL UQ pptimization e
31. ameter File Load Model File D Ug PyL UQ test functions SAC py Choose Model File Step1 Load parameter file and driver file Design of Experiment Method Choose DoE method Morris One at A Time m Morries One At A Time MOAT Configuration Number of total sample points dimensiontl Number of Trajectories Number of Trajectories 50 5 Generate DoE Script Execute DoE Script Step2 Load Design of Experiment results Choose Analysis Method Load parameter file D UQ PyL UQ test_functi ons params SAC txt Choose Parameter File Choose Input File Choose Output File Step3 Choose sensitivity analysis method and show results Step 1 Define parameter and model information lt gt Choose Sensitivity Analysis tab lt gt Load parameter file UQ PyL UQ test functions params SAC txt and model file driver file UQ PyL UQ test functions SAC py 61 Step 2 Load DoE results lt gt Load DOE results sample input file UQ PyL UQ test functions SAC sample output morris 2015 05 19 21 34 2 6 txt and model output file UQ PyL UQ test functions S AC model output morris 2015 05 19 21 34 26 txt Step 3 Choose sensitivity analysis method and show results lt gt Choose sensitivity analysis method Morris and click Show Results button to acquire sensitivity analysis results UQ PyL gives the tabular and graphic results W Figure 1 O OO A BET Morries One at A Time
32. analysis it can obtain tabular and graphic results hOO EA 69 00 87 Model Evaluation Results H La H o o o 2 i E i A m i m 2 if Ei o o z o co 20 30 Sample Number The result is different from run the same algorithm on the original Sobol G function model For Parameter Optimization part we also choose the model file as pickle file then it can run global optimization algorithm on the surrogate model E UQ PyL Uncertainty Quantification Python Laboratory a Definition Design of Experiment Statistical Analysis Sensitivity Analysis Surrogate Modelling Optimization gt Load Data Load Parameter File D UQ PyL Ug test functions params Sobol G txt Choose Parameter File Load Model D Ug PyL SVRmodel pickle Choose Model File Clowes EE ERREECHT Load SVRmodel pickle as model file Optimization Method Shuffled Complex Evolution v Show Results Show Optimization Results Then we do SCE parameter optimization algorithm 1t can obtain tabular and graphic results 70 van Ch 6 47869971 63886959 6 3988961 8 45837269 M 54625876 1 WORSTF 6 576468 WORSTR 6 47925921 0 64045023 6 46132286 8 45816132 0 54382353 Evolution Loop 14 Trial 766 BESTF 6 5 76394 BEST 0 47827325 8 63882823 6 398 7 7878 6 45837516 6 549126 1 WORSTF 6 576468 WORST 6 47766161 3 8 63716216 39665221 6 45862374 M 535954161 M 58434864 6 58441515
33. ast we do parameter optimization using UQ PyL There are two steps 1 Define parameter and model information 2 Choose parameter optimization method and show the results 48 EN UQ PyL Uncertainty Quantification Python Laboratory E Definition Design of Experiment Statistical Analysis Sensitivity Analysis Surrogate Modelling Optimization a Load Data Load Parameter File D UQ PyL UQg test functions params Sobol G txt Choose Parameter File Load Model D UQ PyL UQ test functions Sobol G py Choose Model File Choose ptimization Method Load parameter and model information Opt imization Method Shuffled Complex Evolution Show Results Show Optimization Results Step 1 Define parameter and model information lt gt Switch to Optimization tab lt gt Click Choose Parameter File button to choose UQ PyL UQ test functions params Sobol G txt file lt gt Click Choose Model File button to choose UQ PyL UQ test functions Sobol G py file 49 UQ PyL Uncertainty Quantification Python Laboratory x Definition Design of Experiment Statistical Analysis Sensitivity Analysis Surrogate Modelling Optimization 4 Load Data Load Parameter File D UQ PyL UQ test functions params Sobol G txt Choose Parameter File Load Model D UQ PyL UQ test functionz Sobol G py Choose Model File Choose ptimization Method Optimization Method Shuffled Complex Evolution sel Show Results
34. cal Analysis Statistical moments Confidence interval Hypothesis test 1 2 3 Sensitivity Analysis Morris One at A Time MOAT Derivative based Global Sensitivity Measure DGSM Sobol Sensitivity Analysis Fourier Amplitude Sensitivity Test FAST Metamodel based Sobol Correlation analysis Delta Moment Independent Measure Delta Multivariate Adaptive Regression Splines MARS based sensitivity analysis 1 2 4 Surrogate Modeling Generalized Linear Model Ordinary Least Squares Ridge Regression Lasso Least Angle Regression LARS Lasso Bayesian Regression and Elastic Net Regression 3 Tree Random Forest Nearest Neighbors Support Vector Machine Gaussian Process MARS Stochastic Gradient Descent 1 2 5 Parameter Optimization Shuffled Complex Evolution SCE Dynamically Dimensional Search DDS Adaptive Surrogate Modeling based Optimization ASMO Particle Swarm Optimization PSO Simulated Annealing SA and Monte Carlo Markov Chain MCMC 1 3 Overview about functionality of the UQ PyL package CO sl Oy OF A GQ N np 10 TE E 13 14 12 16 17 18 19 20 21 22 GE 24 25 26 Z 2 0 29 30 J1 ANTE Dy DoE EHE Y main py box behnken py central composite py fast samp Lerc py faure py It2npy finate Ort py frac Factspy fut tact py GLPypy halton py hammersley py lng pY monte carlo py NHOIris odtspy piackett Dburman py Saltelli py Sta py symmetric LH py
