Surrogates.jl: Surrogate models and optimization for scientific machine learning
A surrogate model is an approximation method that mimics the behavior of a computationally expensive simulation. In more mathematical terms: suppose we are attempting to optimize a function $\; f(p)$, but each calculation of $\; f$ is very expensive. It may be the case that we need to solve a PDE for each point or use advanced numerical linear algebra machinery, which is usually costly. The idea is then to develop a surrogate model $\; g$ which approximates $\; f$ by training on previous data collected from evaluations of $\; f$. The construction of a surrogate model can be seen as a three-step process:
- Sample selection
- Construction of the surrogate model
- Surrogate optimization
The sampling methods are super important for the behavior of the surrogate. Sampling can be done through QuasiMonteCarlo.jl, all the functions available there can be used in Surrogates.jl.
The available surrogates are:
- Linear
- Radial Basis
- Kriging
- Custom Kriging provided with Stheno
- Neural Network
- Support Vector Machine
- Random Forest
- Second Order Polynomial
- Inverse Distance
After the surrogate is built, we need to optimize it with respect to some objective function. That is, simultaneously looking for a minimum and sampling the most unknown region. The available optimization methods are:
- Stochastic RBF (SRBF)
- Lower confidence-bound strategy (LCBS)
- Expected improvement (EI)
- Dynamic coordinate search (DYCORS)
Multi-output Surrogates
In certain situations, the function being modeled may have a multi-dimensional output space. In such a case, the surrogate models can take advantage of correlations between the observed output variables to obtain more accurate predictions.
When constructing the original surrogate, each element of the passed y
vector should itself be a vector. For example, the following y
are all valid.
using Surrogates
using StaticArrays
x = sample(5, [0.0; 0.0], [1.0; 1.0], SobolSample())
f_static = (x) -> StaticVector(x[1], log(x[2]*x[1]))
f = (x) -> [x, log(x)/2]
y = f_static.(x)
y = f.(x)
Currently, the following are implemented as multi-output surrogates:
- Radial Basis
- Neural Network (via Flux)
- Second Order Polynomial
- Inverse Distance
- Custom Kriging (via Stheno)
Gradients
The surrogates implemented here are all automatically differentiable via Zygote. Because of this property, surrogates are useful models for processes which aren't explicitly differentiable, and can be used as layers in, for instance, Flux models.
Installation
Surrogates is registered in the Julia General Registry. In the REPL:
using Pkg
Pkg.add("Surrogates")
Contributing
Please refer to the SciML ColPrac: Contributor's Guide on Collaborative Practices for Community Packages for guidance on PRs, issues, and other matters relating to contributing to SciML.
See the SciML Style Guide for common coding practices and other style decisions.
There are a few community forums:
- The #diffeq-bridged and #sciml-bridged channels in the Julia Slack
- The #diffeq-bridged and #sciml-bridged channels in the Julia Zulip
- On the Julia Discourse forums
- See also SciML Community page
Quick example
using Surrogates
num_samples = 10
lb = 0.0
ub = 10.0
#Sampling
x = sample(num_samples, lb, ub, SobolSample())
f = x -> log(x) * x^2 + x^3
y = f.(x)
#Creating surrogate
alpha = 2.0
n = 6
my_lobachevsky = LobachevskySurrogate(x, y, lb, ub, alpha = alpha, n = n)
#Approximating value at 5.0
value = my_lobachevsky(5.0)
#Adding more data points
surrogate_optimize!(f, SRBF(), lb, ub, my_lobachevsky, RandomSample())
#New approximation
value = my_lobachevsky(5.0)
135.03840648224897
Reproducibility
The documentation of this SciML package was built using these direct dependencies,
Status `~/work/Surrogates.jl/Surrogates.jl/docs/Project.toml`
[99985d1d] AbstractGPs v0.5.23
[e30172f5] Documenter v1.9.0
[587475ba] Flux v0.16.3
[91a5bcdd] Plots v1.40.11
[1fd47b50] QuadGK v2.11.2
[6fc51010] Surrogates v6.11.0 `~/work/Surrogates.jl/Surrogates.jl`
[78aa1720] SurrogatesAbstractGPs v0.1.0 `~/work/Surrogates.jl/Surrogates.jl/lib/SurrogatesAbstractGPs`
[4f55584b] SurrogatesFlux v0.1.1 `~/work/Surrogates.jl/Surrogates.jl/lib/SurrogatesFlux`
[778709c9] SurrogatesMOE v0.1.4 `~/work/Surrogates.jl/Surrogates.jl/lib/SurrogatesMOE`
[50679fc6] SurrogatesPolyChaos v0.1.0 `~/work/Surrogates.jl/Surrogates.jl/lib/SurrogatesPolyChaos`
[3fee2672] SurrogatesRandomForest v0.1.1 `~/work/Surrogates.jl/Surrogates.jl/lib/SurrogatesRandomForest`
[eb0c118b] SurrogatesSVM v0.1.0 `~/work/Surrogates.jl/Surrogates.jl/lib/SurrogatesSVM`
[009559a3] XGBoost v2.5.1
[e88e6eb3] Zygote v0.7.4
and using this machine and Julia version.
