1"use strict";(self.webpackChunk=self.webpackChunk||[]).push([[2803],{1295(e,A,n){n.d(A,{z:()=>o});var t=n(96540),i=n(53259),a=n.n(i),r=(n(78478),n(74848)),l=a()({loader:function(){return n.e(1236).then(n.bind(n,91236))},loading:function(e){return e.timedOut?(0,r.jsx)("blockquote",{children:"Error: Loading Plotly timed out."}):(0,r.jsx)("div",{children:"loading..."})},timeout:1e4}),o=t.memo(function(e){var A=e.data;return(0,r.jsx)("div",{className:"plotly-figure",style:{"overflow-x":"auto"},children:(0,r.jsx)(l,{data:A.data,layout:A.layout})})})},7877(e,A,n){n.d(A,{A:()=>a});n(96540);var t=n(29068),i=n(74848);const a=function(e){return(0,i.jsx)(t.A,{language:"python",title:"Output:",children:e.children})}},28453(e,A,n){n.d(A,{R:()=>r,x:()=>l});var t=n(96540);const i={},a=t.createContext(i);function r(e){const A=t.useContext(a);return t.useMemo(function(){return"function"==typeof e?e(A):{...A,...e}},[A,e])}function l(e){let A;return A=e.disableParentContext?"function"==typeof e.components?e.components(i):e.components||i:r(e.components),t.createElement(a.Provider,{value:A},e.children)}},38987(e,A,n){n.d(A,{A:()=>r});n(96540);var t=n(28774),i=n(35088),a=n(74848);const r=function(e){var A=e.githubUrl,n=e.colabUrl;return(0,a.jsxs)("div",{className:"margin-top--sm margin-bottom--lg",children:[(0,a.jsxs)(t.A,{to:A,className:"button button--outline button--primary margin-right--xs",children:["Open in GitHub",(0,a.jsx)(i.A,{})]}),(0,a.jsxs)(t.A,{to:n,className:"button button--outline button--primary margin--xs",children:["Run in Google Colab",(0,a.jsx)(i.A,{})]})]})}},48284(e,A,n){n.r(A),n.d(A,{assets:()=>p,contentTitle:()=>s,default:()=>h,frontMatter:()=>o,metadata:()=>t,toc:()=>d});const t=JSON.parse('{"id":"tutorials/bonsai/index","title":"Simple and interpretable arms with BONSAI.","description":"<LinkButtons","source":"@site/../docs/tutorials/bonsai/index.mdx","sourceDirName":"tutorials/bonsai","slug":"/tutorials/bonsai/","permalink":"/docs/next/tutorials/bonsai/","draft":false,"unlisted":false,"tags":[],"version":"current","lastUpdatedBy":null,"lastUpdatedAt":null,"frontMatter":{"title":"Simple and interpretable arms with BONSAI.","sidebar_label":"Simple and interpretable arms with BONSAI."},"sidebar":"tutorials","previous":{"title":"Utilizing and Creating Ax Analyses","permalink":"/docs/next/tutorials/analyses/"},"next":{"title":"Manually Sampling Specific Parameterizations","permalink":"/docs/next/recipes/attach-trial"}}');var i=n(74848),a=n(28453),r=n(38987),l=n(7877);n(1295);const o={title:"Simple and interpretable arms with BONSAI.",sidebar_label:"Simple and interpretable arms with BONSAI."},s="BONSAI + MAP-SAAS Tutorial:",p={},d=[{value:"Overview",id:"overview",level:2},{value:"1. Imports",id:"1-imports",level:2},{value:"2. Understanding the Components",id:"2-understanding-the-components",level:2},{value:"2.1 Hartmann50 Problem",id:"21-hartmann50-problem",level:3},{value:"2.2 A MAP-SAAS model (<code>EnsembleMapSaasSingleTaskGP</code>)",id:"22-a-map-saas-model-ensemblemapsaassingletaskgp",level:3},{value:"2.3 BONSAI",id:"23-bonsai",level:3},{value:"3. Set Up the Hartmann50 Optimization Problem",id:"3-set-up-the-hartmann50-optimization-problem",level:2},{value:"4. Define the Hartmann50 Objective Function",id:"4-define-the-hartmann50-objective-function",level:2},{value:"5. Configure the Generation Strategy with BONSAI and MAP-SAAS",id:"5-configure-the-generation-strategy-with-bonsai-and-map-saas",level:2},{value:"6. Run the Optimization Loop",id:"6-run-the-optimization-loop",level:2},{value:"7. Visualize Optimization Performance",id:"7-visualize-optimization-performance",level:2},{value:"8. Analyze Objective vs Simplicity trade-offs",id:"8-analyze-objective-vs-simplicity-trade-offs",level:2},{value:"10. Key Takeaways",id:"10-key-takeaways",level:2},{value:"Why BONSAI + MAP-SAAS?",id:"why-bonsai--map-saas",level:3},{value:"When to Use This Approach",id:"when-to-use-this-approach",level:3}];function c(e){const A={a:"a",code:"code",h1:"h1",h2:"h2",h3:"h3",header:"header",img:"img",li:"li",ol:"ol",p:"p",pre:"pre",strong:"strong",ul:"ul",...(0,a.R)(),...e.components};return(0,i.jsxs)(i.Fragment,{children:[(0,i.jsx)(r.A,{githubUrl:"https://github.com/facebook/ax/blob/main/tutorials/bonsai/bonsai.ipynb",colabUrl:"https://colab.research.google.com/github/facebook/ax/blob/main/tutorials/bonsai/bonsai.ipynb"}),"\n",(0,i.jsx)(A.header,{children:(0,i.jsx)(A.h1,{id:"bonsai--map-saas-tutorial",children:"BONSAI + MAP-SAAS Tutorial:"})}),"\n",(0,i.jsxs)(A.p,{children:["This tutorial demonstrates how to use ",(0,i.jsx)(A.strong,{children:"BONSAI"})," (Bayesian Optimization with Natural\nSimplicity and Interpretability) with ",(0,i.jsx)(A.strong,{children:"MAP-SAAS"})," to optimize the high-dimensional\nHartmann50 benchmark problem and simplify proposals in order to only make necessary\nchagnes from the default (status quo) parameter values."]}),"\n",(0,i.jsx)(A.h2,{id:"overview",children:"Overview"}),"\n",(0,i.jsxs)(A.ul,{children:["\n",(0,i.jsxs)(A.li,{children:[(0,i.jsx)(A.strong,{children:"Hartmann50"}),': A 50-dimensional synthetic benchmark where only 6 dimensions are\nrelevant (the true Hartmann function), and 44 dimensions are "dummy" irrelevant\ndimensions.']}),"\n",(0,i.jsxs)(A.li,{children:[(0,i.jsx)(A.strong,{children:"BONSAI"}),": A Bayesian optimization method that removes irrelevant parameter changes\nto simplify proposals from Ax. This simplifies the proposals so that they change fewer\nparameters, making the proposals more interpretable and more likely to avoid\nregressions in metrics not captured in the optimization objective."]}),"\n",(0,i.jsxs)(A.li,{children:[(0,i.jsx)(A.strong,{children:"MAP-SAAS"}),": A fast Gaussian process model that has a SAAS (sparsity) prior."]}),"\n"]}),"\n",(0,i.jsx)(A.p,{children:"This combination is particularly powerful for high-dimensional problems with low\neffective dimensionality."}),"\n",(0,i.jsxs)(A.p,{children:["These methods were proposed in\n",(0,i.jsx)(A.strong,{children:(0,i.jsx)(A.a,{href:"https://arxiv.org/abs/2602.07144",children:"Daulton, et al. BONSAI: Bayesian Optimization with Natural Simplicity and Interpretability, ArXiv, 2026"})}),"."]}),"\n",(0,i.jsx)(A.h2,{id:"1-imports",children:"1. Imports"}),"\n",(0,i.jsx)(A.pre,{children:(0,i.jsx)(A.code,{className:"language-python",children:'import numpy as np\nimport torch\n\nfrom ax.api.client import Client\nfrom ax.api.configs import RangeParameterConfig\nfrom ax.api.utils.generation_strategy_dis