35. chniques and sample sizes to prepare for UQ exercise for a given problem In the second step the different sample parameter sets generated in the last step are fed into the simulation model or mathematical function to enable the execution of simulation model function calculation In the third step a variety of UQ exercises are carried out including statistical analysis SA surrogate modelling and parameter optimization 25 Parameter Name Parameter Range Parameter Distribution i model configuration preparation Control Template Control File a Uncertainty Propagation E I I I I L e I I e Outputs UO Analysis Parameter Y a a A a Tabular and Graphic Analysis Results Fig 1 UQ PyL flowchart 3 2 UQ PyL Main Frame UQ PyL 1s equipped with a Graphic User Interface GUI to facilitate execution of various functions but it can also run as a script program in a batch mode Fig 2 shows the main page of UQ PyL Different tab widgets allow user to execute different steps of UQ process including problem definition DoE Statistical Analysis SA Surrogate Modeling and Parameter Optimization One may click on the desired tab by mouse and or enter the required information via keyboard to perform various tasks After a task 1s completed the software generates tabular results and or graphical outputs The graphical outputs can be saved in a
36. e import numpy as np import random as rd Read the parameter range file param file 2 UQ test functions params Sobol G txt bl np empty 0 bu np empty 0 pf read param file param file for i b in enumerate pf bounds bl np append bl b 0 bu np append bu b 1 dir UQ test functions Ssiutil copy dirt Sobol G py dlire functu py Run SCE UA optimization algorithm SCE sceua bl bu pf ngs 2 4 2 SAC SMA model 4 2 1 Problem Definition The SAC SMA is a rainfall runoff model which has a highly non linear non monotonic input parameter model output relationship There are sixteen parameters in the SAC SMA model Thirteen of them are considered tunable and the other three parameters are fixed at pre specified values according to Brazil 1988 Table 1 describes those parameters and their ranges No Parameter Description 1 10 0 300 0 2 5 0 150 0 3 0 10 0 75 4 PCTIM Impervious fraction of the watershed area decimal 0 0 0 10 fraction 5 ADIMP Additional impervious area decimal fraction 0 0 0 20 6 ZPERC Maximum percolation rate dimensionless 5 0 350 0 REXP Exponent of the percolation equation 1 0 5 0 dimensionless 8 LZTWM Lower zone tension water maximum storage mm 10 0 500 0 52 9 LZFSM Lower zone supplemental free water maximum 5 0 400 0 storage mm 10 LZFPM Lower zone primary free water maximum storage 10 0 drainage ra
37. e samples param file UQ test functions params Sobol G txt pf read param file param file Generate samples choose method here param values lhs sample 50 pf num vars criterion center res discrepancy evaluate param values print res 37 Samples are given in range 0 1 by default Rescale them to your parameter bounds scale samples general param values pf bounds np savetxt Input Sobol txt param values delimiter Run the model and save the output in a text file This will happen offline for external models Y Sobol G predict param values np savetxt Output Sobol txt Y delimiterz Perform the sensitivity analysis uncertainty analysis using the model Output Specify which column of the output file to analyze zero indexed moments analyze Output Sobol txt column 0 4 1 4 Sensitivity Analysis Next we do sensitivity analysis using UQ PyL There are three steps 1 Define parameter and model information 2 Do specific Design of Experiment or load Design of Experiment results Different sensitivity analysis method need different Design of Experiment method 3 Choose sensitivity analysis method and show the results 38 UQ PyL Uncertainty Quantification Python Laboratory zB ETS N E T Problem Definition Design of Experiment Uncertainty Analysis CSensitivity Analysis J Surrogate Modelling Optimi Lah Load parameter file D UQ PyL UQ test fu
38. gned to before global B nus H o mple i pf m T declaretion E E global functn gi Deren value HEITE Te prt Dun vel dic second order ATuc Parameter Mu Sigma Mu Star Mu Star Conf genre o no ser VET MUERTA AA Ge QURE IN x1 0 706156 2 641627 2 640762 0 445780 A aca nem agem pol um men x2 0 127724 1 719118 1 542336 0 482545 RIIT epit peri ape ET o min x3 0 039390 0 588605 0 542817 0 148633 A H ET Gage ep SE x4 0 118547 0 313918 0 295576 0 098900 39 HEEL irum a Ro E Xm Ha x5 0 001200 0 025919 0 024397 0 005641 40 m pee ES x6 0 002367 0 039866 0 034427 0 012365 Bom sees x Ap x7 0 013475 0 027525 0 023902 0 011979 peter OU do oo mt NE eddie x8 0 007195 0 034502 0 029983 0 011464 43 box k 44 4 param value entra oss UD prs 45 5 finit ff sam f ta 46 4 res discrepes evaluate pera lues 47 a d Interactive UQ PyL Software 13 2 2 2 Linux platform Canopy is a globally recommended Python distribution It contains Python and 100 common built it packages It also contains all the package UQ PyL used in one software So you can install Canopy for all the dependences UQ PyL needed Please go to the official website https www enthought com products canopy for more information Step 1 Install Canopy software Canopy is a commercial software However it provide free use for academic usage If you use Canopy for education or academic you ca