Julia Version 1.11.4
Commit 8561cc3d68d (2025-03-10 11:36 UTC)
Build Info:
Official https://julialang.org/ release
Platform Info:
OS: Linux (x86_64-linux-gnu)
CPU: 4 × AMD EPYC 7763 64-Core Processor
WORD_SIZE: 64
LLVM: libLLVM-16.0.6 (ORCJIT, znver3)
Threads: 1 default, 0 interactive, 1 GC (on 4 virtual cores)
Environment:
JULIA_DEBUG = Documenter
A more complete overview of all dependencies and their versions is also provided.
Status `~/work/Surrogates.jl/Surrogates.jl/docs/Manifest.toml`
[a4c015fc] ANSIColoredPrinters v0.0.1
[621f4979] AbstractFFTs v1.5.0
[99985d1d] AbstractGPs v0.5.23
[1520ce14] AbstractTrees v0.4.5
[7d9f7c33] Accessors v0.1.42
[79e6a3ab] Adapt v4.3.0
⌅ [c75e803d] AdaptiveRejectionSampling v0.1.2
[66dad0bd] AliasTables v1.1.3
[dce04be8] ArgCheck v2.4.0
[7d9fca2a] Arpack v0.5.4
[a9b6321e] Atomix v1.1.1
[198e06fe] BangBang v0.4.4
[9718e550] Baselet v0.1.1
[d1d4a3ce] BitFlags v0.1.9
[fa961155] CEnum v0.5.0
[082447d4] ChainRules v1.72.3
[d360d2e6] ChainRulesCore v1.25.1
[aaaa29a8] Clustering v0.15.8
[944b1d66] CodecZlib v0.7.8
[35d6a980] ColorSchemes v3.29.0
[3da002f7] ColorTypes v0.12.0
[c3611d14] ColorVectorSpace v0.11.0
[5ae59095] Colors v0.13.0
[861a8166] Combinatorics v1.0.2
[bbf7d656] CommonSubexpressions v0.3.1
[34da2185] Compat v4.16.0
[a33af91c] CompositionsBase v0.1.2
[2569d6c7] ConcreteStructs v0.2.3
[f0e56b4a] ConcurrentUtilities v2.5.0
[187b0558] ConstructionBase v1.5.8
[6add18c4] ContextVariablesX v0.1.3
[d38c429a] Contour v0.6.3
[9a962f9c] DataAPI v1.16.0
[864edb3b] DataStructures v0.18.22
[e2d170a0] DataValueInterfaces v1.0.0
[244e2a9f] DefineSingletons v0.1.2
[8bb1440f] DelimitedFiles v1.9.1
[163ba53b] DiffResults v1.1.0
[b552c78f] DiffRules v1.15.1
[b4f34e82] Distances v0.10.12
[31c24e10] Distributions v0.25.118
[ffbed154] DocStringExtensions v0.9.3
[e30172f5] Documenter v1.9.0
[f151be2c] EnzymeCore v0.8.8
[460bff9d] ExceptionUnwrapping v0.1.11
[95c220a8] ExtendableSparse v1.7.0
[c87230d0] FFMPEG v0.4.2
[7a1cc6ca] FFTW v1.8.1
[cc61a311] FLoops v0.2.2
[b9860ae5] FLoopsBase v0.1.1
[5789e2e9] FileIO v1.17.0
[1a297f60] FillArrays v1.13.0
[53c48c17] FixedPointNumbers v0.8.5
[587475ba] Flux v0.16.3
[1fa38f19] Format v1.3.7
[f6369f11] ForwardDiff v0.10.38
[d9f16b24] Functors v0.5.2
[38e38edf] GLM v1.9.0
[0c68f7d7] GPUArrays v11.2.2