1patch import choose_generation_strategy\nfrom ax.api.utils.structs import GenerationStrategyDispatchStruct\n\n# Model configuration\nfrom ax.generators.torch.botorch_modular.surrogate import ModelConfig\n\n# BoTorch model (the key component for BONSAI)\nfrom botorch.models.map_saas import EnsembleMapSaasSingleTaskGP\n\nprint(f"Using torch version: {torch.__version__}")\nprint(f"CUDA available: {torch.cuda.is_available()}")\n'})}),"\n",(0,i.jsx)(l.A,{children:"[INFO 09-24 05:13:04] ax.storage.sqa_store.with_db_settings_base: Ax SQL storage initialized with SQLAlchemy 1.4.17\nUsing torch version: 2.14.0+cu130\nCUDA available: False"}),"\n",(0,i.jsx)(A.h2,{id:"2-understanding-the-components",children:"2. Understanding the Components"}),"\n",(0,i.jsx)(A.h3,{id:"21-hartmann50-problem",children:"2.1 Hartmann50 Problem"}),"\n",(0,i.jsx)(A.p,{children:"The Hartmann50 problem is a 50-dimensional optimization problem where:"}),"\n",(0,i.jsxs)(A.ul,{children:["\n",(0,i.jsx)(A.li,{children:"The first 6 dimensions contain the actual Hartmann function (which has 6 local minima)"}),"\n",(0,i.jsx)(A.li,{children:'The remaining 44 dimensions are "dummy" and do not affect the objective value'}),"\n",(0,i.jsx)(A.li,{children:"This makes it an ideal test case for algorithms that can identify and focus on\nrelevant dimensions"}),"\n",(0,i.jsx)(A.li,{children:"The global minimum is approximately -3.32237"}),"\n"]}),"\n",(0,i.jsxs)(A.h3,{id:"22-a-map-saas-model-ensemblemapsaassingletaskgp",children:["2.2 A MAP-SAAS model (",(0,i.jsx)(A.code,{children:"EnsembleMapSaasSingleTaskGP"}),")"]}),"\n",(0,i.jsx)(A.p,{children:"This is a Gaussian Process model that uses an ensemble of independent GPs with different\nsamples of the global sparsity level (integrating over the global sparsity level). It\nuses Maximum A Posteriori (MAP) estimation for fitting each member in the ensemble,\nwhich is significantly faster than using MCMC as in SAASBO. It levers the same sparsity\nprior as SAASBO, but is significantly faster."}),"\n",(0,i.jsx)(A.h3,{id:"23-bonsai",children:"2.3 BONSAI"}),"\n",(0,i.jsxs)(A.p,{children:["BONSAI (Bayesian Optimization with Natural Simplicity and Interpretability) is a\ntechnique for post-processing candidates generated by BO to prune irrelevant parameter\nchanges from the default (status quo or target) values. It is compatabile with any\nacquisition function and is easily enabled by specifying\n",(0,i.jsx)(A.code,{children:"simplify_parameter_changes=True"})," in Ax."]}),"\n",(0,i.jsx)(A.h2,{id:"3-set-up-the-hartmann50-optimization-problem",children:"3. Set Up the Hartmann50 Optimization Problem"}),"\n",(0,i.jsx)(A.p,{children:"We'll create a Client and configure the experiment with 50 parameters."}),"\n",(0,i.jsx)(A.pre,{children:(0,i.jsx)(A.code,{className:"language-python",children:'# Create a client\nclient = Client()\n\n# Define 50 parameters (x0 through x49) in the unit hypercube [0, 1]\nparameters = [\n RangeParameterConfig(\n name=f"x{i}",\n parameter_type="float",\n bounds=(0.0, 1.0),\n )\n for i in range(50)\n]\n\n# Configure the experiment\nclient.configure_experiment(parameters=parameters)\n\n# Define the center of the search space as the pruning target\n# Parameters that are "pruned" will be set to these default values\npruning_target = {f"x{i}": 0.5 for i in range(50)}\n\n# Configure optimization to minimize the objective\nmetric_name = "hartmann"\nobjective = f"-{metric_name}" # Negative sign indicates minimization\nclient.configure_optimization(\n objective=objective,\n pruning_target_parameterization=pruning_target,\n)\n\nprint(f"Experiment configured with {len(parameters)} parameters")\nprint(f"Objective: minimize {metric_name}")\nprint(f"Pruning target: center of search space (0.5 for all parameters)")\n'})}),"\n",(0,i.jsx)(l.A,{children:"Experiment configured with 50 parameters\nObjective: minimize hartmann\nPruning target: center of search space (0.5 for all parameters)"}),"\n",(0,i.jsx)(A.h2,{id:"4-define-the-hartmann50-objective-function",children:"4. Define the Hartmann50 Objective Function"}),"\n",(0,i.jsx)(A.p,{children:"The Hartmann50 function uses the 6D Hartmann function on the first 6 dimensions, with 44\ndummy dimensions that don't affect the output."}),"\n",(0,i.jsx)(A.pre,{children:(0,i.jsx)(A.code,{className:"language-python",children:'from botor
1ch.test_functions import Hartmann\n\n# Create the 6D Hartmann function\nhartmann_6d: Hartmann = Hartmann(dim=6, negate=False)\n\n\ndef hartmann50(**parameters) -> float:\n """Evaluate the Hartmann50 function.\n\n Only the first 6 dimensions (x0-x5) affect the output.\n The remaining 44 dimensions (x6-x49) are ignored.\n\n Args:\n **parameters: Dict of parameter values (x0 through x49)\n\n Returns:\n The Hartmann function value (to be minimized).\n """\n # Extract the first 6 parameters that actually matter\n x = torch.tensor([[parameters[f"x{i}"] for i in range(6)]], dtype=torch.double)\n return hartmann_6d(x).item()\n\n\n# Test the function\ntest_params = {f"x{i}": 0.5 for i in range(50)}\nprint(f"Test evaluation at center: {hartmann50(**test_params):.4f}")\nprint("Global optimum is approximately: -3.32237")\n'})}),"\n",(0,i.jsx)(l.A,{children:"Test evaluation at center: -0.5053\nGlobal optimum is approximately: -3.32237"}),"\n",(0,i.jsx)(A.h2,{id:"5-configure-the-generation-strategy-with-bonsai-and-map-saas",children:"5. Configure the Generation Strategy with BONSAI and MAP-SAAS"}),"\n",(0,i.jsxs)(A.p,{children:["We use ",(0,i.jsx)(A.code,{children:"choose_generation_strategy"})," with\n",(0,i.jsx)(A.code,{children:'GenerationStrategyDispatchStruct(method="custom", simplify_parameter_changes=True)'})," to\nspecify that we want to use BONSAI and we specify to use\n`EnsembleMapSaasSingleTaskGP`` to leverage MAP-SAAS."]}),"\n",(0,i.jsx)(A.pre,{children:(0,i.jsx)(A.code,{className:"language-python",children:'# Configuration parameters\nNUM_SOBOL_TRIALS = 10 # Number of initial quasi-random trials\n\n# Configure the model for BONSAI with MAP-SAAS\nmodel_config = ModelConfig(\n botorch_model_class=EnsembleMapSaasSingleTaskGP,\n name="BONSAI",\n)\n\n# Create the BONSAI generation strategy using choose_generation_strategy\ngeneration_strategy = choose_generation_strategy(\n struct=GenerationStrategyDispatchStruct(\n method="custom",\n initialization_budget=NUM_SOBOL_TRIALS,\n initialize_with_center=True,\n simplify_parameter_changes=True,\n ),\n model_config=model_config,\n)\n\n# Set the generation strategy on the client\nclient.set_generation_strategy(generation_strategy=generation_strategy)\n\nprint(f"Generation strategy configured: {generation_strategy.name}")\nprint(" - 1 Center trial")\nprint(f" - {NUM_SOBOL_TRIALS - 1} Sobol trials")\nprint(" - BONSAI with MAP-SAAS")\nprint(" - simplify_parameter_changes=True (pruning irrelevant dimensions)")\n'})}),"\n",(0,i.jsx)(l.A,{children:"Generation strategy configured: Center+Sobol+MBM:BONSAI\n- 1 Center trial\n- 9 Sobol trials\n- BONSAI with MAP-SAAS\n- simplify_parameter_changes=True (pruning irrelevant dimensions)"}),"\n",(0,i.jsx)(A.h2,{id:"6-run-the-optimization-loop",children:"6. Run the Optimization Loop"}),"\n",(0,i.jsx)(A.pre,{children:(0,i.jsx)(A.code,{className:"language-python",children:'import logging\n\n# Set the Ax logger to show only warnings and errors.