39. hoose Executable File Generate Driver lt gt Choose Problem Definition tab click on Driver Generator widget lt gt Click Choose Model Input File to load model configuration file for SAC model is UQ PyL UQ test functions SAC ps testO01 sac ze Click Generate Template File to generate model configuration template file this file will be used in model driver file 54 E UQ PyL Uncertainty Quantification Python Laboratory Design of Experiment Uncertainty Analysis Sensitivity Analysis Surrogate Modelling Optimiigk Generate Template File Load Model Input File D UQ PyL UQ test functions SAC ps test l sac Choose Model Input File 4 lal Generate Template File Input Variables Generate Driver Load Parameter File D Ug PyL UQ test functions params SAC txt Choose Parameter File Load Model Input File D UQ PyL UQ test functions SAC ps test l sac Choose Model Input File Driver Generator Load Executable File 0D UQ PyL UQ test_functions SAC mopexcal exe Choose Executable File Generate Driver Generate Python driver file lt gt Click Choose Parameter File to load model parameter file for SAC model is UQ PyL UQ test functions params SAC txt lt Click Choose Model Input File to load model configuration file for SAC model is UQ PyL UQ test functions SAC ps test01l sac lt gt Click Choose Executable File to load
40. ill introduce how to run simulations on surrogate models you built This step can also 1mplemented using python script Python script file Sobol G Surrogate py Optional turn off bytecode pyc files import sys sys dont write bytecode True from UQ DoE import sobol from UQ RSmodel import SVR from UQ test functions import Sobol G from UQ util import scale samples general read param file discrepancy import numpy as np import random as rd 47 Set random seed does not affect quasi random Sobol sampling seed 1 np random seed seed rd seed seed Read the parameter range file and generate samples param file UQ test functions params Sobol G txt pf read param file param file Generate samples choose method here param values sobol sample 500 pf num vars Samples are given in range 0 1 by default Rescale them to your parameter bounds scale samples general param values pf bounds np savetxt Input Sobol txt param values delimiter Run the model and save the output in a text file This will happen offline for external models Y Sobol G predict param values np savetxt Output SobolX txt Y delimiter Perform regression analysis using the model output Specify which column of the output file to analyze zero indexed model SVR regression Input Sobol txt Output Sobol txt column 0 cv True 4 1 6 Parameter Optimization At l
41. is home quanjp swefs software Canopy you can see the file inside it quanjp login02 Canopy 11 total 336 drwxrwxr x drwxrwxr x rw rw r rwXr Xr x LA co quan p quanjp quan p quanjp quan p quanjp quanjp quanjp quanjp quanjp quanjp quanjp quan p quanjp quan p quanjp quanjp quanjp quan p quanjp E Li zl h m i A e O cua Pd _ boot py canopy canopy cli canopy desktop canopy mime xml LO A h3 J CD IWH C E us hn LA io coc de pa e e rwXIWXIr X rw rIw r FrW IWw I drwxrwxr x drwxrwxr x zb His CH e cC d d cu EUN LM LLL Pt C L ma m LL e Oh ca Ki C L f bi Run canopy to setting up Canopy software 16 Canopy System and User environment locations Your Canopy environment will be installed in the location shown below You may change it if you wish to What s this Canopy environment directory home quanjp swgfs software Python Continue Enter the Canopy environment directory for me is home quanjp swgfs software Python click Continue to continue Your python installation will in this directory Setting up your Canopy environment E After that a dialogue will display Do you want to make Canopy your default Python environment This will give you direct access to Canopy Python and to utilities like IPython easy install nosetests from your terminal
42. le button to choose UQ PyL UQ test_functions params Sobol_G txt file Click Choose Model File button to choose UQ PyL UQ test_functions Sobol_G py file UQ PyL Uncertainty Quantification Python Laboratory E Problem Definition Design of Experiment Statistical Analysis Sensitivity Analysis Surrogate Modelling Optimi tad Perform Design of Experiment Load parameter file D UQ PyL UQ test_functions params Sobol_G txt Choose Parameter File Load Model File D Ug PvL Ug test functions Sobol G py Choose Model File Choose DoE method Quasi Monte Carlo Number of Sample Points 500 Generate DoE Script Execute DoE Script Choose Analysis Method Load parameter file D Ug PvyL Ug test functions params Sobol G txt Choose Parameter File Load data file input file output file Choose Input File Choose Output File Surrogate Model Method SYM v Show Results Do Design of Experiment and load results OR Load Design of Experiment results directly Step 2 Do DoE for surrogate modeling method and load results gt gt gt 4 Choose DoE method for example Quasi Monte Carlo Set Number of Trajectories for example 500 Click Generate DoE Script button to generate script Click Execute DoE Script button to run script and acquire DoE result Load input output file you just generated 1 Click Choose Input File button to load sample file for example UQ PyL