[46192b85] GPUArraysCore v0.2.0
[28b8d3ca] GR v0.73.13
[d54b0c1a] GaussQuadrature v0.5.8
[cc18c42c] GaussianMixtures v0.3.12
[d7ba0133] Git v1.3.1
[42e2da0e] Grisu v1.0.2
[cd3eb016] HTTP v1.10.15
[076d061b] HashArrayMappedTries v0.2.0
[34004b35] HypergeometricFunctions v0.3.28
[88f59080] ILUZero v0.2.0
[b5f81e59] IOCapture v0.2.5
[7869d1d1] IRTools v0.4.14
[22cec73e] InitialValues v0.3.1
[18e54dd8] IntegerMathUtils v0.1.2
[3587e190] InverseFunctions v0.1.17
[92d709cd] IrrationalConstants v0.2.4
[42fd0dbc] IterativeSolvers v0.9.4
[82899510] IteratorInterfaceExtensions v1.0.0
⌅ [033835bb] JLD2 v0.4.54
[1019f520] JLFzf v0.1.9
[692b3bcd] JLLWrappers v1.7.0
[682c06a0] JSON v0.21.4
[0f8b85d8] JSON3 v1.14.1
[b14d175d] JuliaVariables v0.2.4
[63c18a36] KernelAbstractions v0.9.34
[ec8451be] KernelFunctions v0.10.65
[2d691ee1] LIBLINEAR v0.7.1
[b1bec4e5] LIBSVM v0.8.1
[929cbde3] LLVM v9.2.0
[b964fa9f] LaTeXStrings v1.4.0
[23fbe1c1] Latexify v0.16.6
[73f95e8e] LatticeRules v0.0.1
[0e77f7df] LazilyInitializedFields v1.3.0
[2ab3a3ac] LogExpFunctions v0.3.29
[e6f89c97] LoggingExtras v1.1.0
[c2834f40] MLCore v1.0.0
[7e8f7934] MLDataDevices v1.9.1
[d8e11817] MLStyle v0.4.17
[f1d291b0] MLUtils v0.4.7
[1914dd2f] MacroTools v0.5.15
[d0879d2d] MarkdownAST v0.1.2
[739be429] MbedTLS v1.1.9
[442fdcdd] Measures v0.3.2
[128add7d] MicroCollections v0.2.0
[e1d29d7a] Missings v1.2.0
[872c559c] NNlib v0.9.29
[77ba4419] NaNMath v1.1.2
[71a1bf82] NameResolution v0.1.5
[b8a86587] NearestNeighbors v0.4.21
[6fe1bfb0] OffsetArrays v1.16.0
[0b1bfda6] OneHotArrays v0.2.6
[4d8831e6] OpenSSL v1.4.3
[3bd65402] Optimisers v0.4.5
[bac558e1] OrderedCollections v1.8.0
[90014a1f] PDMats v0.11.32
[69de0a69] Parsers v2.8.1
[b98c9c47] Pipe v1.3.0
[ccf2f8ad] PlotThemes v3.3.0
[995b91a9] PlotUtils v1.4.3
[91a5bcdd] Plots v1.40.11
[8d666b04] PolyChaos v0.2.11
[aea7be01] PrecompileTools v1.2.1
[21216c6a] Preferences v1.4.3
[8162dcfd] PrettyPrint v0.2.0
[27ebfcd6] Primes v0.5.6
[33c8b6b6] ProgressLogging v0.1.4
[43287f4e] PtrArrays v1.3.0
[1fd47b50] QuadGK v2.11.2
[8a4e6c94] QuasiMonteCarlo v0.3.3
[c1ae055f] RealDot v0.1.0
[3cdcf5f2] RecipesBase v1.3.4
[01d81517] RecipesPipeline v0.6.12
[189a3867] Reexport v1.2.2
[2792f1a3] RegistryInstances v0.1.0
[05181044] RelocatableFolders v1.0.1
[ae029012] Requires v1.3.1
[79098fc4] Rmath v0.8.0