\nlogging.getLogger("ax.api.client").setLevel(logging.WARNING)\n\n# Total number of trials\nTOTAL_TRIALS = 50\n\n# Track best values for visualization\nbest_values = []\nall_values = []\ncurrent_best = float("inf")\n\nprint(f"Starting optimization with {TOTAL_TRIALS} trials...")\nprint("-" * 60)\n\nfor trial_idx in range(TOTAL_TRIALS):\n # Get the next trial(s) from the generation strategy\n trials = client.get_next_trials(max_trials=1)\n\n for index, parameters in trials.items():\n # Evaluate the objective function\n result = hartmann50(**parameters)\n all_values.append(result)\n\n # Update best value (we\'re minimizing)\n if result < current_best:\n current_best = result\n improvement_marker = " *NEW BEST*"\n else:\n improvement_marker = ""\n\n best_values.append(current_best)\n\n # Report the result back to Ax\n client.complete_trial(\n trial_index=index,\n raw_data={metric_name: result},\n )\n\n # Determine which phase we\'re in\n if trial_idx == 0:\n phase = "Center"\n elif trial_idx < NUM_SOBOL_TRIALS:\n phase = "Sobol"\n else:\n phase = "BONSAI"\n\n # Print progress (every 5 trials or when there\'s improvement)\n if trial_idx % 5 == 0 or improvement_marker:\n print(\n f"Trial {trial_idx + 1:3d}/{TOTAL_TRIALS} [{phase:6s}]: "\n f"value = {result:8.4f}, best = {current_best:8.4f}{improvement_marker}"\n )\n\nprint("-" * 60)\nprint("Optimization complete!")\nprint(f"Best value found: {current_best:.4f}")\nprint("Global optimum: -3.32237")\nprint(f"Gap to optimum: {current_best - (-3.32237):.4f}")\n'})}),"\n",(0,i.jsx)(l.A,{children:"Starting optimization with 50 trials...\n------------------------------------------------------------\nTrial 1/50 [Center]: value = -0.5053, best = -0.5053 *NEW BEST*\nTrial 6/50 [Sobol ]: value = -0.0376, best = -0.5053\nTrial 7/50 [Sobol ]: value = -0.5351, best = -0.5351 *NEW BEST*\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 11/50 [BONSAI]: value = -0.5053, best = -0.5351\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 13/50 [BONSAI]: value = -0.7964, best = -0.7964 *NEW BEST*\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 15/50 [BONSAI]: value = -0.8017, best = -0.8017 *NEW BEST*\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 16/50 [BONSAI]: value = -0.9318, best = -0.9318 *NEW BEST*\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 21/50 [BONSAI]: value = -0.9225, best = -0.9318\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 24/50 [BONSAI]: value = -0.9488, best = -0.9488 *NEW BEST*\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 26/50 [BONSAI]: value = -0.7886, best = -0.9488\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 27/50 [BONSAI]: value = -0.9664, best = -0.9664 *NEW BEST*\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 28/50 [BONSAI]: value = -0.9686, best = -0.9686 *NEW BEST*\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 31/50 [BONSAI]: value = -0.9667, best = -0.9686\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 33/50 [BONSAI]: value = -1.1020, best = -1.1020 *NEW BEST*\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 36/50 [BONSAI]: value = -1.2304, best = -1.2304 *NEW BEST*\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 39/50 [BONSAI]: value = -1.2711, best = -1.2711 *NEW BEST*\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 41/50 [BONSAI]: value = -1.3695, best = -1.3695 *NEW BEST*\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 45/50 [BONSAI]: value = -1.9519, best = -1.9519 *NEW BEST*\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 46/50 [BONSAI]: value = -1.9041, best = -1.9519\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\n/opt/hostedtoolcache/Python/3.14.7/x64/lib/python3.14/site-packages/linear_operator/utils/cholesky.py:41: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\nwarnings.warn(\nTrial 50/50 [BONSAI]: value = -2.0499, best = -2.0499 *NEW BEST*\n------------------------------------------------------------\nOptimization complete!\n
1Best value found: -2.0499\nGlobal optimum: -3.32237\nGap to optimum: 1.2725"}),"\n",(0,i.jsx)(A.h2,{id:"7-visualize-optimization-performance",children:"7. Visualize Optimization Performance"}),"\n",(0,i.jsx)(A.pre,{children:(0,i.jsx)(A.code,{className:"language-python",children:"import matplotlib.pyplot as plt\n\nfig, ax = plt.subplots(figsize=(12, 6))\n\ntrials_range = range(1, len(best_values) + 1)\n\n# Plot best values over trials (convergence plot)\nax.plot(trials_range, best_values, 'b-', linewidth=2, label='Best value found')\nax.axvline(x=1, color='purple', linestyle='--', alpha=0.5, label='Center')\nax.axvline(x=NUM_SOBOL_TRIALS, color='r', linestyle='--', alpha=0.7, label='Sobol \u2192 BONSAI')\nax.axhline(y=-3.32237, color='g', linestyle=':', alpha=0.7, label='Global optimum (-3.32)')\n\n# Scatter plot of all trial values\ncolors = ['purple'] + ['orange'] * (NUM_SOBOL_TRIALS - 1) + ['blue'] * (TOTAL_TRIALS - NUM_SOBOL_TRIALS)\nax.scatter(trials_range, all_values, c=colors, alpha=0.6, s=50)\n\nax.set_xlabel('Trial', fontsize=12)\nax.set_ylabel('Objective Value', fontsize=12)\nax.set_title('BONSAI Optimization Progress on Hartmann50', fontsize=14)\nax.legend(loc='upper