43. ls _ ANTE apy discrepancy py spyderlib spyderplugins Statistics moments method MOAT sensitivity analysis Sobol sensitivity analysis Metamodel based sobol sensitivity Ensure all needed files are loaded Por GUL uses GLP Bayesian Ridge regression Decision Tree regression GLP Elastic Net regression Gaussian Process regression k nearest neighbor regression GLP LAR regression GLP Lars regression GLP Lasso regression HE HE YS 4 4 MARS regression GLP Ordinary Least Squares t Random Forest regression GLP Ridge regression Stochastic Gradient Descent regression Support Vector Machine regression Ensure all needed files are loaded For GUI uses ASMO optimization DDS Optimization Monte Carlo Markov Chain optimization Particle Swarm Optimization Simulated Annealing optimization dB 4 Shuffled Complex Evolution Ensure all needed files are loaded Compute discrepancy of design Spyder package Spyder package 2 Installation 2 1 Dependencies UQ PyL is an open source package written in Python language It runs on all major platforms Windows Linux MacOS It requires some pre installed standard Python packages Python version 2 7 6 Numpy gt 1 7 1 Scipy gt 0 16 0 Matplotlib gt 1 4 3 PyQt4 If you use graphic user interface Scikit learn 0 14 1 9995429 2 2 Detailed Installation 2 2 1 Windows platform For Windows platform
44. model executable file for SAC model is UQ PyL UQ test functions SAC mopexcal exe ze Click Generate Driver button to acquire model driver file The driver file UQ PyL UQ test_functions SAC py shows below import os import math import string import numpy as np from util import read param file tat He at ae TE TE eae E E E E aE AEE AEE EE AAA USER SPECIFIC SECTION controlFileName D UQ PyL UQ test functions params SAC txt 55 appinputriles ps test l sac appInputTmplts applInputFiles Tmplt at He tate ae eae ae AE aE aE a AEE aaa EEE aE PEPE EEE FUNCTION GENERATE MODEL INPUT FILE genAppInputFile inputData appTmpltFile appInputFile nInputs inputName s infile open appTmpltFile r outfile open appInputFile w while 1 lineln infile readline E linea Y break lineLen len lineln newLine lineln if nInputs gt 0 for find in range nInputs strLen len inputNames fInd sind string find newLine inputNames fInd if sind gt O0 sdata 7 3f inputData fInd strdata str sdata next sInd strLen lineTemp newLine 0 sInd strdata newLine next lineLen 1 newLine lineTemp lineLen len newLine outfile write newLine infile close outfile closet return E E AE TE TE TE TE TE TE TE TE TE dd AE AE dd dd dd dd dd dd dd dg FUNCTION RUN MODEL def runApplication sysComm mopexcal exe os system sysComm return
45. n download canopy 1 5 5 full rh5 64 sh from our website or from Canopy official website After downloading you should install Canopy by steps below chmod 755 canopy 1 5 5 full rh5 64 sh canopy 1 5 5 full rh5 64 sh Welcome to the Canopy 1 5 5 installer Io continue the installation you must review and approve the license term agreement Tess Enter to continue If you approve the license term press Enter to continue 14 Canopy Product License Express Canopy Express Software License Agreement Basic Professional Canopy Subscription License Agreement Academic Canopy Software License for Academic Use Please review your applicable license carefully By installing or using a Canopy product you siqnify your assent to and acceptance of the terms of the applicable license to Canopy If vou do not accept the terms of the applicable license then you must not use the Canopy products Should you have any questions regarding licensing please contact us at support enthought com ENTHOUGHT CANOPY EXPRESS Software License Agreement Ihis Enthought Canopy Express Software License Agreement the greement is between Enthought Inc a Delaware corporation 2 nthought and the licensee subscriber who accepts the terms of this Agreement the ustomer The effective Do you approve the license terms yes no no gt gt gt yes no y Type yes then press Enter to continue Canopy will be installed
46. nctions params Sobol G txt Choose Parameter File Load Model File D AUQ PyL UQ test_functi ons S obol G py Choose Model File EES Define parameter and model information Choose DoE method Morris One at A Time Morries One At A Time MOAT Configuration Number of total sample points dimension 1 Number of Trajectories Number of Trajectories E baal Generate DoE Script Execute DoE Script Choose Analysis Method Load parameter file D UQ PyL UQ test_functions params Sobol_G txt Choose Parameter Fila Load data file input file output file Choose Input File Choose utput File Sensitivity Analysis Method Morris D Show Re sults Step 1 Define parameter and model information lt gt Switch to Sensitivity Analysis tab lt Click Choose Parameter File button to choose UQ PyL UQ test functions params Sobol G txt file lt Click Choose Model File button to choose UQ PyL UQ test functions Sobol G py file 39 uo UQ PyL Uncertainty Quantification Python Laboratory BS Problem Definition Design of Experiment Uncertainty Analysis Sensitivity Analysis Surrogate Modelling Optimiigk Perform Design of Experiment Load parameter file D UQ PyL UQ test functions params Sobol G txt Choose Parameter File Load Model File D Ug PvyL UQg test functions Sobol G py Choose Model File Desizn of Experime