[6e75b9c4] ScikitLearnBase v0.5.0
[7e506255] ScopedValues v1.3.0
[6c6a2e73] Scratch v1.2.1
[efcf1570] Setfield v1.1.2
[1277b4bf] ShiftedArrays v2.0.0
[605ecd9f] ShowCases v0.1.0
[992d4aef] Showoff v1.0.3
[777ac1f9] SimpleBufferStream v1.2.0
[699a6c99] SimpleTraits v0.9.4
[ed01d8cd] Sobol v1.5.0
[a2af1166] SortingAlgorithms v1.2.1
[dc90abb0] SparseInverseSubset v0.1.2
[a0a7dd2c] SparseMatricesCSR v0.6.9
[e56a9233] Sparspak v0.3.9
[276daf66] SpecialFunctions v2.5.0
[171d559e] SplittablesBase v0.1.15
[860ef19b] StableRNGs v1.0.2
[90137ffa] StaticArrays v1.9.13
[1e83bf80] StaticArraysCore v1.4.3
[10745b16] Statistics v1.11.1
[82ae8749] StatsAPI v1.7.0
[2913bbd2] StatsBase v0.34.4
[4c63d2b9] StatsFuns v1.3.2
[3eaba693] StatsModels v0.7.4
[09ab397b] StructArrays v0.7.0
[856f2bd8] StructTypes v1.11.0
[6fc51010] Surrogates v6.11.0 `~/work/Surrogates.jl/Surrogates.jl`
[78aa1720] SurrogatesAbstractGPs v0.1.0 `~/work/Surrogates.jl/Surrogates.jl/lib/SurrogatesAbstractGPs`
[89f642e6] SurrogatesBase v1.1.0
[4f55584b] SurrogatesFlux v0.1.1 `~/work/Surrogates.jl/Surrogates.jl/lib/SurrogatesFlux`
[778709c9] SurrogatesMOE v0.1.4 `~/work/Surrogates.jl/Surrogates.jl/lib/SurrogatesMOE`
[50679fc6] SurrogatesPolyChaos v0.1.0 `~/work/Surrogates.jl/Surrogates.jl/lib/SurrogatesPolyChaos`
[3fee2672] SurrogatesRandomForest v0.1.1 `~/work/Surrogates.jl/Surrogates.jl/lib/SurrogatesRandomForest`
[eb0c118b] SurrogatesSVM v0.1.0 `~/work/Surrogates.jl/Surrogates.jl/lib/SurrogatesSVM`
[3783bdb8] TableTraits v1.0.1
[bd369af6] Tables v1.12.0
[62fd8b95] TensorCore v0.1.1
[3bb67fe8] TranscodingStreams v0.11.3
[28d57a85] Transducers v0.4.84
[5c2747f8] URIs v1.5.1
[1cfade01] UnicodeFun v0.4.1
[1986cc42] Unitful v1.22.0
[45397f5d] UnitfulLatexify v1.6.4
[013be700] UnsafeAtomics v0.3.0
[41fe7b60] Unzip v0.2.0
[009559a3] XGBoost v2.5.1
[e88e6eb3] Zygote v0.7.4
[700de1a5] ZygoteRules v0.2.7
⌅ [68821587] Arpack_jll v3.5.1+1
[6e34b625] Bzip2_jll v1.0.9+0
[4ee394cb] CUDA_Driver_jll v0.12.1+1
[76a88914] CUDA_Runtime_jll v0.16.1+0
[83423d85] Cairo_jll v1.18.4+0
[ee1fde0b] Dbus_jll v1.14.10+0
[2702e6a9] EpollShim_jll v0.0.20230411+1
[2e619515] Expat_jll v2.6.5+0
⌅ [b22a6f82] FFMPEG_jll v4.4.4+1
[f5851436] FFTW_jll v3.3.10+3
[a3f928ae] Fontconfig_jll v2.15.0+0
[d7e528f0] FreeType2_jll v2.13.3+1
[559328eb] FriBidi_jll v1.0.16+0