right')\nax.grid(True, alpha=0.3)\nax.set_xlim([1, TOTAL_TRIALS])\n\nplt.tight_layout()\nplt.show()\n"})}),"\n",(0,i.jsx)(A.p,{children:(0,i.jsx)(A.img,{src:"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAABKYAAAJOCAYAAACN2Q8zAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjExLjIsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvgI3uAAAAAAlwSFlzAAAPYQAAD2EBqD+naQAA3DFJREFUeJzs3Xd4FNXXwPHv7qb3kEJIIAFCCb0jLdKLIBaKImABAQUb/sQurw0rYkNFFAsoiKKIBQUUUHrvvacQQgIhvW523j+uu2TJJqSRTcL5PM8+ZGfuzNyd3QmZs+eeq9M0TUMIIYQQQgghhBBCiEqmt3cHhBBCCCGEEEIIIcT1SQJTQgghhBBCCCGEEMIuJDAlhBBCCCGEEEIIIexCAlNCCCGEEEIIIYQQwi4kMCWEEEIIIYQQQggh7EICU0IIIYQQQgghhBDCLiQwJYQQQgghhBBCCCHsQgJTQgghhBBCCCGEEMIuJDAlhBBCCCGEEEIIIezCwd4dEEIIUf3Fx8dz4sQJq2XOzs6EhoZSu3btEu3j/PnzxMbG4uTkRP369fH09Cyy7f79+0lJScHNzY327dsXWm80GtmyZQs+Pj60bNmy0Pr8/HyOHz9OWloawcHBBAcHo9PprtrHnTt3kpWVRdOmTQkICCjydRw/fpywsDDq1at31X1eKTMzk9OnT5OVlUWdOnUICQkp9T7K4uzZs5w+fZpWrVrh7e1d7fZf3foBlz8rZjqdDjc3Nxo2bGj3vomyO3XqFHFxcbRo0QJfX1+bbbZv305ubi7du3ev5N7VfFdeVwXp9Xq6detW7LYxMTH4+fnRoEGDa9VFIYQQV9KEEEKIcpozZ44G2HyEh4dr33//fZHbfvfdd1qrVq2stnFwcNAGDBigbd++3eY2PXv21ABNr9dr+/fvL7Q+MTFRA7S+fftaLc/Oztaefvppzdvb2+p4Xl5e2vjx47Xdu3cX2c+zZ89qBoNBA7TJkycX2e7zzz/XAO3VV18tso0thw8f1oYNG6Y5Oztb9a1Ro0bap59+qplMplLtz5bo6Ght/fr1WkpKSqF1M2fO1ABt7dq15T6OLdd6/wXZ83WWhvmzcuVDp9NpAwYM0I4fP27vLooyeOihhzRA++2334psExISohkMhmvaj8OHD2vr16/X8vPzr+lxqpqiritAc3Z2trlNTEyMNnDgQE2n01natmrVqsj/g4QQQlQsGconhBCiwoSFhdG9e3e6d+9Ox44dqVWrFidPnuTOO+9k2bJlhdo//PDD3HXXXezfvx9PT086dOhAq1atAFi1ahXdunXjhx9+KPJ4JpOJF154ocT9Gz16NG+99RYpKSkEBQXRoUMH6tevT0ZGBl9++SVff/11kdsuWLCA/Px83Nzc+O6778jJySnxca/m77//pmPHjixdupS8vDyaNWtGx44d8fX15cSJEzz44IOMGjWK/Pz8ch1n/vz5REZGsm/fvkLr6tatS/fu3a9Zps613n9B9nydZREaGmq5biIiItDpdKxatYrevXuTlpZm7+6JauqFF14gMjKSzMxMe3fFLiIiIizXVcHHlbKysujXrx8rV67E1dWVjh074u/vz/79++nfvz+nTp2yQ++FEOL6IoEpIYQQFebBBx9kw4YNbNiwge3btxMfH88TTzwBwNy5c63afvbZZ3z88cc4OTkxe/ZsLly4wI4dO9i3bx/nz5/ngQceIC8vj3vvvZdDhw7ZPF6PHj345Zdf2LZt21X7tn//fpYuXYq7uzu///47586dY8eOHZw+fZrk5GS+/vpr2rVrV+T28+fPp1atWjz//PMkJyfbDLSVxdmzZ7njjjvIyMhg+PDhREdHc+jQIbZv305iYiKLFi3C09OTH374gddff71CjmnLqFGj2LBhQ7HnoCrvv7r1o6CJEydarpvDhw9z5MgRwsLCiI2NLTYwK4Qo2quvvmq5rsyP1atXF2r3ySefcPToUTp27Eh0dDTbt28nLi6OMWPGkJyczIsvvmiH3gshxPVFAlNCCCGuGUdHR5588kkAzp07Z1mem5vLSy+9BMC8efN4+OGHcXJysqyvVasWn376KePGjSM7O5tXXnnF5v5fffVV9Ho9zz333FX7cvToUQDuvPNOhgwZYrXOw8ODe++9l3vvvdfmtps3b+bIkSPccccd3Hfffej1er766qurHrMk3n77bS5dusTAgQNZsmSJVU0pg8HAXXfdxc8//2zV1iwqKooNGzaQkZEBqPooO3bsIC4urtBxDhw4QHR0NKCCdOYbtf379wMqQLZhwwZSUlKK3f/OnTu5ePFiof2fOnWKHTt2WNpe6cr9p6enF7ppLPiwFWzMz8/n1KlTbNu2jZiYGJvHKcvrLCgqKopt27YVqpl2ZZuC58UcVC2qT6XVuHFj7r//fgBLrZwdO3awa9cuS5tTp06xadMm0tPTrbZNTExkx44dHDhwgLy8vKse68SJE+zcudNyPs6fP8+GDRtISkqytLl06RIbNmywXMOZmZns2bPHZkZabGws27Zt48iRI5hMpiKPq2kasbGx7Ny5k5iYGDRNK1e74phMJo4ePcq2bduIjY0tst3BgwfZsGGD5XlMTAzbt2/nwoULpT5meeTm5nLkyBF27NhBYmJike2u7O/Zs2fZsmULiYmJbN261XKdbt682XIdnD592ua2UVFR7N69u9DnKT8/n8OHD7Nnzx6MRuM16ffVzvO1fl++++47AD7++GP8/PwA9X/XJ598goeHB0uXLiU7O7tCjiWEEKII9h1JKIQQoiYw15h64403Cq3btm2bBmg333yzZdnq1as1QGvRokWx+42NjdX0er3m6uqq5eTkWJaba0wlJiZqY8eO1QDt77//tqy3VWNq8+bNGqANGTKk1K9v0qRJGqCtX79e0zRN69Onj6bX67XY2NhCbUtbYyo4OFgDtF27dhXbrm/fvhqgLVq0yLLs6aef1gBtw4YN2qhRo6zqowwcOFA7f/58oe2vfPTs2VPTNNu1l8z737hxozZ69GhNr9dbaoA999xzmqZp2pEjR7T27dtb9ufh4aF9++23hfp/5f63b99eZB0YQKtdu7Zl27Nnz2pTpkzRPD09rdo0adJEW7lypc3zVJrXqWmatnLlSi0iIsJqm9DQUG3x4sWFXov5vGzZskWbOHGipfYYoPXr109LSkoq9r00K+6zcu+992qA9sorr2iapmlhYWFa06ZNtT179mgtW7a0HG/btm2apmnasWPHtD59+lh9Bjw9PbUXX3zRZo2hnTt3Wu3HxcVFe+GFF7TPPvtMA7Sff/7Z0va3337TAG327Nnaa6+9pnl4eGiA1rlzZ6vzd2WtuKCgIG3BggWFjr106VKtSZMmVm0DAwO1N998s0ztivPZZ59ptWv
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
1a5uiINn066ZcuUcvxOqsPJoQQQgghhBCiTCQwVQY6fRUt+nZhixq+V1QdKJ0O3OqpgEfSLvC/oWT7da0DtTpC/GoVpLJV/8iYpf6t3atMXRf/0ekgeCAE9YHk/ZB7Sc1Q6NMSnHzs3bviGQzQqRPGhAT1sxBCCCGEEEIIcRUSmKpJMmMATc1oVxQHNzUsLzMWKGFgSqeDerdD0k41nMyjgaopZZafDeknwO8G9RDlp3eEWu3t3QshhBBCCCGEEOKaqqJj0qo2U77J3l0ogg5KMseiphU961tRfNtAsyfAyRtSDkHaSciIVjPxZUSpgFSLZ6Uo9/XMaITVq3Fav179LIQQQgghhBBCXIVkTJWBll+S6I8deDRUs+/lZ4OhiGlS8/6rE+XeoPT7r90LvJrC+X/hwmZV6Nw1GGr3Bv8uRR9TXB+MRnQffIBbbi4MGQJOxWTuCSGEEEIIIYQQSGCqZvHrpIJT6afAs0nhrChNU8P9fFqrDKiycK0D9UephxBCCCGEEEIIIUQ5yFC+msTgDI0mgaMXpB4FY8bldXnpkHoYnP1VG5289UIIIYQQQgghhLAviU7UNP5doOV08G4OWedUPaiUQ5CTAL5todVL4Nva3r0UQgghhBBCCFEBdu/eTcuWLcnNzS3Xfo4ePUrLli1JTk6umI4JUUISmKqJ/DpBx9nQ9i2IeEI92r0DHT6QoJQQQgghhBBCVBHHjx9n8uTJREZG0qdPH5577jni4uJKtY+MjAwOHjyIyVS+SbqysrI4ePAgxms0kVFcXBwtW7a0PNq3b88tt9zC999/b7P9smXLGDlyJB07duTGG2/k2WefJT4+3qrNL7/8QsuWLXnooYcKbd+2bVtWrFhheb5mzRpGjRrFDTfcwNChQ/nwww/Jysqyeexp06bRsmVLdu3aVeTr2L59e2leviiGBKZqKr0D+HWEereph28bGb4nhBBCCCGEEFXEiRMn6NSpExkZGcyYMYNXXnkFX19fbrzxRnt37ZrIzc3l4MGDPPHEEyxevJjPP/+cvn37MmbMGBYsWGDVdtKkSYwbN45u3brx6aef8sILL3D48GFatWrF3r17Le0uXbrEoUOH+PTTT9mxY4fVPg4cOGDJ/vrhhx8YOHAgbdq04YMPPuChhx4iKiqKsWPHFupnamoqc+bMISMjg88++6zI15GRkVFonSgbKX4uhBBCCCGEEEJUssWLF+Pu7s78+fPR/TdxVY8ePXj00Uet2kVHR/Paa6+xc+dOvLy8uOOOO3jggQcs25jt2LGDjz76iFOnTtGiRQtef/116tSpU+r9XGsNGjSgZcuWAHTo0IGlS5eyatUq7rnnHgC+++47Pv/8c/7++2/69u1r2a5///4MGjSIUaNGcfDgQfR6lXjh5OTErbfeylNPPcWaNWtsHvOzzz5j5MiRPPvss5ZlgwYNIicnp1DbRYsWERISwqxZsxg7dizvvvsubm5uFfb6RWGSQlMGegc5bUIU4uiI9tRTZDz0EDg62rs3QgghhBBCQHZ20Y8razKVt20p6XQ6MjIySEpKslru7Oxs+TkjI4MePXoQHx/PrFmzGDduHM899xwzZswotL8777yT2267jbfffpvo6Gj69+9Pfn5+qfdTFi+//DK//fZbqbeLi4vjyJEjtGjRwrLs888/p1u3blZBKVDn6/nnn+fIkSOsW7fOat1rr73Ghg0b+PPPP20eR6fTERcXV2i4Y8FzbTZv3jwmTJjAkCFD8Pb2ZsmSJaV+XaJ0JGOqDHT6yo0oC1EtGAzQowd5CQnqZyGEEEIIIext5Mii13XsCC++ePn52LFgI4MGgJYt4Y03Lj+//35ITbVuU8rAzLhx45g7dy5NmjRhyJAhdOnShb59+9K0aVNLm7lz55Kbm8sPP/xgCaLk5eXxyCOPMHXqVDw9PS1t3377bUaNGgVAmzZtqFevHj/++CN33nlnqfZTFpGRkYwaNYpp06bx1FNPXfV1u7u7k5ubS1RUFGPGjOHJJ5+0rN+/fz8ji3jf2rRpA8C+ffvo1auXZXmjRo2YOHEizzzzDAMHDrRkU5k9+eST3HbbbTRt2pSbbrqJG264gf79+xMYGGjVbu/evezbt48//vgDvV7P/fffzxdffMG9995bmtMhSklSf4QQQgghhBBCiEoWHBzMoUOHmDlzJkajkXfeeYeIiAiGDx9umWFv165dREZGWmX29O/fn8zMTI4dO2a1v969e1t+9vX1pW3btuzevbvU+ynKQw89ZFW8vODj0UcfRa/X8/TTT/Pee+8Vu5//+7//Y/Hixfzwww98/vnn/Pbbb8yePduyPjc3t8ihc66uruh0OpszEL744oucOnWKb7/9ttC6AQMGcOTIESZNmsTp06d59NFHqVu3Lm8UDDaisqVuvfVWS8Dq/vvvZ/PmzRw9evSq50eUnWRMlYFm0uzdBSGqnvx82LgRx0uXYPBg0EvcWwghhBBC2Flxw7Cu/HvVRkCjyLZffFH2PhXg4eHB+PHjGT9+PAC//vort99+O59//jkPPfQQGRkZhbKZzEGbK4tvu7q6FmpnblOa/RTl6aefZvLkyTbXpaamMnHiRBo0aMDo0aOL3U/BGlNt27bl3LlzvPDCCzz66KMYDAZCQ0M5deqUzW3PnDmDpmmEhYUVWhcYGMgTTzzB9OnTufPOOwutDw0N5cknn+TJJ58kPz+fF154geeee46hQ4fSsmVLsrOzWbhwIc7Ozpb+Aej1er744gvefvvtYl+XKDsJTJWByVi+aTiFqJHy8tC9/TbuubnQv7/UmRJCCCGEEPbn4mL/tqVwyy23ULduXU6ePAmoIWpXFvQ+cOAAAOHh4VbLDx06RPfu3QEwmUwcPnyYwYMHl3o/RQkNDS1y3R133EH79u2ZN2+ezbpNxfH19SUzM5OMjAy8vLy47bbbmDVrFnFxcQQHB1u1nTdvHu7u7vTr18/mvqZNm8acOXOsMrBsMRgMTJo0iTfffJNTp07RsmVLli5dipubGytWrLBqu2nTJv7v//6P119/HQcHCaFcC5LSIIQQQgghhBBCVLL58+ezaNEiq5nhfvrpJ2JiYujRowcAEyZMYN++fSxYsACAtLQ0nn/+eYYOHUpISIjV/qZPn05mZiYA7777LklJSZbspdLspywWLFjAN998U+qgVEpKCgsWLKBdu3Z4eXkBKrgUEhLC6NGjOXv2LACapvHdd9/x4Ycf8tZbb+Hr62tzfx4eHkyfPp3XX38dTbs80unFF19k3bp1luLnubm5fPLJJ7i4uNChQwfg8jC+K4cp3nPPPaSnp5epuLsoGQlMCSGEEEIIIYQQlaxLly6sWLECf39/GjZsSGBgIJMnT+att95i2LBhADRt2pQvv/ySxx9/nODgYGrXro2DgwNz584ttL927doRHBxMUFAQM2bMYP78+ZZaSaXZT1m4lCKDbNy4cbRs2ZKIiAjq1KmDXq9n8eLFlvXe3t6sX7+eoKAgGjduTOPGjQkMDOS5556zDHEszgMPPICfn5/VDHw9evTg5ZdfxsvLi8aNG+Pn58fy5cv56aefCAkJ4dSpU/zzzz/cdtttNl/bgAED+KKChm+KwnRawTCiKFZqaire3t4kJSbh6287QivEdSs7G23ECHJyc3Fatgx9EQULqwJNg8OH4Z9/YP9+9bxZM+jdW024IuWxRHVkMplISEggMDCw0Ew0QohrR649IezHfP15eXkRFRVFgwYNShUgqSpMJhPR0dG4uroSGBiITld4Fnij0ciZM2fw9PSkdu3aVusyMzM5deoULVq0wGg0Eh0dTUhIiM1zUdx+srOzOXHiBM2aNcNwDWbZzsvLsyoi7uDgQEhISLEzAmZmZnL27FlcXV2pW7duofXJycnExcXRvHlzq+Xx8fFcuHCBevXq4e3tbVmek5NDTEwM/v7++Pj4WJanpKQQExNT5GtPTEzkwoULNGvWzPI6GjRogLu7e2lOQbWSnZ3N6dOnbV5XycnJ+Pr6kpKSYsl0Kw8JTJWCBKaEKEY1CUzl56talUuXQloaeHiATgfp6eDmpuq2P/SQlMgS1Y/cHAthH3LtCWE/NSUwJURVVJmBKancJYS4rvzwAyxcCH5+EBqqglJmSUnw008qWDVhgv36KIQQQgghhBDXC/laRwhx3UhNhZ9/Bnd3CAiwDkoB1KoFvr7w+++QmGifPgohhBBCCCHE9UQCU2WgMxQe8yvEdc/BAe2xx8icOBGq6DSq27ZBfDzUqVN0m9q14cIF2Ly58volhBBCCCGEENerqnn3WMXpDRLPE6IQBwfo25fchIQqG5i6dEn9W1z39HqVSWVuK4QQQgghhBDi2pEIixDiuuHkBCaTmoWvKJqmHk5OldcvIYQQQgghhLheSWCqDDSTTGQoRC