47. nt Method Choose DoE method Morris One at A Time Morries One At A Time MDAT Configuration Number of total sample points dimension 1 Number of Trajectories Number of Trajectories 50 Generate DoE Script Execute DoE Script Do specific Design of Experiment and load results OR Load Design of Experiment results directly Choose Analysis Method Load parameter file D UQ PyL UQ test_functions params Sobol_G txt Choose Parameter File Load data file input file output file Choose Input File Choose Output File Sensitivity Analysis Method Morris X Show Results Step 2 Do specific DoE for specific sensitivity analysis method For example we do Morris analysis in this chapter Then load DoE results Choose DoE method for this experiment is Morris One at A Time Set Number of Trajectories for example 50 Click Generate DoE Script button to generate script Click Execute DoE Script button to run script and acquire DoE result Load input output file you just generated 1 Click Choose Input File button to load sample file for example UQ PyL sample output morris 2015 05 19 17 54 55 txt 2 Click Choose Output File button to load model output file for example UQ PyL model output morris 2015 05 19 17 54 55 txt Po 40 E UQ PyL Uncertainty Quantification Python Laboratory 2s Problem Definition Design of Experiment Uncertainty Analysis ensitivit
48. odel SVR regression Input SAC Output SAC column 0 cv True 4 2 5 Parameter Optimization 2 UQ PyL Uncertainty Quantification Python Laboratory O File About Optimization Load Parameter File D UQ PyL UQg test functions params SAC txt Choose Parameter File y Definition Design of Experiment Uncertainty Analysis Sensitivity nalysis Surrogate Modelling Load Data Load Model D UQ PyL UQ test functions SAC py m L choose Model File Step1 Load parameter file and model driver Optimization Method Shuffled Complex Evolution v Show Results TM Step2 Choose optimization method and show results Show Optimization Results Step 1 Define parameter and model information lt gt Choose Optimization tab lt gt Load parameter file UQ PyL UQ test functions params SAC txt and model file driver file UQ PyL UQ test functions SAC py 66 Step 2 Choose optimization method and show results lt gt Choose optimization method Shuffled Complex Evolution and click Show Results button to acquire optimization results UQ PyL gives the tabular and graphic results C 8 42378595 6 771927931 Evolution Loop 22 Trial 1195 BESTF 6 866000 BEST 6 49999998 0 50025442 0 44093632 6 52111712 0 44434248 0B 60504633 8 43312661 98 778055538 WORSTF 6 660001 WORSTX 8 49999913 80 5027474 8 44747874 0 51922003 8 44448914 0B 60441023
49. ose Parameter File Choose Model File D UQg PyL UQ test functionz Sobol G py Choose Model File Desi fE t Method e d cam iis Load parameter file and model file Choose DoE method Latin Hypercube E Latin Hypercube Configuration Choose different Latin Hypercube method Random Latin Hypercube 0 Center Latin Hypercube C Maximin Latin Hypercube Center Maximin Latin Hypercube C Correlation Latin Hypercube Number of Sample Points 50 Generate DoE Script Execute DoE Script Show Design of Experiment Result Choose Result File Choose Result File Display Result Step 1 Define parameter and model information lt gt Switch to Design of Experiment tab lt gt Click Choose Parameter File button to choose UQ PyL UQ test functions params Sobol G txt file lt gt Click Choose Model File button to choose UQ PyL UQ test functions Sobol G py file 30 ES UQ PyL Uncertainty Quantification Python Laboratory pas Problem Definition Design of Experiment y Uncertainty Analysis Sensitivity Analysis Surrogate Modelling Optimi dar Load Model Information Choose Parameter File D UQ PyL UQ test functions params Sobol G txt Choose Model File D Ug PyL UQ test functions Sobol G py Choose DoE method Latin Hypercube Latin Hypercube Configuration Choose different Latin Hypercube method Number of Sample Points Generate DoE Script Choose Parameter File
50. ose surrogate modeling method and show results lt gt Choose surrogate modeling method SVM lt gt Click Show Results button to acquire surrogate modeling results UQ PyL gives the tabular and graphic results 64 This step can also implemented using python script Python script file SAC Surrogate py Optional turn off bytecode pyc files import sys sys dont write bytecode True from UQ DoE import monte carlo from UQ test functions import SAC from UQ util import scale samples general read param file discrepancy import numpy as np import random as rd Set random seed does not affect quasi random Sobol sampling seed 1 np random seed seed rd seed seed Read the parameter range file and generate samples param file UQ test functions params SAC txt pf read param file param file Generate samples choose method here param values monte carlo sample 500 pf num vars 65 Samples are given in range 0 1 by default Rescale them to your parameter bounds scale samples general param values pf bounds np savetxt Input SAC txt param values delimiterz Run the model and save the output in a text file This will happen offline for external models Y 2 SAC predict param values np savetxt Output SAC txt Y delimiter Perform regression analysis using the model output Specify which column of the output file to analyze zero indexed m