[0656b61e] GLFW_jll v3.4.0+2
[d2c73de3] GR_jll v0.73.13+0
[78b55507] Gettext_jll v0.21.0+0
[f8c6e375] Git_jll v2.47.1+0
[7746bdde] Glib_jll v2.82.4+0
[3b182d85] Graphite2_jll v1.3.14+1
[2e76f6c2] HarfBuzz_jll v8.5.0+0
[1d5cc7b8] IntelOpenMP_jll v2025.0.4+0
[aacddb02] JpegTurbo_jll v3.1.1+0
[c1c5ebd0] LAME_jll v3.100.2+0
[88015f11] LERC_jll v4.0.1+0
[dad2f222] LLVMExtra_jll v0.0.35+0
[1d63c593] LLVMOpenMP_jll v18.1.7+0
[dd4b983a] LZO_jll v2.10.3+0
⌅ [e9f186c6] Libffi_jll v3.2.2+2
[d4300ac3] Libgcrypt_jll v1.11.0+0
[7e76a0d4] Libglvnd_jll v1.7.0+0
[7add5ba3] Libgpg_error_jll v1.51.1+0
[94ce4f54] Libiconv_jll v1.18.0+0
[4b2f31a3] Libmount_jll v2.40.3+0
[89763e89] Libtiff_jll v4.7.1+0
[38a345b3] Libuuid_jll v2.40.3+0
[856f044c] MKL_jll v2025.0.1+1
[e7412a2a] Ogg_jll v1.3.5+1
[458c3c95] OpenSSL_jll v3.0.16+0
[efe28fd5] OpenSpecFun_jll v0.5.6+0
[91d4177d] Opus_jll v1.3.3+0
[36c8627f] Pango_jll v1.56.1+0
[30392449] Pixman_jll v0.44.2+0
⌅ [c0090381] Qt6Base_jll v6.7.1+1
⌅ [629bc702] Qt6Declarative_jll v6.7.1+2
⌅ [ce943373] Qt6ShaderTools_jll v6.7.1+1
⌃ [e99dba38] Qt6Wayland_jll v6.7.1+1
[f50d1b31] Rmath_jll v0.5.1+0
[a44049a8] Vulkan_Loader_jll v1.3.243+0
[a2964d1f] Wayland_jll v1.21.0+2
[2381bf8a] Wayland_protocols_jll v1.36.0+0
[a5c6f535] XGBoost_jll v2.0.1+0
[02c8fc9c] XML2_jll v2.13.6+1
[aed1982a] XSLT_jll v1.1.42+0
[ffd25f8a] XZ_jll v5.6.4+1
[f67eecfb] Xorg_libICE_jll v1.1.1+0
[c834827a] Xorg_libSM_jll v1.2.4+0
[4f6342f7] Xorg_libX11_jll v1.8.6+3
[0c0b7dd1] Xorg_libXau_jll v1.0.12+0
[935fb764] Xorg_libXcursor_jll v1.2.3+0
[a3789734] Xorg_libXdmcp_jll v1.1.5+0
[1082639a] Xorg_libXext_jll v1.3.6+3
[d091e8ba] Xorg_libXfixes_jll v6.0.0+0
[a51aa0fd] Xorg_libXi_jll v1.8.2+0
[d1454406] Xorg_libXinerama_jll v1.1.5+0
[ec84b674] Xorg_libXrandr_jll v1.5.4+0
[ea2f1a96] Xorg_libXrender_jll v0.9.11+1
[14d82f49] Xorg_libpthread_stubs_jll v0.1.2+0
[c7cfdc94] Xorg_libxcb_jll v1.17.0+3
[cc61e674] Xorg_libxkbfile_jll v1.1.2+1
[e920d4aa] Xorg_xcb_util_cursor_jll v0.1.4+0
[12413925] Xorg_xcb_util_image_jll v0.4.0+1
[2def613f] Xorg_xcb_util_jll v0.4.0+1
[975044d2] Xorg_xcb_util_keysyms_jll v0.4.0+1
[0d47668e] Xorg_xcb_util_renderutil_jll v0.3.9+1
[c22f9ab0] Xorg_xcb_util_wm_jll v0.4.1+1
[35661453] Xorg_xkbcomp_jll v1.4.6+1
[33bec58e] Xorg_xkeyboard_config_jll v2.39.0+0