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
1nXsmz3l6btdbXVsltW1xbTFsXMa1ZueTNN29XMsjT3FRFl3GCAc35/nAtKoMAwcmbg9Xw85uGZc86c+bB8D/D2u5S7uJDl0HKxQEizLCktTa70dOfWVAYAADhODEO6/35p3jwpKcnet3691K6dNGnS4fO2b5e+/NIOrn7+2R4aF8rq1bN7g91/v70C4GWXEUoBqFgB95hatWqVOnTooNjYWHXo0EFTpkwpOBYREaE333xTI0aMCEqRAAAAABAKOnWSfvrp8Cp5WVnSjTdKCxdKjRvbw/727LHPjYqyeyBdd5100UWOll3ppKVJe/dKfv/hoBBAeDIsK7CuDb169VK1atU0ceJEeTweGYah/EutWLFCN9xwg1auXBnUYp3m8/kUHx+vjF0ZSkhMcLocILRkZcnq00fZfr+iZsyQq1o1pysCqhTTNJWWlqbk5GS5XEwhCVQU2l7VlZ0t3XWX9Oabh/fFxkrt20sNG9pD5Q4elLZskTwe+9xLL3Wu3spi2TLp00+lH36Q/H5TjRunqUaNZPXs6dKFF9o92wAcX3v37lVCQoIyMzMVFxdX7usF/NPzq6++0rPPPiuPx1Pk2Mknn6w1a9aUqzAAAAAACFUej/TGG9K//23PzSRJ+/ZJX39t9+aR7JXtmjSxtydMOLwfgZk7114Rce5cuzdanTpSdLS9cuJTT0nvvMOMEkA4Cngon9/vV2xsbMFz44hoOjMzU5GRkeWrDAAAAABC3FVXSe+/bw/hy8qSDh2ye/Sce66UmGif43LZq93961/ShRc6W++xNGggnXZaaPY6WrdOevVVKSdHOuMMu0bDsCemP+00aedO6YMP7CCwY0enqwVQFgEHU02aNNHMmTM1YMAASYWDqZkzZ6p58+blrw4AAAAAQthff9m9dHr3tueZ2rbNfv7dd0XPXbVKevLJiq+xLJKT7fmw8h+nnx4aQdWXX0rp6VKzZsXXk5wsZWRIs2dLHTqERs0ASifgYGrgwIG66667lJWVpauvvlqSdODAAX388ccaPny4XnrppWDVGHq4yQFFGYbUsKHMQ4f4TQAAAFQZpmk/qle355D6/nvpl1+cripwaWnS1Kn2Q7KHyx0ZVJW2R9XOnXZQ98MPdk+yhg3t3mLnnCO5A/grdNEiKS7u2O+dlCStWGEHVLVrl/09ADgj4MnPTdPUgAEDNHny5IJ9ERERysvL00033aR33nknaEWGivzJz4M1wRdQ2TABLOAc2h/gDNoeNm6Uhgyx55OqVcvet327tHXr4fmOLEvatUtq2VI66yznaj2WvDw7UFu0yJ4r62hKE1QtWGAPu9uxw54Dyu22J4KPjLQnh3/gAalmzdLXZlmS12uvwFev3uH9hmEqKSlNu3Yly7Jc8vmkzEx73q8GDUp/fQBlE+zJzwPuMeVyuTRp0iTdfPPN+vTTT7Vjxw4lJibq8ssv18UXX1zuwgAAAAAg1DVqJLVpI82bJyUk2CFNvXqFA5T0dKluXemll6RTT3Ws1FLJzbVXvvvqK/vx96Bq505pyhT7IRUNqvbtk1580V618Iwz7Pm18u3fL/33v/b2448fnjS+JIZhv8/q1cc+78ABKSbG7lkFIHwEHEzl69y5szp37hyMWgAAAAAgrBiG1Lev9Ouv0h9/SCedZK/YJ9lD/Hbtknbvlq65RjrlFGdrLQ232w7a2rSRhg0re1BVvbodDp1+uuTzSfHxh3tU1aghnXCCPVH8ihXS2WeXvq5u3ezX5OYWPxTQsuzP87XXSkes0QUgDAQ8lK8qyh/Kl5GeoYTaCU6XA4SW7GxZ99yjQ4cOKXr8eLliYpyuCKhSGE4EOIO2h3zLl0svvyytX2+vHOdy2cFUQoLUs6c0cKAUFeV0leVXUlD1dzEx9txPR/aO8vnsoXZl6T3m90srV9rvVaOG/fl1uSy1aJGp2rXjtHatS7VqSc88Ex4BIBDOgj2UL+Bg6qabbirxnHfffTeQS4esgmBqV4YSEgmmgEKysmT16aNsv19RM2bIVa2a0xUBVQp/HAPOoO3hSFlZdm+g1avtICU5WerYUUpJcbqy4+fIoOo//7E//ry8iq0hLs5S27aGnnhCOu+8in1voCoKmTmmZs+eXei5ZVlKT0+XaZpKTk6WwapcAAAAAKqQ6OjDcy1VFUcO/bv4Yunuu+3hfLt3S9u22ROg5+Qc3xp8PkMLFkgtWtjzWjGUDwgvAQdTO3bsKLLP7/fr888/17Rp0/Taa6+VqzAAAAAAQPg49VS7d9ju3fYKhC1b2sMZDx06fE5urvTXX/ZKhuVdM+vHH03dfbepjRvdys2Vnn9e+vBDe/J1r7foaoEAQlO5Jz8/UlRUlK688kpFR0dr6NChmjhxYjAvDwAAAAAIUdWqST16SK+/bq+QV726PRdU9er2ccuS1q6VTj5ZuvJKe/6t8qhXT2rRIl0TJyZrzBiXsrOlrVvtCdD//W/pX/+STjut3B8WgOPsuAyE79Chg2bMmHE8Lg0AAAAACFFXXy117ixt2GA/Dh6059tKT5dWrZLi4qR77il/KJUvJkZ65BH72pdeenj//PnSmWdKDz1kh2QAQtdxCaYWL17M5I8AAAAAUMXExNhh0J13SnXrStu3Sxs3StnZUvfu0tNPS23bBv99Tz5ZmjVLmjFDOuEEe19OjjR6tD3v1IwZdo8tAKEn4KF8kydPLrJv3759WrlypSZNmqSrr766XIWFNMYqA0UZhpScLDMriwH9AAAAVVh0tNS3rz1cb/16u8dUUpJUv/7xfV/DkK64QurWzQ7AnnvODqc2bZJ695YuuUR65RU7xAIQOgzLCiw3Ptqqe9HR0brllls0ZswYxVay5RB8Pp/i4+ODtiQiUNmwZDbgHNof4AzaHuCcktrfn3/aPbfmzTu8z+ORhg+XHnzQ7t0VzvbtkxYvlr78Utq1y16N8Pzz7VUh69RxujpUZnv37lVCQkLQspGAe0ytXLmyyL74+HjVr19fERER5SoKAAAAAIDyaNJEmjNHmjZNuvdee2L07Gzp8celSZOkceOknj0Lv2bLFvthGFLjxnZPr1C0YYPdK2z1asnttiee37ZNWrZM+vhj++Nt187pKoHSCTiYat68eTDrAAAAAAAgqAxD8nrtYXxPPCGNHSvl5krr1kmXXWYP/XvpJXvI34cfSt9/L/l89usSEqQLLpCuu85eATBU+Hx2KPXbb3b4FhV1+JhpSn/9JT37rPTMM/ZxINTR3zgAeTl5TpcAhB6/X8b99yv20UftiQQAAACAEFGjhh3WLF9uD3XL95//SE2b2qsJfv65PdTvtNOkU06xj6emSqNG2b2RQsXChXZPqdNOKxxKSZLLZde+a5f02WfO1AeUVal7TI0aNarMF3/qqafK/JqwwGoOQFGmKa1Zowi/394GAAAAQkyzZtKCBXbvqPvvl3bskLKypJUrpbg4qXZtO9xxuex5mmrXtkOgN96QHnssNNb4mT9fioy0H8UxDLvur7+WbrlFio+v2PqAsip1MPX000+X+eKVNpgCAAAAAIQlw5D69bPnlxo8WJoyxd7v89m9pk46SWrVSsqfOrl6dTvMattWatjQsbIL/PGHlJcnZWQc/ZzcXGnvXum778q/GmLjxvYcVsDxUupgKsDF+wAAAAAACDnx8VL37nbQs327tHOnvX/9evvxd5deWrH1BUP37uW/RkKCtGJFaIRyqJyYYwoAAAAAUCXl5trzT/XqJV14oRQd7XRFoWfPHmnyZKerQGUW8Kp8+bZu3aq1a9cqJyenyLGuXbuW9/IAAAAAABwXSUmSZdmP006TTjxRWrVK2r//8Dm5ufbzNm3seaectn+/tHSpvZJgbGzRea+ysqTsbOnss8u3mmBOjjRxor09Z440YkTg1wKOJeBgas+ePerfv78+O8ZU/wz/AwAAAACEqg4d7PBlxw57LiaPxw50jvTXX1JKij0BusfjTJ1/9+WX0tix9jxSderYc0D5/fZwxIgI6dprpVtvLf9k7d98I61da//r89kTxAPBFvBQvpEjR+rPP//URx99JEmaNWuWXnnlFbVu3Vp9+vTRvHnzglYkgDARFyerRg2nqwAAAABKJSlJuvJKO3TZudPuOZXPNKXNm+1w57rrQieUkqQuXaSnn7bnkMoPpPbts0O1hx4KTiglST162P/m5toTwAPHg2EF2K2pYcOGmjx5si688EIZhlGod9Sdd96pM844Q0OGDAlaoaHA5/MpPj5emZmZiiMqBoowTVNpaWlKTk6Wy8UUdkBFov0BzqDtAc4JVvvLzZXefVf65BO7B1JUlB1Q5eTYwdVNN0lXXBGcoOd4SEuTMjOlmBipQYPg1vnZZ9Jll9nbt90mjR8fvGsjfO3du1cJCQlBy0YCDqbcbrd8Pp+qVaumiIgIHTx4UJ7/Rcg7d+5Ux44dtWbNmnIXGEoIpoBj45dzwDm0P8AZtD3AOcFufxs2SAsX2kP3XC7pjDOkCy6Q6tYtf63h6sABqVYtu1fWiSdK69aFbkCHihPsYCrgOaby8vJUrVo1SVJCQoLWrVunpk2bSpIiIyO1devWchcHAAAAAEBFOPFE+4HDqleXzj/fntNqwwZpzRqpSROnq0JlE5T/1jnvvPP0+OOPa//+/crOztbIkSPVuHHjYFw6JOXl5DldAhB6/H4ZDz2kGk8/bf+XCgAAAICw17374e3Zs52rA5VXwMFUnSPWyXzooYc0ffp0xcXFqXr16nrzzTf1yCOPBKXAkMRig0BRpin9+qvcf/xhbwMAAAAIe/kToEsEUzg+yjSU79ChQ4qJiZEk7dixo2B/hw4d9OOPPxas0Hf55ZerXbt2QSwTAAAAAABUtObNpfr1pW3bpK++krKypOhop6tCZVKmYKp+/fq6/vrrdeutt6ply5aFjrVo0UItWrQIanEAAAAAAMA5hmEP53vnHenQIWnRIqlbN6erQmVSpqF8Z555pl599VW1atVKbdq00Ztvvql9+/Ydr9oAAAAAAIDDjhzON2eOc3WgcipTMPX111/rjz/+0LBhw7R582YNHjxY9erV08CBA/Xtt98erxoBAAAAAIBDunaVXP9LD5hnCsFmWJYV0FTeubm5+vTTTzVhwgTNmTNHeXl5at68uQYNGqT+/furVq1awa61iBUrVui9996TaZoaO3Zsieffe++9RXp49erVS7169SrV+/l8PsXHxytjV4YSEhMCqhmotLKyZPXpo2y/X1EzZshVrZrTFQFVimmaSktLU3JyslyuoCy6C6AUaHuAc2h/FatdOym/P8qmTVLDhs7WA+fs3btXCQkJyszMVFxcXLmvF3Drdbvd6t27tz777DNt3LhRTz75pA4cOKB77rmnYC6q46lHjx7q37+/Vq1apUmTJpXqNZMmT
