diff --git a/Manifest.toml b/Manifest.toml index 4e8695f..c6518c0 100644 --- a/Manifest.toml +++ b/Manifest.toml @@ -1,8 +1,8 @@ # This file is machine-generated - editing it directly is not advised -julia_version = "1.10.0" +julia_version = "1.9.3" manifest_format = "2.0" -project_hash = "f82bc1ed2fe8b41d9591a6c5820977796331ff2f" +project_hash = "e2dfd06edc6a9e3f2fe028723bfb75222df5a783" [[deps.ADTypes]] git-tree-sha1 = "f5c25e8a5b29b5e941b7408bc8cc79fea4d9ef9a" @@ -347,9 +347,9 @@ uuid = "ade2ca70-3891-5945-98fb-dc099432e06a" [[deps.DelaunayTriangulation]] deps = ["DataStructures", "EnumX", "ExactPredicates", "Random", "SimpleGraphs"] -git-tree-sha1 = "26eb8e2331b55735c3d305d949aabd7363f07ba7" +git-tree-sha1 = "d4e9dc4c6106b8d44e40cd4faf8261a678552c7c" uuid = "927a84f5-c5f4-47a5-9785-b46e178433df" -version = "0.8.11" +version = "0.8.12" [[deps.DelimitedFiles]] deps = ["Mmap"] @@ -626,9 +626,9 @@ version = "3.3.9+0" [[deps.GLMakie]] deps = ["ColorTypes", "Colors", "FileIO", "FixedPointNumbers", "FreeTypeAbstraction", "GLFW", "GeometryBasics", "LinearAlgebra", "Makie", "Markdown", "MeshIO", "ModernGL", "Observables", "PrecompileTools", "Printf", "ShaderAbstractions", "StaticArrays"] -git-tree-sha1 = "e53267e2fc64f81b939849ca7bd70d8f879b5293" +git-tree-sha1 = "31571f931b22f0ebb98cace13b74c0d4516c8c2b" uuid = "e9467ef8-e4e7-5192-8a1a-b1aee30e663a" -version = "0.9.5" +version = "0.9.8" [[deps.GPUArrays]] deps = ["Adapt", "GPUArraysCore", "LLVM", "LinearAlgebra", "Printf", "Random", "Reexport", "Serialization", "Statistics"] @@ -662,9 +662,9 @@ version = "1.3.3" [[deps.GeometryBasics]] deps = ["EarCut_jll", "Extents", "GeoInterface", "IterTools", "LinearAlgebra", "StaticArrays", "StructArrays", "Tables"] -git-tree-sha1 = "424a5a6ce7c5d97cca7bcc4eac551b97294c54af" +git-tree-sha1 = "5694b56ccf9d15addedc35e9a4ba9c317721b788" uuid = "5c1252a2-5f33-56bf-86c9-59e7332b4326" -version = "0.4.9" +version = "0.4.10" [[deps.Gettext_jll]] deps = ["Artifacts", "CompilerSupportLibraries_jll", "JLLWrappers", "Libdl", "Libiconv_jll", "Pkg", "XML2_jll"] @@ -761,10 +761,10 @@ uuid = "c817782e-172a-44cc-b673-b171935fbb9e" version = "0.1.7" [[deps.ImageCore]] -deps = ["AbstractFFTs", "ColorVectorSpace", "Colors", "FixedPointNumbers", "MappedArrays", "MosaicViews", "OffsetArrays", "PaddedViews", "PrecompileTools", "Reexport"] -git-tree-sha1 = "fc5d1d3443a124fde6e92d0260cd9e064eba69f8" +deps = ["ColorVectorSpace", "Colors", "FixedPointNumbers", "MappedArrays", "MosaicViews", "OffsetArrays", "PaddedViews", "PrecompileTools", "Reexport"] +git-tree-sha1 = "b2a7eaa169c13f5bcae8131a83bc30eff8f71be0" uuid = "a09fc81d-aa75-5fe9-8630-4744c3626534" -version = "0.10.1" +version = "0.10.2" [[deps.ImageIO]] deps = ["FileIO", "IndirectArrays", "JpegTurbo", "LazyModules", "Netpbm", "OpenEXR", "PNGFiles", "QOI", "Sixel", "TiffImages", "UUIDs"] @@ -902,7 +902,7 @@ version = "1.14.0" [deps.JSON3.weakdeps] ArrowTypes = "31f734f8-188a-4ce0-8406-c8a06bd891cd" - + [[deps.JpegTurbo]] deps = ["CEnum", "FileIO", "ImageCore", "JpegTurbo_jll", "TOML"] git-tree-sha1 = "fa6d0bcff8583bac20f1ffa708c3913ca605c611" @@ -964,9 +964,9 @@ uuid = "dd4b983a-f0e5-5f8d-a1b7-129d4a5fb1ac" version = "2.10.1+0" [[deps.LaTeXStrings]] -git-tree-sha1 = "f2355693d6778a178ade15952b7ac47a4ff97996" +git-tree-sha1 = "50901ebc375ed41dbf8058da26f9de442febbbec" uuid = "b964fa9f-0449-5b57-a5c2-d3ea65f4040f" -version = "1.3.0" +version = "1.3.1" [[deps.LabelledArrays]] deps = ["ArrayInterface", "ChainRulesCore", "ForwardDiff", "LinearAlgebra", "MacroTools", "PreallocationTools", "RecursiveArrayTools", "StaticArrays"] @@ -1143,16 +1143,16 @@ uuid = "1914dd2f-81c6-5fcd-8719-6d5c9610ff09" version = "0.5.10" [[deps.Makie]] -deps = ["Animations", "Base64", "CRC32c", "ColorBrewer", "ColorSchemes", "ColorTypes", "Colors", "Contour", "DelaunayTriangulation", "Distributions", "DocStringExtensions", "Downloads", "FFMPEG_jll", "FileIO", "FilePaths", "FixedPointNumbers", "Formatting", "FreeType", "FreeTypeAbstraction", "GeometryBasics", "GridLayoutBase", "ImageIO", "InteractiveUtils", "IntervalSets", "Isoband", "KernelDensity", "LaTeXStrings", "LinearAlgebra", "MacroTools", "MakieCore", "Markdown", "MathTeXEngine", "Observables", "OffsetArrays", "Packing", "PlotUtils", "PolygonOps", "PrecompileTools", "Printf", "REPL", "Random", "RelocatableFolders", "Scratch", "Setfield", "ShaderAbstractions", "Showoff", "SignedDistanceFields", "SparseArrays", "StableHashTraits", "Statistics", "StatsBase", "StatsFuns", "StructArrays", "TriplotBase", "UnicodeFun"] -git-tree-sha1 = "a37c6610dd20425b131caf65d52abdf859da5ab1" +deps = ["Animations", "Base64", "CRC32c", "ColorBrewer", "ColorSchemes", "ColorTypes", "Colors", "Contour", "DelaunayTriangulation", "Distributions", "DocStringExtensions", "Downloads", "FFMPEG_jll", "FileIO", "FilePaths", "FixedPointNumbers", "Formatting", "FreeType", "FreeTypeAbstraction", "GeometryBasics", "GridLayoutBase", "ImageIO", "InteractiveUtils", "IntervalArithmetic", "IntervalSets", "Isoband", "KernelDensity", "LaTeXStrings", "LinearAlgebra", "MacroTools", "MakieCore", "Markdown", "MathTeXEngine", "Observables", "OffsetArrays", "Packing", "PlotUtils", "PolygonOps", "PrecompileTools", "Printf", "REPL", "Random", "RelocatableFolders", "Scratch", "ShaderAbstractions", "Showoff", "SignedDistanceFields", "SparseArrays", "StableHashTraits", "Statistics", "StatsBase", "StatsFuns", "StructArrays", "TriplotBase", "UnicodeFun"] +git-tree-sha1 = "40c5dfbb99c91835171536cd571fe6f1ba18ff97" uuid = "ee78f7c6-11fb-53f2-987a-cfe4a2b5a57a" -version = "0.20.4" +version = "0.20.7" [[deps.MakieCore]] deps = ["Observables", "REPL"] -git-tree-sha1 = "ec5db7bb2dc9b85072658dcb2d3ad09569b09ac9" +git-tree-sha1 = "248b7a4be0f92b497f7a331aed02c1e9a878f46b" uuid = "20f20a25-4f0e-4fdf-b5d1-57303727442b" -version = "0.7.2" +version = "0.7.3" [[deps.MappedArrays]] git-tree-sha1 = "2dab0221fe2b0f2cb6754eaa743cc266339f527e" @@ -1193,9 +1193,9 @@ version = "0.3.2" [[deps.MeshIO]] deps = ["ColorTypes", "FileIO", "GeometryBasics", "Printf"] -git-tree-sha1 = "8be09d84a2d597c7c0c34d7d604c039c9763e48c" +git-tree-sha1 = "8c26ab950860dfca6767f2bbd90fdf1e8ddc678b" uuid = "7269a6da-0436-5bbc-96c2-40638cbb6118" -version = "0.4.10" +version = "0.4.11" [[deps.Missings]] deps = ["DataAPI"] @@ -1232,6 +1232,11 @@ git-tree-sha1 = "8d852646862c96e226367ad10c8af56099b4047e" uuid = "3b2b4ff1-bcff-5658-a3ee-dbcf1ce5ac09" version = "0.4.4" +[[deps.Multisets]] +git-tree-sha1 = "8d852646862c96e226367ad10c8af56099b4047e" +uuid = "3b2b4ff1-bcff-5658-a3ee-dbcf1ce5ac09" +version = "0.4.4" + [[deps.MultivariatePolynomials]] deps = ["ChainRulesCore", "DataStructures", "LinearAlgebra", "MutableArithmetics"] git-tree-sha1 = "f9978f23952b52b8d958b72f8b5368f84254dc02" @@ -1315,6 +1320,18 @@ git-tree-sha1 = "a4ca623df1ae99d09bc9868b008262d0c0ac1e4f" uuid = "18a262bb-aa17-5467-a713-aee519bc75cb" version = "3.1.4+0" +[[deps.OpenEXR]] +deps = ["Colors", "FileIO", "OpenEXR_jll"] +git-tree-sha1 = "327f53360fdb54df7ecd01e96ef1983536d1e633" +uuid = "52e1d378-f018-4a11-a4be-720524705ac7" +version = "0.3.2" + +[[deps.OpenEXR_jll]] +deps = ["Artifacts", "Imath_jll", "JLLWrappers", "Libdl", "Zlib_jll"] +git-tree-sha1 = "a4ca623df1ae99d09bc9868b008262d0c0ac1e4f" +uuid = "18a262bb-aa17-5467-a713-aee519bc75cb" +version = "3.1.4+0" + [[deps.OpenLibm_jll]] deps = ["Artifacts", "Libdl"] uuid = "05823500-19ac-5b8b-9628-191a04bc5112" @@ -1339,10 +1356,10 @@ uuid = "efe28fd5-8261-553b-a9e1-b2916fc3738e" version = "0.5.5+0" [[deps.Optim]] -deps = ["Compat", "FillArrays", "ForwardDiff", "LineSearches", "LinearAlgebra", "NLSolversBase", "NaNMath", "Parameters", "PositiveFactorizations", "Printf", "SparseArrays", "StatsBase"] -git-tree-sha1 = "01f85d9269b13fedc61e63cc72ee2213565f7a72" +deps = ["Compat", "FillArrays", "ForwardDiff", "LineSearches", "LinearAlgebra", "MathOptInterface", "NLSolversBase", "NaNMath", "Parameters", "PositiveFactorizations", "Printf", "SparseArrays", "StatsBase"] +git-tree-sha1 = "d024bfb56144d947d4fafcd9cb5cafbe3410b133" uuid = "429524aa-4258-5aef-a3af-852621145aeb" -version = "1.7.8" +version = "1.9.2" [[deps.Opus_jll]] deps = ["Artifacts", "JLLWrappers", "Libdl", "Pkg"] @@ -1448,6 +1465,12 @@ git-tree-sha1 = "f9501cc0430a26bc3d156ae1b5b0c1b47af4d6da" uuid = "eebad327-c553-4316-9ea0-9fa01ccd7688" version = "0.3.3" +[[deps.PkgVersion]] +deps = ["Pkg"] +git-tree-sha1 = "f9501cc0430a26bc3d156ae1b5b0c1b47af4d6da" +uuid = "eebad327-c553-4316-9ea0-9fa01ccd7688" +version = "0.3.3" + [[deps.PlotThemes]] deps = ["PlotUtils", "Statistics"] git-tree-sha1 = "1f03a2d339f42dca4a4da149c7e15e9b896ad899" @@ -1761,9 +1784,9 @@ version = "1.1.1" [[deps.ShaderAbstractions]] deps = ["ColorTypes", "FixedPointNumbers", "GeometryBasics", "LinearAlgebra", "Observables", "StaticArrays", "StructArrays", "Tables"] -git-tree-sha1 = "db0219befe4507878b1a90e07820fed3e62c289d" +git-tree-sha1 = "79123bc60c5507f035e6d1d9e563bb2971954ec8" uuid = "65257c39-d410-5151-9873-9b3e5be5013e" -version = "0.4.0" +version = "0.4.1" [[deps.SharedArrays]] deps = ["Distributed", "Mmap", "Random", "Serialization"] @@ -1794,9 +1817,9 @@ version = "0.8.6" [[deps.SimplePartitions]] deps = ["AbstractLattices", "DataStructures", "Permutations"] -git-tree-sha1 = "e9330391d04241eafdc358713b48396619c83bcb" +git-tree-sha1 = "e182b9e5afb194142d4668536345a365ea19363a" uuid = "ec83eff0-a5b5-5643-ae32-5cbf6eedec9d" -version = "0.3.1" +version = "0.3.2" [[deps.SimplePolynomials]] deps = ["Mods", "Multisets", "Polynomials", "Primes"] @@ -1866,9 +1889,9 @@ weakdeps = ["ChainRulesCore"] [[deps.StableHashTraits]] deps = ["Compat", "PikaParser", "SHA", "Tables", "TupleTools"] -git-tree-sha1 = "662f56ffe22b3985f3be7474f0aecbaf214ecf0f" +git-tree-sha1 = "10dc702932fe05a0e09b8e5955f00794ea1e8b12" uuid = "c5dd0088-6c3f-4803-b00e-f31a60c170fa" -version = "1.1.6" +version = "1.1.8" [[deps.StackViews]] deps = ["OffsetArrays"] @@ -2052,9 +2075,9 @@ uuid = "781d530d-4396-4725-bb49-402e4bee1e77" version = "1.4.0" [[deps.TupleTools]] -git-tree-sha1 = "155515ed4c4236db30049ac1495e2969cc06be9d" +git-tree-sha1 = "41d61b1c545b06279871ef1a4b5fcb2cac2191cd" uuid = "9d95972d-f1c8-5527-a6e0-b4b365fa01f6" -version = "1.4.3" +version = "1.5.0" [[deps.URIs]] git-tree-sha1 = "074f993b0ca030848b897beff716d93aca60f06a" diff --git a/Project.toml b/Project.toml index 59e7476..6d587bf 100644 --- a/Project.toml +++ b/Project.toml @@ -11,6 +11,7 @@ DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0" FileIO = "5789e2e9-d7fb-5bc7-8068-2c6fae9b9549" GLMakie = "e9467ef8-e4e7-5192-8a1a-b1aee30e663a" IJulia = "7073ff75-c697-5162-941a-fcdaad2a7d2a" +LaTeXStrings = "b964fa9f-0449-5b57-a5c2-d3ea65f4040f" JSON3 = "0f8b85d8-7281-11e9-16c2-39a750bddbf1" ParametricMCPs = "9b992ff8-05bb-4ea1-b9d2-5ef72d82f7ad" PlotlyJS = "f0f68f2c-4968-5e81-91da-67840de0976a" diff --git a/experiments/plot_activation.jl b/experiments/plot_activation.jl deleted file mode 100644 index 8412e2f..0000000 --- a/experiments/plot_activation.jl +++ /dev/null @@ -1,15 +0,0 @@ -using Plots - -function f(δ, k) - return 1/(1 + exp(-2 * δ * k)) -end - -k = 1:1:10 -δ = -2:0.05:2 - -p = plot() -for i in 1:length(k) - plot!(p, δ, f.(δ, k[i]), label="k = $(k[i])", linewidth=7) -end -plot!(p, size=(800, 600), title="Logistic function", textfontsize=20, legendfontsize=20, tickfontsize=10, legend=:topleft, titlefontsize=20, labelfontsize = 20, xlabel="δ", ylabel="f(δ, k)") -display(p) \ No newline at end of file diff --git a/experiments/tower_defense.jl b/experiments/tower_defense.jl index caa47fe..9f907e1 100644 --- a/experiments/tower_defense.jl +++ b/experiments/tower_defense.jl @@ -3,25 +3,83 @@ using BlockArrays using LinearAlgebra: norm_sqr, norm using Zygote using Colors -using GLMakie: Figure, Axis, Colorbar, heatmap!, text!, surface!, scatter!, Axis3, save +using GLMakie: + Figure, + Axis, + Colorbar, + heatmap!, + text!, + surface!, + scatter!, + Axis3, + save, + image, + DataAspect, + rotr90, + hidedecorations!, + hidezdecorations! +record, empty!, resize_to_layout! +using FileIO +using LaTeXStrings + +# -------------------------------------------------------------------------------------------------------------------------------- +# ------------------------------------------------------- INTRO/GUIDE ------------------------------------------------------------ +# -------------------------------------------------------------------------------------------------------------------------------- + +# This script contains the code for the tower defense game. It is organized as follows: +# 1. Costs: Contains the cost functions for the tower defense game +# 2. Stage 1 Solver: Contains the code for solving the Stage 1 optimization problem +# 3. Stage 2 Solver: Contains the code for solving the Stage 2 optimization problem +# 4. Visualization: Contains the code for visualizing the results of the optimization problem +# In this problem we assume that there are 3 worlds and 3 signals. In Stage 2 there are 3 possible directions of attack + +# Nomenclature for tower defense game +# N : Number of worlds (=3) +# ps = [P(w₁),..., P(wₙ)] : prior distribution of k worlds for each signal, nx1 vector +# βs : vector containing P2's cost parameters for each world. vector of nx1 vectors +# x[Block(1)] : u(0), P1's action given signal s¹=0 depends on r +# x[Block(2)] : u(1), P1's action given signal s¹=1 +# x[Block(3)] : u(2), P1's action given signal s¹=2 +# x[Block(4)] : u(3), P1's action given signal s¹=3 +# x[Block(5)] ~ x[Block(7)] : v(wₖ, 0), P2's action for each worlds given signal s¹=0 depends on r +# x[Block(8)] : v(wₖ, 1), P2's action for world 1 given signal s¹=1 +# x[Block(9)] : v(wₖ, 2), P2's action for world 2 given signal s¹=2 +# x[Block(10)] : v(wₖ, 3), P2's action for world 3 given signal s¹=3 +# θ = rₖ = [r₁, ... , rₙ] : r, Scout allocation in each direction +# J : Stage 1's objective function + +# -------------------------------------------------------------------------------------------------------------------------------- +# ------------------------------------------------------- TOWER DEFENSE COSTS----------------------------------------------------- +# -------------------------------------------------------------------------------------------------------------------------------- -using Infiltrator +"Defender cost function" +function J_1(u, v, β) + -J_2(u, v, β) +end + +""" +Attacker cost function. Sigmoidal cost function with force multipliers -""" Nomenclature - N : Number of worlds (=3) - ps = [P(w₁),..., P(wₙ)] : prior distribution of k worlds for each signal, nx1 vector - βs : vector containing P2's cost parameters for each world. vector of nx1 vectors - x[Block(1)] : u(0), P1's action given signal s¹=0 depends on r - x[Block(2)] : u(1), P1's action given signal s¹=1 - x[Block(3)] : u(2), P1's action given signal s¹=2 - x[Block(4)] : u(3), P1's action given signal s¹=3 - x[Block(5)] ~ x[Block(7)] : v(wₖ, 0), P2's action for each worlds given signal s¹=0 depends on r - x[Block(8)] : v(wₖ, 1), P2's action for world 1 given signal s¹=1 - x[Block(9)] : v(wₖ, 2), P2's action for world 2 given signal s¹=2 - x[Block(10)] : v(wₖ, 3), P2's action for world 3 given signal