[c5fb5394] Xorg_xtrans_jll v1.5.1+0
[3161d3a3] Zstd_jll v1.5.7+1
[35ca27e7] eudev_jll v3.2.9+0
[214eeab7] fzf_jll v0.56.3+0
[1a1c6b14] gperf_jll v3.1.1+1
[a4ae2306] libaom_jll v3.11.0+0
[0ac62f75] libass_jll v0.15.2+0
[1183f4f0] libdecor_jll v0.2.2+0
[2db6ffa8] libevdev_jll v1.11.0+0
[f638f0a6] libfdk_aac_jll v2.0.3+0
[36db933b] libinput_jll v1.18.0+0
[275f1f90] liblinear_jll v2.47.0+0
[b53b4c65] libpng_jll v1.6.47+0
[08558c22] libsvm_jll v3.25.0+0
[f27f6e37] libvorbis_jll v1.3.7+2
[009596ad] mtdev_jll v1.1.6+0
[1317d2d5] oneTBB_jll v2022.0.0+0
⌅ [1270edf5] x264_jll v2021.5.5+0
⌅ [dfaa095f] x265_jll v3.5.0+0
[d8fb68d0] xkbcommon_jll v1.4.1+2
[0dad84c5] ArgTools v1.1.2
[56f22d72] Artifacts v1.11.0
[2a0f44e3] Base64 v1.11.0
[ade2ca70] Dates v1.11.0
[8ba89e20] Distributed v1.11.0
[f43a241f] Downloads v1.6.0
[7b1f6079] FileWatching v1.11.0
[9fa8497b] Future v1.11.0
[b77e0a4c] InteractiveUtils v1.11.0
[4af54fe1] LazyArtifacts v1.11.0
[b27032c2] LibCURL v0.6.4
[76f85450] LibGit2 v1.11.0
[8f399da3] Libdl v1.11.0
[37e2e46d] LinearAlgebra v1.11.0
[56ddb016] Logging v1.11.0
[d6f4376e] Markdown v1.11.0
[a63ad114] Mmap v1.11.0
[ca575930] NetworkOptions v1.2.0
[44cfe95a] Pkg v1.11.0
[de0858da] Printf v1.11.0
[3fa0cd96] REPL v1.11.0
[9a3f8284] Random v1.11.0
[ea8e919c] SHA v0.7.0
[9e88b42a] Serialization v1.11.0
[6462fe0b] Sockets v1.11.0
[2f01184e] SparseArrays v1.11.0
[f489334b] StyledStrings v1.11.0
[4607b0f0] SuiteSparse
[fa267f1f] TOML v1.0.3
[a4e569a6] Tar v1.10.0
[8dfed614] Test v1.11.0
[cf7118a7] UUIDs v1.11.0
[4ec0a83e] Unicode v1.11.0
[e66e0078] CompilerSupportLibraries_jll v1.1.1+0
[deac9b47] LibCURL_jll v8.6.0+0
[e37daf67] LibGit2_jll v1.7.2+0
[29816b5a] LibSSH2_jll v1.11.0+1
[c8ffd9c3] MbedTLS_jll v2.28.6+0
[14a3606d] MozillaCACerts_jll v2023.12.12
[4536629a] OpenBLAS_jll v0.3.27+1
[05823500] OpenLibm_jll v0.8.1+4
[efcefdf7] PCRE2_jll v10.42.0+1
[bea87d4a] SuiteSparse_jll v7.7.0+0
[83775a58] Zlib_jll v1.2.13+1
[8e850b90] libblastrampoline_jll v5.11.0+0
[8e850ede] nghttp2_jll v1.59.0+0
[3f19e933] p7zip_jll v17.4.0+2
Info Packages marked with ⌃ and ⌅ have new versions available. Those with ⌃ may be upgradable, but those with ⌅ are restricted by compatibility constraints from upgrading. To see why use `status --outdated -m`
You can also download the manifest file and the project file.