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",alt:""})}),"\n",(0,i.jsx)(A.h2,{id:"8-analyze-objective-vs-simplicity-trade-offs",children:"8. Analyze Objective vs Simplicity trade-offs"}),"\n",(0,i.jsx)(A.p,{children:"One of the key benefits of BONSAI is that it can prune irrelevant parameters (set them\nto default values). Let's analyze how the number of active parameters relates to the\nbest objective value found."}),"\n",(0,i.jsx)(A.pre,{children:(0,i.jsx)(A.code,{className:"language-python",children:'def count_active_parameters(\n parameters: dict[str, float], default_value: float = 0.5, tol: float = 1e-6\n) -> int:\n """Count the number of parameters that differ from the default value."""\n return sum(1 for v in parameters.values() if abs(v - default_value) >
1 tol)\n\n\n# Get all trials and their parameters\nexperiment = client._experiment\ntrials_data = []\n\nfor trial_index, trial in experiment.trials.items():\n arm = trial.arm\n if arm is not None:\n params = arm.parameters\n num_active = count_active_parameters(params)\n # Get the objective value for this trial\n trial_data = trial.lookup_data()\n if not trial_data.df.empty:\n obj_value = trial_data.df[trial_data.df["metric_name"] == metric_name][\n "mean"\n ].values[0]\n trials_data.append(\n {\n "trial_index": trial_index,\n "num_active_params": num_active,\n "objective_value": obj_value,\n }\n )\n\n# Convert to arrays for plotting\nnum_active_params = [d["num_active_params"] for d in trials_data]\n\nprint(f"Collected data for {len(trials_data)} trials")\nprint("\\nNumber of active parameters per trial:")\nprint(f" Min: {min(num_active_params)}")\nprint(f" Max: {max(num_active_params)}")\nprint(f" Mean: {np.mean(num_active_params):.1f}")\n\n# Compute best observed objective value for each number of active parameters\nfrom collections import defaultdict\n\n# Group trials by number of active parameters\nparams_to_best_value = defaultdict(lambda: float(\'inf\'))\nfor d in trials_data:\n n_active = d["num_active_params"]\n obj_val = d["objective_value"]\n if obj_val < params_to_best_value[n_active]:\n params_to_best_value[n_active] = obj_val\n\n# Sort by number of active parameters\nsorted_n_active = sorted(params_to_best_value.keys())\n\nfig, ax = plt.subplots(figsize=(8, 5))\n\n# Modify best_values_by_n_active to represent the best objective for any point with <=k parameters active\ncumulative_best_values = []\ncurrent_best = float(\'inf\')\nfor n in sorted_n_active:\n current_best = min(current_best, params_to_best_value[n])\n cumulative_best_values.append(current_best)\n\n# Line plot: Best observed value versus number of active parameters based on cumulative best values\nax.plot(sorted_n_active, cumulative_best_values, color=\'steelblue\', marker=\'o\', linestyle=\'-\', linewidth=2)\nax.axhline(y=-3.32237, color=\'g\', linestyle=\':\', linewidth=2, label=\'Global optimum (-3.32)\')\nax.axvline(x=6, color=\'r\', linestyle=\'--\', alpha=0.7, label=\'True relevant dims (6)\')\nax.set_xlabel(\'Number of Active Parameters (<=k)\', fontsize=12)\nax.set_ylabel(\'Best Objective Value\', fontsize=12)\nax.set_title(\'Best Observed Value by Number of Active Parameters\', fontsize=12)\nax.legend(loc=\'upper right\')\nax.grid(True, alpha=0.3)\n\nplt.tight_layout()\nplt.show()\n'})}),"\n",(0,i.jsx)(l.A,{children:"Collected data for 50 trials\nNumber of active parameters per trial:\nMin: 0\nMax: 50\nMean: 12.9"}),"\n",(0,i.jsx)(A.p,{children:(0,i.jsx)(A.img,{src:"data:image/png;base64,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
1xESEoKkpCRpWUREBIYPH44aNWqgYcOGmDhxIlJSCv5/++DBg6hcubLO9gDgwIED6NKlC6pWrYo2bdpgzZo1Rb6/wtocOnQITz31FBITE7XaLFiwAK1atQKQ80t97r6oVasWXn/9ddy8eVOr/uTJkzF8+HDMnTsXLVq0QO3atfHBBx9A9d+sj99++y1WrlyJffv2Sevat2+f3niXLl2K+vXro0GDBhg7dixSU1O1yr/77juMHz9ea9tvvvkmPv74YzRr1gw1atTAlClTkJqaio8++gh16tRB/fr18d1332mt59atW+jfvz9q1qyJFi1aYMGCBdBoNHpj+uuvv3Dv3j1069ZNWrZ+/Xrs2bMH27dvR8eOHVG1alWMHz9eKzYA6NGjBzZs2ICsLOtfXcLEggpUqawP4FYogPbtcx4KhbWjISIiS8vIKPiRmVm8dYs17AyEh4dj4sSJmDVrFo4ePYqaNWsiNTUVkZGRWnUzMzPx4MED6QA1Li4OzZs3h4uLCzZt2oTly5fj0qVLePXVV43e3q5du/Dqq6/itddew86dOzFx4kRMmzZN6xfy/Ipq07x5cyQnJ2PLli1a7VasWIE2bdoAAPr37y+dudm4cSM8PDzQtm1bpKWlSfXj4+OxatUq3Lp1C8uWLcOCBQuwdu1afP311wByLgnq06cPWrRoIa2rRYsWOvH+9NNPeP/99zF+/HisW7cOCoUC8+bN06qT/1Ko+Ph4rFmzBvHx8VixYgXmzJmDefPmoXbt2khLS8OGDRswZcoUjB8/HocOHZLavfLKK7C3t8fmzZuxYMEC3Lt3D9u2bdO7Hw8dOoSwsDA4OztLyzZu3IgWLVqgWrVqWnXlcu3D96ZNmyI1NRWnTp3Su+6SZGftAKj0qpR3ytmyeCmUvT0wbpy1oyAiopKSZwCujkaNgOnTn7wu7B5HtWsDs2c/eT10KJDnDAEAYMcO0+MswPfff4/nnnvOqDYLFy5EpUqVMH/+fGnZ//73P/j5+eHq1auoVauWwdubMmUKpk2bhiFDhgAAqlWrhlmzZmHOnDkYNWqU3nUU1cbe3h6vvfYa/ve//+HNN98EkHPW5fLly9i0aRMAwNXVFa6uOT9mhoSEYPHixdi+fTv27dun9Qt+1apVsXjxYshkMtSuXRuDBw/GH3/8gSlTpsDNzQ2urq5wdHRESEhIge/5888/x9ixY/HGG28AyBn8fODAgQLr53rmmWfw7bffQiaTISwsDO3bt0dERISU2NSqVQuLFi3Cn3/+iTZt2iA7OxtXr17FkiVLEBYWBgBo1KhRgWcswsPDERQUpLXsxo0baNOmDSZPnoytW7fCw8MDnTp1wuTJk+Hi4iLV8/Pzg52dHe7evYvmzZsX+V4siYkFFcjHTQlnpR3SVGrey4KIiMiCZDIZGjRoYHS7Y8eO4fz586hcuTKEENIDAG7evFlgYpF/e2lpaTh//jwiIyPx7bffSuvJyMjA48ePodFodH4pN7TNgAED0KpVK0RGRqJChQr43//+h2effRY1atQAkHOGYM6cOdi/fz+io6ORnZ2NmJgY3L17V2t7tWrVgkwmk14HBAQYNaYiMzMT//zzD7766iut5a1atZIuBStI/m37+fnB3d1dq46fn58Uj0KhQM+ePTFgwAAMHz4cbdu2RZMmTaAo4AoKtVoNOzvtw/KsrCysWrUKo0ePxubNmxEdHY1Ro0bhypUr+OWXX7Tq2tnZQa1WF74DSgATCyqQTCZDJV9X/PMgAdGJ6UjPVMPJoQx9ZIR48muUUgnk+UIgIqJy6L9fwPXKd1CMwgbn5q+7cqXpMRlILpfD3t7e6HYZGRl46aWXdA6WARQ6IDz/9nLHKnzxxRfo0KGD3vr5GdqmefPmqFy5MtavX48JEyZI/+YaPHgwYmNjMXv2bFSuXBlKpRJt27aV1p9L30F5bhJlCLVaDY1GAwcHB63l+V/ro2/bRcWzYcMG7NixA7t378bAgQOhVquxdetWPPvsszrtgoKCcPXqVa1l/v7+0qButVqNOnXq4LPPPkOfPn2QkJAAT09PAEBycjIyMjJ0znhYQxk6SiRrqOiXk1gAOZdDVa/gadV4jKJSPTktvmkT4Oho3XiIiMiyjPmet1TdYuTu7q4zCDs8PFzrdVhYGA4cOIAKFSroPfg3lJeXl3Rwa+iUrca0ef311/G///0PjRo1QmRkJPr16yeV/fHHH9i8eTPat28PIOdAOSIiwuj3oFAoCk00nJ2dUaFCBVy6dEnaFgBcuHDB6G0ZQi6Xo3v37ujevTs0Gg26d++Ozz//XGe8CZAzTmL+/PnIysqSEr6mTZvizz//1Krn5pZzmXpGnnE+Z86cgUKhQJMmTSzyPozBwdtUKK1xFrwcioiIqMTUr18fKpUKP/30EwDgwYMH+PTTT7XqvPPOOwgPD8eECROkwc537tzBiBEjdH7xL8oHH3yAhQsXYuvWrdBoNFCr1Thy5AhmzpxpdpsBAwbg3Llz+Pjjj9GhQwcEBARIZcHBwdi9ezc0Gg3S0tIwcuRIrQNnQwUHB+P27duFzo40cuRIfP3117h+/TqAnAHSBc0eZY7Y2FiMGzdOSpBSU1MRFxcHPz8/vfU7dOgApVKpNfj77bffxo0bN7Bx40YAQGJiIubOnYvnnnsOgYGBUr3du3ejY8eOOpdmWQMTCypUxbI+MxQREVEZVbFiRcyfPx9vv/02/P390bJlS3Tt2lWrTs2aNbF3714cPXoU7u7u8PT0xAsvvICGDRtCqVQatb1x48bh888/x5gxY+Dm5gYvLy98/PHH6NSpk9ltqlevjmeffRaHDx/GgAEDtMq+//57bNmyBZ6envD19YUQQmcmJEP0798f9vb28PHxKXC62YkTJ6Jt27YICwuDp6cn5s6di759+xq9raJ4eXnh6aefRvPmzeHt7Y2AgAD4+fnhs88+01vf2dkZb7/9NlasWCEtq1atGrZs2YJJkybBz88PQUFBcHR0xObNm6U6WVlZWLduHcaVkslqZMKYi9NsXFJSEjw8PJCYmGiVrFCj0eDRo0fw9/c363SnMR4lpmPgt/sBAM89449P+jYuke0Wi4yMcnUplDX6n0oP9r9tY/8/kZGRgTt37uCpp56CYxn9Xn/48CFcXV2ly1pUKhViY2MRHByst352djbi4+Ph4+MDtVqN6OhoBAcHaw0mBnIuIRJCFHmMUtT2gJxf3N3d3XXGHyQmJiIrKwu+vr4Gt8mVkJCAlJQUBAYG6gxUzm2fO7PTo0eP4OzsLM0WlZCQAI1GA29vb6l+SkoKUlNTtc5+ADnHa0lJSfD19S3wM5KSkgK5XA5nZ2ckJSVBpVJJZxPyv0d9246PjweQk0Dkevz4MeRyuTT2IW9dNzc3ve85r8TERNSsWRN//PGHNJMUkDNuIyoqCj4+Pjr7dvHixdixYwd27dpV6LqLUtjflTHHvxxjQYXyc3eEk4MC6ZnZuMczFkRERGbLP8hWqVQWepAvl8vh4eEBALC3ty9wOtXcRKUoRW0PgN7EAYAUhzFtcnl6euocdBfU3t/fX6dtfnmnqc3L3d29yAPgvO3y183/HvVtO29CkStv4lFUXX08PDxw5coVvYP4fX199SYmffr0MXhMTEmw7Z89qEgymQwVfXO+qKIT0pGRaf2pzIiIiIjKIy8vL73JUkF8fHy0bqpnbUwsqEgV/XI+4ALA/bjUwisTERERkU3ipVBUpEr5BnA/E1TwadBSRS4HWrR48pyIiIiILIaJBRUp94wFUMamnHVwAD780NpREBEREdkE/oxLRcp7L4vw2DKUWBARUbnFSS2Jik9x/T0xsaAi+Xs4QWmfc9v6e7GcGYqIiKwnd8ac3JvBEZH5cv+e9M1IZQxeCkVFkstkqOjrin8fJuLh4zSosrKlRKNUK2f3sSAiIkChUMDT0xOPHj0CkHNjsfz3cyhvhBBQq9Wws7Mr9++VdFmy/4UQSEtLw6NHj+Dp6QmFwrzjOyYWZJDcxEIAiIhLQZXAMjKAm4iIyp3AwEAAkJKL8k4IAY1GA7lczsTCBpVE/3t6ekp/V+ZgYkEG0RpnEcPEgoiIrEcmkyEoKAj+/v7IysqydjgWp9FoEBcXBx8fH5u/87otsnT/29vbm32mIhcTCzJIpbwzQ3EANxERlQIKhaLYDohKM41GA3t7ezg6OjKxsEFlqf9Ld3RUalTMdy8LIiIiIqK8mFiQQQI8neFgl/NxKVP3siAiIiKiEsHEggyikMuksxaR8anIVGdbOSIiIiIiKk2YWJDBchMLjQAexKVaORoDyOVAo0Y5j1J+TSIRERFRWcfB22Swivlmhnoq