51. ou plan to manually specify the full path to Canopy Python you must specify Canopy s User Python rather than the Canopy installation Python Learn More L Choose Yes then click Start using Canopy 22 eoo Welcome to Canopy m CHTHOUONT Hi welcome to Canopy CANOPY Log in to your Enthought account or create one ne A E E y Editor Package Manager Doc Browser Training on Demand Recent files No recent files Restore previous session G Open an existing file F Version 1 5 5 3123 No updates found Also you can check your python installation in your python installation path All files are in YourPythonPath User for me IS Users wangchen Library Enthought Canopy_64bit User The python executable file is in YourPythonPath User bin Step 3 Test your Python installation If you have multiple python environment please specific one For MacOS you could add a line like this to the etc launchd conf file export PY THONPATH Users wangchen Library Enthought Canopy_64bit User bin 23 Then enter command source launchd conf to make your launchd conf file renew Type python or python2 7 command if you can see Enthought Canopy Python that means you already accomplished the installation ouchenmatoMacBook Pro UQ PyL Linux wangchen python Enthought Canopy Python 2 7 9 64 bit default Jun 38 2815 19 41 21 GCC 4 2 1 Based on Apple
52. r bashrc file renew Type python or python2 7 command if you can see Enthought Canopy Python that means you already accomplished the installation n M LE Enthought Canopy Python 2 7 9 64 bit default Jun 30 2015 22 40 22 GCC 4 1 2 20080704 Red Hat 4 1 2 55 on linux Iype help copyright credits or license for more information You can check if all the packages UQ PyL needed are already installed Using import command if no error messages that means you already have all the packages ba Enthought Canopy Python 2 7 95 d Dit default Jun 30 2015 22 40 22 GEC 4 1 2 20080704 4 1 2 55 on linux Ivpe help copyright redits HE license for more information gt gt gt import numpy gt gt gt numpy version _ gt gt gt import matplotlib gt gt gt matplotlib version 1 4 3 gt gt gt import aklearn gt gt gt gklearn version 0 16 1 import FyQt4 Step 4 Install UQ PyL software Download UQ PyL Linux version unzip the source code using command tar xvf UQ PyL_Linux tar gz Then enter into the UQ PyL directory cd UQ PyL Linux Enter command to run UQ PyL main page python main pyw or python2 7 main pyw 19 Or Interactive UQ PyL Software python main interactive pyw or python2 7 main interactive pyw You can see the main page of UQ PyL software Driver Generator 2 2 3 MacOS platform For MacOS platfo
53. r Opinia e 66 4 Run simulation On surrogate mode ete ape er a Sa at ep Eege 68 dak Wise merac ito VO PVL S ege 72 4 4 1 How to run interactive UQ PyL Software 72 4 4 2 How to use interactive UO ETA Be EE 73 1 Introduction 1 1 A Quick Start UQ PyL Uncertainty Quantification Python Laboratory is a software platform for performing various uncertainty quantification UQ activities such as Design of Experiments DoE Statistical Analysis Sensitivity Analysis SA Surrogate Modeling and Parameter Optimization This document describes how to set up problems and use these UQ methods to solve them through UQ PyL The mathematics of those UQ methods can be found in the separate theory manual We request that you cite the following paper when you report the results obtained by using the UQ PyL software platform C Wang O Duan Charles H Tong 2015 UQ PyL A GUI platform for uncertainty quantification of complex models Under review for Environmental Modeling Software 1 2 Available UQ PyL Capabilities 1 2 1 Design of Experiment Full Factorial design Fractional Factorial design Plackett Burman design Box Behnken design Central Composite design Monte Carlo design Latin Hypercube design random center maxmin center maxmin correlate Symmetric Latin Hypercube design Improved Distributed Hypercube design Sobol sequence Halton sequence Faure sequence Hammersley sequence Good Lattice Point 1 2 2 Statisti
54. rm Canopy also has a MacOS version You can download Canopy software and UQ PyL MacOS version from our website The installation process is very similar with Linux platform Step 1 Install Canopy software First double click the dmg file to start the installation 20 Drag Canopy into your Applications folder to install Canopy Applications ZENTHOUGHT Pull Canopy icon to Application folder Canopy Step 2 Setting up Canopy environment Double click Canopy icon to start setting Canopy environment 21 eoo Canopy Environment Setup Canopy System and User environment locations Your Canopy environment will be installed in the location shown below You may change it if you wish to What s this Canopy environment directory Change Users wangchen Library Enthought Canopy_64bit Write Canopy environment directory click Continue to continue Your python installation will be in this directory eoo Canopy Setting up your Canopy environment After that a dialogue will display eoo Make Canopy your default Python environment Do you want to make Canopy your default Python environment Yes Recommended This will give you direct access to Canopy Python and to utilities like IPython easy install nosetests from your terminal command prompt Learn More No Later on if you want to make Canopy Python the default you can do so from the preferences dialog Warning If y