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
1RIvFScVDewC3EeMs1Gpg6dKcBxMLIiIiIosyObGYNm0a7t27h4MHDyItTftgb/jw4Vi2bJnZwRVk27Zt2Lt3L2bOnIkLFy7g66+/xvz58/Hee+8V2m7QoEE4cuQIfv/9d+zZswcnT55E3759LRYnFY9Ar7xTznJmKCIiIqLSyOQxFhs3bsTGjRvRvHlznbKwsDCcPHnSrMAK8+qrr+LVV1+VXr/wwgt47733MHv2bHz33Xd621y7dg3btm3Dn3/+ifr16wMAvvvuO7Rp0wYXLlxAvXr1LBYvmSfA40liYfTMUERERERUIkw+YxEdHa11MC6TyaTnSqUSGRkZ5kVmpKSkJDg7OxdYfvToUSgUCrRp00Za1rJlSzg6OuLo0aMlESKZKO/dt3nGgoiIiKh0MvmMRUBAAC5evIhmzZoB0E4sTp06hUqVKpkfnYHu3LmD7777DmPGjCmwTmRkJLy9vWFn9+Qty+Vy+Pr64uHDh3rbqFQqqFQq6XVSUhIAQKPRQKPRFFP0htNoNBBCWGXb1uTjpoQMgEDOGQuD379GA5kQAACh0QBlfL/Zav9TDva/bWP/2zb2v22zdv8bs12TE4tu3bphzJgx+PXXXxESEiItj4qKwsSJE9G9e3ej1hcYGIiUlJQCy+vWrYtjx47pLI+NjcVLL72EZ599FtOnTy90G3I99zKQy+UQ/x185jd79mzMnDlTZ3lMTEyJn5EBcjo2MTERQgi976U883Kxx+PULDyMT8WjR48Ma5SRAc//BvQnPHoEODpaMELLs+X+J/a/rWP/2zb2v22zdv8nJycbXNfkxGLGjBlo0qQJqlWrhiZNmkAIgW7duuHw4cPw8/PDlClTjFrfrVu3CjzABwCFQvfeBXFxcWjfvj38/PywY8cO2NvbF9je398fcXFx0Gg0Wp0SGxsLf39/vW0mT56MCRMmSK+TkpIQGhoKPz8/uLu7G/K2ipVGo4FMJoOfn5/NfbEEebngcWoCkjPUCE8CGj7tB4VcVmib7LR0pGXLkKXWIDpDgVohBrTRCFy5/xiPU1TwdlUiLNTbIm1MYcv9T+x/W8f+t23sf9uVrRG4FB6H8DgNKjnYoU4lH4scYxTG0YgfZs26FOrEiROYOXMmduzYAYVCgdOnT6NPnz749NNP4eXlZdT6XFxcjKr/+PFjdOjQAZ6envj9998LHV8BAM2aNYNarcbff/+NFi1aAABOnDiBtLQ0NG3aVG8bpVIJpVKps1wul1vtD1smk1l1+9Zw9J+HuBmdJL3+eMMZ+Lo5YmSnWmhZM6jANkt3XYLXM50BADc3nIO3x40i2yzZcxWxyU/ORhmyHWPbmMMW+5+eYP/bNva/bWP/2x7dY4y7Fj3GKIgxnzmZKOw0QSkVHx+P9u3bw83NDTt37iwwKQkJCcGoUaOksydt27aFSqXC1q1bIZfL0atXL6jVar2XWOmTlJQEDw8PJCYmWu2MxaNHj+Dv728zXyxH/3mITzafLbD8o14Ndf64SnMbc9hi/9MT7H/bxv63bex/21PSxxiFMeb41+QzFta0ceNGnDt3Do6OjggICNAqe/DgATw8PAAAKSkpWjfN27BhA9566y1UrFgRANCxY0etG/tR6ZKtEViy52qhdRbtvoKn/N0h/++0oEYjsHD3lVLZxlwajQZxSSpk26fxPxYbxP63bex/28b+ty2GHGMs3XsVzaoHlvhlUUUx+IzF+fPnAUC6B0Tu60JXLpPBx8cHFSpUKNY/BLVaXeDgaVdXV+l5amoq7O3t4eDgoFUnd3S9vnEbheEZi5J14W4cPlh73OT28uxs1L17CQBwsXIdaIzsbyIiIqLS6suBTVGvso/Ft2ORMxYNGjQAAGmAde5rQ1SqVAk//fST3pvpmcLOzk4rgShIQZdI2cJBeXnwOMW8mbcUmmz0+PtXAMCVirWYWBAREVG5Ye5xkiUYnFhs2rSp0NcFSU5Oxi+//IJ3330Xp0+fNi46smnerobNQlC/sg+8XHMG2cenqHD+blypbGMuIQRUGSooHZVa940h28D+t23sf9vG/rcthh5jGHqcVJIMTix69epV6OvCvPrqqzpjIYiKUruiN3zdHLVmXMrPz90Rn/dvIl1jmK0RGPTt/lLZxly2dikcaWP/2zb2v21j/9sWQ48xalf0LsGoDFMin053d3fs2rWrJDZF5YhCLsPITrUKrTOiYy2tA/fS3IaIiIioKGX5GMOsxOLBgwcYO3YsateujaCgINSpUwdjx45FZGSkTt3nn3/enE2RjWpZMwgf9WoIXzft031+7o4FTrVWmtsQERERFaWsHmOYfB+Lq1evonXr1khOTkbDhg0REBCA6OhonDlzBp6enjh8+DBq1KhR3PFaFWeFsp5sjcDle4/xOCUD3q45p/8MufN2SrdXkJWtQeSyNQh7poJBd9E2ejsmtDGFLfc/sf9tHfvftrH/bVfOnbdjcedBDJ4K9kOdSr4lfqaiRO5jMX78eFSrVg0bN25ESEiItPz+/fvo3bs3xo8fz8ufqNgo5DKjp1RTyGXwcM6Zati3kg9gwB+iqdspieneiIiIyLYo5DLUreSDQKds+Pv7FNu9sSzF5MTiyJEjOHXqlFZSAQChoaFYsWIFmjRpYnZwRGaxtwcmTXrynIiIiIgsxuTEwsXFBcHBwXrLgoODC7yHBFGJUSiAli2tHQURERGRTTD5Qr033nijwHtZbNiwAUOGDDE5KCIiIiIiKlsMPmNx8+ZNrdfDhg3D9OnTcfXqVXTr1k0avL19+3Y8fPgQM2fOLPZgiYySnQ38/XfO82bNcs5gEBEREZFFGJxYPPPMMwWWzZ8/X2fZxo0bYeKEU0TFIysL+OKLnOebNjGxICIiIrIggxOL5cuXWzIOIiIiIiIqwwxOLIYNG2bJOIiIiIiIqAwzeVao2NhYHDp0CPfu3YMQApUqVcLzzz8PHx/O509EREREZGuMTixUKhU+/PBDLFmyBCqVSqtMqVRi9OjRmD17NhwcHIotSCIiIiIiKt2MSiyys7PRo0cP7Nq1Cx07dsSLL76IihUrQiaTITw8HDt37sS8efNw48YN/Prrr7ztPBERERGRjTAqsfjhhx9w+PBh7Nq1C507d9YpHzt2LHbv3o1evXrhxx9/xODBg4srTiIiIiIiKsWMOqWwevVqzJkzR29Skatz586YPXs2Vq9ebXZwRGaxswPGjct52Jk8nIiIiIiIDGBUYnHp0iX06dOnyHqvvfYaLl68aHJQRMXCzg5o3z7nwcSCiIiIyKKMSizS09Ph7u5eZD0PDw+kpaWZHBQREREREZUtRiUWFSpUwPnz54usd+HCBQQHB5saE1HxyM4GTp3KeWRnWzsaIiIionLNqMSiY8eOmDp1KrKysgqsk5WVhSlTpuCFF14wOzgis2RlAbNm5TwK+cwSERERkfmMSiw++OADHD9+HO3bt8fx48d1ynPLTpw4gQ8++KDYgiQiIiIiotLNqBGtVapUwaZNm9CnTx80a9YM3t7eqFixIgAgPDwc8fHxcHNzw6ZNm1ClShWLBExERERERKWP0Xew69KlC65cuYIJEyYgKCgIN2/exK1bt1ChQgWMHz8ely9fRpcuXSwRKxERERERlVImzcEZGhqKr7/+Gl9//XVxx0NERERERGWQ0WcsiIiIiIiI8mNiQUREREREZuPtiKn8srMDRox48pyIiIiILIZHW1R+2dkBXbtaOwoiIiIim8BLoYiIiIiIyGwmJxabN282q5zI4jQa4NKlnIdGY+1oiIiIiMo1ky+F6t27N4QQJpcTWVxmJjBlSs7zTZsAR0frxkNERERUjlnkUiiVSgWFQmGJVRMRERERUSlk1BmLy5cvF/oayEkqdu7cieDgYPMiIyIiIiKiMsOoxKJOnTqFvs4lk8nw5Zdfmh4VERERERGVKUYlFmvXrpWeDxw4UOt1LhcXF4SFhaFatWrmR0dERERERGWCUYnFgAEDpOd3797Vek1ERERERLbL5MHb06ZN07s8OTnZ5GCIiIiIiKhsMjmxiIqKwqxZs6TXkZGRqFevHtzd3dGoUSPExsYWS4BEJrOzA4YMyXnY8SbzRERERJZkcmIxa9YsBAYGSq8/++wz3L17F++99x5SUlIwZ86cYgmQyGR2dkDPnjkPJhZEREREFmVyYrFnzx507txZer19+3ZMnToVc+fOxf/+9z9s27atOOIjIiIiIqIywOSfcSMjI+Hv7w8AuH37NiIiItClSxcAQFhYGCIjI4snQiJTaTTArVs5z6tUAeQWuR8kEREREcGMMxYBAQG4ePEiAODXX3+Fl5cXwsLCAACPHj2Skg4iq8nMBCZMyHlkZlo7GiIiIqJyzeQzFt27d8drr72GF198Ef/73//Qq1cvyP/7RfjEiRNo1qx
1ZsQVJRERERESlm8mJxSeffIJHjx7hl19+QbNmzfD5559LZfPnz8fcuXOLJUAiIiIiIir9TE4s3N3dsX79er1lhw8fhkKhMDkoIiIiIiIqW8wazSqEwO7duzF58mS89dZb0vKjR48iOzvb7OCIiIiIiKhsMPmMRVpaGrp164b9+/dDqVRCpVLh+++/BwAsW7YMMTEx6NWrV7EFSkREREREpZfJZyymTZuGe/fu4eDBg0hLS9MqGz58OJYtW2Z2cEREREREVDaYfMZi48aN2LhxI5o3b65TFhYWhpMnT5oVGJHZ7OyAfv2ePCciIiIiizH5aCs6Ohr16tWTXstkMum5UqlERkaGeZERmcvODnj9dWtHQURERGQTiuUGeYB2YnHq1ClUqlTJvMiIiIiIiKjMMDmx6NatG8aMGYOIiAit5VFRUZg4cSK6d+9udnAFUalUWLp0KRo1agQXFxc8/fTTmDZtWpFnSXx9fSGTybQeM2bMsFicZGVCAPfu5TyEsHY0REREROWayZdCzZgxA02aNEG1atXQpEkTCCHQrVs3HD58GH5+fpgyZUpxxqll165duHjxIpYtW4YaNWrg4sWLeO211/Do0SNpZqqCrF+/Hn379rVYbFSKqFTA6NE5zzdtAhwdrRsPERERUTlm1qVQJ06cwODBg3Hz5k0oFAqcPn0affr0wbFjx+Dl5VWccWp55ZVXsHjxYjz77LNwcXFBs2bNMH78eGzcuNFi2yQiIiIiooKZdYO8gIAALF68GPfv30dWVhYePnyI77//Hv7+/sUVn8FiYmLg4eFRZL1Ro0bB0dER1atXx8yZM6FSqUogOiIiIiKi8q1czMF548YNLFy4EJMmTSq0Xs+ePTFq1ChUrVoVR44cwdChQxEeHo5Vq1bpra9SqbQSj6SkJACARqOBRqMpvjdgII1GAyGEVbZdJmk0kP03tkJoNEAZ32/sf9vG/rdt7H/bxv63bdbuf2O2KxPC/FGtQgjs2LEDly9fRmBgIHr37g03Nzej1uHq6orU1NQCy+vVq4fz58/rLH/48CFatWqFGjVq4Ndff4VCoTB4m+vWrcPAgQMRGxsLHx8fnfIZM2Zg5syZOstv3Lhh9PsrDhqNBomJifDw8IBcbtbJJtuQkQHPt94CACR8/32ZH2PB/rdt7H/bxv63bex/22bt/k9OTka1atWQmJgId3f3QusadcYiKysLY8aMwfr16+Ho6IjRo0dj6tSp6NatG3bv3i3V+/TTT3HixAn4+fkZvO6UlBRjQgGQMwNVu3btULVqVWzZssWopAIA6tSpAwC4efOm3sRi8uTJmDBhgvQ6KSkJoaGh8PPzK3LHWoJGo4FMJoOfnx+/WAyRkQGZgwMA5FyeVw4SC/a/7WL/2zb2v21j/9s2a/e/oxHHT0YlFgsWLMDy5cvRokULAMDMmTMRGRmJs2fPYv78+ahWrRquX7+Ozz//HF9++SW++uor4yI3QnR0NNq1a4eKFSti27ZtUCqVRq/j0qVLAIAKFSroLVcqlXrXK5fLrfaHLZPJrLr9MkUuB/67v4pMLs95Xcax/20b+9+2sf9tG/vftlmz/43ZplGJxY8//ohvv/0WY8aMAQB89913ePfdd/Hbb7+ha9euAIAuXbqgSpUqmDRpksUSi5iYGK2koqBMytPTE+PGjcOMGTOwdetWXLlyBf369UNQUBCOHDmCDz74AK+++ipCQ0MtEidZmZ0d0KPHk+dEREREZDFGpT13797F66+/Lr3u378/AKB169Za9dq0aYPw8PBiCE+/LVu24OrVq9izZw+cnJy0bniXkJCgt02XLl1gb2+Pl156Cf7+/pgwYQJGjRqFdevWWSxOsjI7O+DNN3MeTCyIiIiILMqowdsymQz5q+tbVtjysiwpKQkeHh4GDV6xBI1Gg0ePHsHf35+nQm0Q+9+2sf9tG/vftrH/bZu1+9+Y41/+jEvllxBATEzOcz8/abwFERERERU/oxOLpUuXGrSMyOpUKmDo0JznmzaV+VmhiIiIiEozoxOLkSNHGrSMiIiIiIhsh1GJxR9//GGpOIiIiIiIqAwzKrHo0KGDpeIgIiIiIqIyjFMLEBERERGR2ZhYEBERERGR2ZhYEBERERGR2XgfCyq/FArgxRefPCciIiIii2FiQeWXvT3AqZCJiIiISgQvhSIiIiIiIrPxjAWVX0IASUk5z93dAZnMuvEQERERlWNMLKj8UqmAAQNynm/aBDg6WjceIiIionKMl0IREREREZHZmFgQEREREZHZmFgQEREREZHZmFgQEREREZHZmFgQEREREZHZmFgQEREREZHZON0slV8KBdC+/ZPnRERERGQxTCyo/LK3B8aNs3YURERERDaBl0IREREREZHZeMaCyi8hcu6+DQBKJSCTWTceIiIionKMZyyo/FKpgN69cx65CQYRERERWQQTCyIiIiIiMhsTCyIiIiIiMhsTCyIiIiIiMhsTCyIiIiIiMhsTCyIiIiIiMhsTCyIiIiIiMhvvY0Hll1wOtGjx5DkRERERWQwTCyq/HByADz+0dhRERERENoE/4xIRERERkdmYWBARERERkdmYWFD5lZEBdOuW88jIsHY0REREROUaEwsiIiIiIjIbEwsiIiIiIjIbEwsiIiIiIjIbEwsiIiIiIjIbEwsiIiIiIjIbEwsiIiIiIjIb77xN5ZdcDjRq9OQ5EREREVkMEwsqvxwcgOnTrR0FERERkU3gz7hERERERGQ2JhZERERERGQ2JhZUfmVkAL165TwyMqwdDREREVG5xjEWVL6pVNaOgIiIiMgm8IwFERERERGZjYkFERERERGZjYkFERERERGZjYkFERERERGZjYkFERERERGZjbNCUfkllwO1az95TkREREQWU+YTi8zMTDg4OBjVRq1WQwgBe3t7C0VFpYKDAzB7trWjICIiIrIJZfJnXCEE1q9fj2effRZeXl5wdnZGly5dcO3atULbRUdHo1u3bnBycpLaREZGllDURERERETlV5lMLJKTk7Fv3z788MMPSE5Oxv379+Hg4IAXX3yx0HavvfYa4uPjERkZiaioKKSnp6Nnz54QQpRQ5ERERERE5VOZTCzc3d2xYsUK1K5dG3K5HD4+PnjjjTdw584dJCYm6m1z7tw5HD58GF999RX8/Pzg4+ODr776CidOnMCJEydK+B1QicjIAPr3z3lkZFg7GiIiIqJyrUwmFrnS09MRHx+PCxcu4Ntvv0XPnj3h4eGht+6JEydgZ2eHpk2bSssaN24MZ2dnJhblWVJSzoOIiIiILKrUDN5OTEws9JIkhUIBNzc3rWWTJk3CmjVrkJycjEaNGuGnn34qsP2jR4/g4+MDmUymtdzPzw+PHj3S20alUkGlUkmvk/47QNVoNNBoNEW+p+Km0WgghLDKtsskjQay/z5TQqMByvh+Y//bNva/bWP/2zb2v22zdv8bs91Sk1iEhYUhJSWlwPI6dergyJEjWsu+/fZbfPvtt4iOjsaoUaPQqlUrXLp0Cc7OznrXoW/HqNVqnWQj1+zZszFz5kyd5TExMciwwqU1Go1GSsDknD61aBkZ8MzMBAAkPHoEODpaOSDzsP9tG/vftrH/bRv737ZZu/+Tk5MNrltqEouIiAiT2wYEBGDBggUIDQ3F4cOH0blzZ506wcHBePz4MdRqNezsct62RqNBXFwcKlSooHe9kydPxoQJE6TXSUlJCA0NhZ+fH9zd3U2O11QajQYymQx+fn78YjFERgZk/01F7O/vXy4SC/a/7WL/2zb2v21j/9s2a/e/oxHHT6UmsTBX7mVKee9pkZiYCKVSCUdHR7Rs2RLZ2dk4ePAgOnToAAA4cuQIMjIy0LJlS73rVCqVUCqVOsvlcrnV/rBlMplVt1+myOXAf2ejZHJ5ubhJHvvftrH/bRv737ax/22bNfvfmG2WyU/n+vXr8cEHH+DMmTOIiorCoUOHMHjwYISFhaFVq1ZSvUqVKmHOnDkAgOrVq6Nnz5549913cfr0aZw7dw5jxozBiy++iLp161rrrRARERERlQtl8oxF7969kZCQgDFjxiA8PByBgYHo0qUL3nvvPa27aXt4eGidvvnhhx/w4YcfSveu6Nq1K7788ktrvAUqCXI58MwzT54TERERkcXIBO8OZ7CkpCR4eHggMTHRamMsHj16BH9/f54KtUHsf9vG/rdt7H/bxv63bdbuf2OOf/npJCIiIiIiszGxICIiIiIiszGxoPJLpQKGDs155LnRIREREREVvzI5eJvIIEIAuXdV51AiIiIiIoviGQsiIiIiIjIbEwsiIiIiIjIbEwsiIiIiIjIbEwsiIiIiIjIbEwsiIiIiIjIbZ4Wi8ksmA0JDnzwnIiIiIothYkHll1IJLF5s7SiIiIiIbAIvhSIiIiIiIrMxsSAiIiIiIrMxsaDyS6UCRo3KeahU1o6GiIiIqFzjGAsqv4QA7t9/8pyIiIiILIZnLIiIiIiIyGxMLIiIiIiIyGxMLIiIiIiIyGxMLIiIiIiIyGxMLIiIiIiIyGycFYrKL5kM8Pd/8pyIiIiILIaJBZVfSiWwcqW1oyAiIiKyCbwUioiIiIiIzMbEgoiIiIiIzMbEgsqvzExgwoScR2amtaMhIiIiKtc4xoLKL40G+PffJ8+JiIiIyGJ4xoKIiIiIiMzGxIKIiIiIiMzGxIKIiIiIiMzGxIKIiIiIiMzGxIKIiIiIiMzGWaGofHN3t3YERERERDaBiQWVX46OwP/+Z+0oiIiIiGwCL4UiIiIiIiKzMbEgIiIiIiKzMbEwQY2FNRAyLwQH7x7UWn4n/g5C5oUgZF4I3tn5jk67l9e/LJXnt+b8Gqls6z9btcqSVckImReCivMrYtS+UTpth/w6RGoblxanVfbbjd+ksu/PfK/T9ukFTyNkXgg6r+usUzZx70Sp7fXY61plf9//WyqbfWS2TttG3zdCyLwQNPq+kU7Z7COzpbZ/3/9bq+x67HWpbOLeiTptO6/rjJB5IXh6wdM6Zd+f+V5q+9uN34DMTGDyZGDyZMQlPJTKhvw6RKdt/639pfJkVbJW2dZ/tkpla86v0WmbW/by+pd1yt7Z+Y5Ufif+jlbZwbsHpbJ5f8/TaVt3SV2EzAtBy1UtdcpmHpwptT0TeUar7FL0Jals6r6pOm3b/dAOIfNCUGNhDZ2yhScXSm333tqrVfYw+ck+fHvH2zpte2/qLZWr1Cqtsp8u/SSV/XTpJ60ylVollfXe1FtnvW/veFsqf5j8UKts7629UtnCkwt12ub+rbb7oZ1O2dR9U6W2l6IvaZWdiTwjlc08OFOnbctVLREyLwR1l9TVKZv39zypbbF+R1xYg4brGqLi/IoFfkeEzAtB/639ddryOyKHzndEHnFpcaX6O+Kb49/otOV3RA5b+o7o/nN3s44j+B1RNr8jKs6viIbrGpaa44jCcIyFCR4mPwSyoPPFmC2y8SD5AQAgPiNep11MWoxUnl9qZqpUlpaVplUmIKSyxx6PddrGpcVJ5Rqh0SpLz0qXylIyU3TaPkh+gMzsTPi7+OuUxWfES23VGrVWmSpbJZUlqZJ02kalRBX4
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",alt:""})}),"\n",(0,i.jsx)(A.h2,{id:"10-key-takeaways",children:"10. Key Takeaways"}),"\n",(0,i.jsx)(A.h3,{id:"why-bonsai--map-saas",children:"Why BONSAI + MAP-SAAS?"}),"\n",(0,i.jsxs)(A.ol,{children:["\n",(0,i.jsxs)(A.li,{children:["\n",(0,i.jsxs)(A.p,{children:[(0,i.jsx)(A.strong,{children:"MAP-SAAS"}),": MAP-SAAS is a variant of the Sparse Axis Aligned Subspace prior\n(Eriksson & Jankowiak. High-dimensional Bayesian optimization with sparse\naxis-aligned subspace, UAI, 2021), and places a half-Cauchy prior on the GP\nlengthscales. As a result, SAAS models encourage model sparsity, where less relevant\ninputs are driven toward long lengthscales. This improves performance on\nhigh-dimensional tasks, and is synergistic with BONSAI. Standard SAAS models use a\ntime consuming, Bayesian (MCMC) inference procedure; MAP-SAAS provides many of the\nbenefits of the fully Bayesian MAP SAAS by ensembling over just a few models\nestimated via MAP with significantly lower computational costs."]}),"\n"]}),"\n",(0,i.jsxs)(A.li,{children:["\n",(0,i.jsxs)(A.p,{children:[(0,i.jsx)(A.strong,{children:"BONSAI"}),": BONSAI prunes irrelevant dimensions via ",(0,i.jsx)(A.code,{children:"simplify_parameter_changes=True"}),"\nand sets them to the ",(0,i.jsx)(A.code,{children:"pruning_target_parameterization"})," (the status\nquo/default/production values or a target point of interest). This simplifies the\nproposals so that they change fewer parameters, making the proposals more\ninterpretable and more likely to avoid regressions in metrics not captured in the\noptimization objective."]}),"\n"]}),"\n"]}),"\n",(0,i.jsx)(A.h3,{id:"when-to-use-this-approach",children:"When to Use This Approach"}),"\n",(0,i.jsxs)(A.ul,{children:["\n",(0,i.jsxs)(A.li,{children:[(0,i.jsx)(A.strong,{children:"Real-world optimization"})," where simple, interpretable changes are desired."]}),"\n"]})]})}function h(e={}){const{wrapper:A}={...(0,a.R)(),...e.components};return A?(0,i.jsx)(A,{...e,children:(0,i.jsx)(c,{...e})}):c(e)}}}]);
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.