55. sample output sobol 2015 10 11 17 54 55 txt 2 Click Choose Output File button to load model output file for example UQ PyL model output sobol 2015 10 11 17 54 55 txt 44 i5 UQ PyL Uncertainty Quantification Python Laboratory Problem Definition Design of Experiment Statistical Analysis Sensitivity Analysis Surrogate Modelling Optimifak Perform Design of Experiment Load parameter file Load Model File Choose DoE method Number of Sample Points Generate DoE Script Execute DoE Script Choose Analysis Method Load parameter file D Ug PyL UQg test functions params Sobol G txt D UQ PyL UQ test functionz Sobol G py Quasi Monte Carlo Y 4 500 D Ug PyL Ug test functions params Sobol G txt Load data file input file output file D UQ PyL sample output sobol 2015 10 11 20 28 40 txt Surrogate Model Method Choose Parameter File Choose Model File Choose Parameter File Choose Input File Choose Output File Show Results Choose Surrogate Modeling method and show results Step 3 Choose surrogate modeling method and show results lt gt Choose surrogate modeling method like SVM lt gt Click Show Results button to show sensitivity analysis results UQ PyL gives the tabular and graphic results D lt gt UG PyL gt python E m UQ R model m sum I D UQ PyL sample_output_sobol_2615_1 H 11 ZH 28 4BH txt Y D lQ PuL model output sobol 2H815 1H
56. spyder to achieve this function The left part of the interface is a code editor you can type your python code here After run the python code you can see internal variable values in the upper right of the interface and output results 1n the lower right part 4 4 2 How to use interactive UQ PyL Software Method One You can write your own python code in the editor part then click Run gt display on the upper right part and lower right part of the interface button to run the python script Variable values and output values will be Method Two Also you can click Open button D to load a exist python script gt file for example AUQ PyL python example py then click Run button to run the python script You can see the variable values below Key Type Size Value Y float 4 90 array 6 92016797 0 93247791 1 17311737 6 65173187 0 64312815 param file str 1 UQ test functions params Sobol G txt param values float64 90 8 array 0 66666667 6 a 0 33333333 0 66666667 1 fed i num vars 8 names xl x2 x3 x4 x5 x6 x x8 amp boun pf dit 3 seed int 1 1 And tabular and graphic outputs 73 Parameter Mu Sigma Mu Star Mu Star Conf xl 0 706156 2 641627 2 640762 0 445780 xz 0 127724 1 719118 1 542336 0 482545 x3 0 039390 0 588605 0 542817 0 148633 x4 0 118547 0 313918 0 295576 0 098500 xo 0 001200 0 025919 0 024397 0 005641 Sp 0 002367 0 039866 0 0
57. te day 12 Lower zone primary free water lateral drainage rate 0 001 0 05 day directly to lower zone free water decimal fraction 14 0 30 dimensionless Laud to lower zone tension water decimal fraction Table 6 Parameters of SAC SMA model So we generate the parameter file UQ PyL UQ test_functions params SAC txt as UZTWM 10 300 UZFWM 5 150 DAR ist e Ta PETIM 0 Qu ll ADIMPS Q 02 APERC 5 3900 REXP A LZTWM 10 500 LZFSM 5 400 LZFPM 10 1000 BASE OS 05959 LABR Us 001 005 PEREEB O 0 9 SAC SMA model is an executable file on Windows or Linux or MacOS system In order to using UQ PyL we need to generate a python driver to couple SAC SMA model and UQ PyL platform The driver file can be generated automatically by UQ PyL s GUI 53 2 UQ PyL Uncertainty Quantification Python Laboratory Problem Definition Design of Experiment Uncertainty Analysis Sensitivity nalysis Surrogate Modelling Optimi tad Generate Template File po Load Model Input File D UQ PyL Ug test functions SAC ps test l sac Choose Model Input File gt em US Generate Template File Input Variables A EH Generate Template file A Load Parameter File D UQ PyL UQ test functions params SAC txt Choose Parameter File Load Model Input File D UQ PyL UQ test functions SAC ps test l sac Choose Model Input File ee Load Executable File D UQ PyL UQ test_functi ons SAC mopexcal exe C
58. there is a software integrate Python and some common packages called Python xy It contains all the packages UQ PyL needed You can just install Python xy and UQ PyL to run UQ analysis Step 1 Install Python xy software You can download Python xy from our website Double click the Installation file to start installation td Prthon z y 2 7 6 0 Setup License Agreement pythonix y Please review the license terms before installing Pythontx v eB Press Page Down to see khe rest of the agreement Copyright E 2008 Pierre Raybaut Licensed under the GNU General Public License version 3 Python components are distributed as they were received from their copyright holder under their own copyright and or license and without any linking with each other Pethor 5 4 software collection Le the cofec anof software libranes and documents is licensed under the terms of the GNU General Public License version 3 http Zw gnu arg licenses gpl tek GNU GENERAL PUBLIC LICENSE Version 3 29 June 007 IF you accept the terms of the agreement click I Agree to continue You must accept the agreement to install Pythonis y 2 7 6 0 Pythonis vy Ehe Python Distribution made by Scientists Far Scientists Click I Agree to continue Choose Users Y python x y Choose for which users you want to install Python x y 2 7 6 0 Select whether you want to install Python x y 2 7 6 0 for yourself only or for all users of
59. ut SAC txt Output SAC txt column 0 63 4 2 4 Surrogate Modeling 2 UQ PyL Uncertainty Quantification Python Laboratory E Problem Definition Design of Experiment Uncertainty Analysis Sensitivity Analysis urrogate Modelling Optimi Lah Perform Design of Experiment Load parameter file D UQ PyL UQ test_functions params SAC txt Choose Parameter File Load Model File D UQ PyL UQ test functions SAC py Choose Model File Choose DoE method Monte Carlo Number of Sample Points 200 F Stepi Load parameter file and driver file Generate DoE Script Execute DoE Script Step2 Load Design of Experiment results Choose Analysis Method Load parameter file D UQ PyL UQ test functions params SAC txt Choose Parameter File Load data file input file output file Choose Input File Choose Output File Surrogate Model Method SYM v Show Results Step3 Choose surrogate modeling method and show results Step 1 Define parameter and model information lt gt Choose Surrogate Modeling tab lt gt Load parameter file UQ PyL UQ test functions params SAC txt and model file driver file UQ PyL UQ test functions SAC py Step 2 Load DoE results for surrogate modeling lt Choose DoE results sample input file UQ PyL UQ test functions SAC sample output mc 2015 05 19 21 45 26 tx t and model output file UQ PyL UQ test functions SAC model output mc 2015 05 19 21 45 26 txt KK Step 3 Cho
60. ut file output file Choose Input File Choose utput File Basic Statistical Analysis Methods Statistical Moments Methods Y Show Results Advanced Statistical Analysis Methods Pearson Spearman Correlations Analysis Y Show Results Step 1 Define parameter and model information lt gt Switch to Statistical Analysis tab lt gt Click Choose Parameter File button to choose UQ PyL UQ test functions params Sobol G txt file lt Click Choose Model File button to choose UQ PyL UQ test functions Sobol G py file 34 2 UQ PyL Uncertainty Quantification Python Laboratory Problem Definition Design of Experiment ncertainty Analysis Sensitivity Analysis Surrogate Modelling Optimiigk Perform Design of Experiment Load parameter file D UQ PyL UQ test functions params Sobol G txt Choose Parameter File Load Model File D UQ PyL VQ test_functions Sobol_6 py Choose Model File Choose DoE method Monte Carlo y Nunber of Sample Points 50 T Generate DoE Script Execute DoE Script Load Design of Experiment results Choose Analysis Method Load parameter file D Ug PyL Ug test functions params Sobol G txt Choose Parameter File Load data file input file output file Choose Input File Choose Output File Basic Statistical Analysis Methods Statistical Moments Methods x Show Results Advanced Statistical Analysis Methods Pearson Spearman Correlations Analysis Y Show Results Step 2 Lo
61. xamples line py Extraction progress After unzip there will be two shortcut on the desktop one is refer to UQ PyL software main page the other is refer to interactive version of UQ PyL software Double click the shortcuts can start the UQ PyL software If the shortcut doesn t work 11 please go to your install path double click the main pyw file or main interactive pyw file to start these In UQ PyL main page you can do uncertainty quantification analysis through pull down menus In interactive version of UQ PyL software you can write python script to run uncertainty quantification analysis and can see output results and internal variables values through the software s interface Loading Uncertainty Quantification Python Laboratory Version 1 0 PyL UQ PyL Splash Page 12 File About Problem Definition Statistical Analysis Sensitivity Analysis Surrogate Modelling Optimiisk Add Input Variables Parameter Name Parameter Lower Bound 0 00 Input Variables Parameter Upper Bound 1 00 Parameter Distribution A Driver Generator Show input variables Parameter Name Parameter Lower Bound Parameter Upper Bound Parameter Distribution Save to Parameter File UQ PyL Software Main Page So Pythonexamplepy UQ PyL Interactive Environment AMES File Edit About D spetta 1h Optional turn off bytecode pyc files Key T Size Vel poe ype ue
62. y Analysis gt Surrogate Modelling Optimitah Perform Design of Experiment Load parameter file D UQ PyL UQ test functions params Sobol G txt Choose Parameter File Load Model File D UQ PyL UQ test functions Sobol G py Choose Model File Design of Experiment Method Choose DoE method Morris One at A Time Morries One At A Time MDAT Configuration Number of total sample points dimension 1 Number of Trajectories Number of Trajectories 50 Generate DoE Script Execute DoE Script Choose sensitivity analysis method and show results Choose Analysis Method Load parameter file D UQ PyL UQ test functions params Sobol G txt Choose Parameter File Load data file input file output file D UQ PyL sample output morris 2015 05 19 17 54 55 txt Choose Input File D Ug PvL model output morris 2015 05 19 17 54 55 txt Choose Output File Sensitivity Analysis Method Morris v Show Results LESE E l Step 3 Choose sensitivity analysis method and show results lt gt Choose sensitivity analysis method like Morris lt gt Click Show Results button to show sensitivity analysis results UQ PyL gives the tabular and graphic results EN CAWindowsisystem321cmd exe o Soho 1 G txt I D lQ PuL sample output morris 2H15 H5 19 17 54 55 txt Y D zU FyuL model output morris 2Hi15 H5 17 17 54 55 txt Parameter Mu Sigma Mu Star Mu Star Conf 4 248158 2 559118 2 864968 0 422817 HBH 29H318 1 61

UQ-PyL User Manual version 1.1

Contents

Download Pdf Manuals

Related Search

Related Contents