s¹=3 - θ = rₖ = [r₁, ... , rₙ] : r, Scout allocation in each direction - J : Stage 1's objective function +Inputs: + u: vector containing P1's (defender) strategy for each world. + v: vector containing P2's (attacker) strategy for each world. + β: vector containing P2's (attacker) preference parameters for each world. +Outputs + J_2: Attacker's cost function """ +function J_2(u, v, β) + δ = [β[ii] * v[ii] - u[ii] for ii in eachindex(β)] + -sum([activate(δ[j]) * (β[j] * v[j] - u[j])^2 for j in eachindex(β)]) +end + +"Activation function for attacker cost function" +function activate(δ; k = 10.0) + return 1 / (1 + exp(-2 * δ * k)) +end + +# -------------------------------------------------------------------------------------------------------------------------------- +# ------------------------------------------------------- STAGE 1 SOLVER --------------------------------------------------------- +# -------------------------------------------------------------------------------------------------------------------------------- """ Solve Stage 1 to find optimal scout allocation r. @@ -32,6 +90,8 @@ Inputs: r_init: initial guess scout allocation Outputs: r: optimal scout allocation + +TODO: Instead of using one large game, use the complete/incomplete information games """ function solve_r( ps, @@ -92,14 +152,426 @@ function solve_r( end """ -Temp. script to calculate and plot heatmap of Stage 1 cost function +Project onto simplex using Fig. 1 Duchi 2008 """ -function run_visualization(;βs =nothing, save_name="") - dr = 0.01 - ps = [1/3, 1 / 3, 1 / 3] - if βs == nothing - βs = [[4.,2.,2.], [2., 3., 2.], [2., 2., 3.]] +function project_onto_simplex(v; z = 1.0) + μ = sort(v, rev = true) + ρ = findfirst([μ[j] - 1 / j * (sum(μ[1:j]) - z) <= 0 for j in eachindex(v)]) + ρ = isnothing(ρ) ? length(v) : ρ - 1 + θ = 1 / ρ * (sum(μ[1:ρ]) - z) + return [maximum([v[i] - θ, 0]) for i in eachindex(v)] +end + +""" +Compute derivative of Stage 1's objective function w.r.t. x +""" +function compute_dKdx(r, x, ps, βs) + gradient(x -> compute_K(r, x, ps, βs), x)[1] +end + +""" +Compute full derivative of Stage 1's objective function w.r.t. r + +Inputs: + x: decision variables of Stage 2 + ps: prior distribution of k worlds, nx1 vector + +Outputs: + djdq: Jacobian of Stage 1's objective function w.r.t. r +""" +function compute_dKdr(r, x, ps, βs, game) + dKdx = compute_dKdx(r, x, ps, βs) + dKdr = gradient(r -> compute_K(r, x, ps, βs), r)[1] + dxdr = compute_dxdr(r, x, ps, βs, game) + n = length(ps) + for idx in 1:(1 + n^2) + dKdr += (dKdx[Block(idx)]' * dxdr[Block(idx)])' + end + dKdr +end + +""" +Solve stage 2 and return full derivative of objective function w.r.t. r + +Inputs: + r: scout allocation + ps: prior distribution of k worlds, nx1 vector + βs: vector containing P2's cost parameters for each world. vector of nx1 vectors + +Outputs: + dxdr: Blocked Jacobian of Stage 2's decision variables w.r.t. Stage 1's decision variable +""" +function compute_dxdr(r, x, ps, βs, game; verbose = false) + n = length(ps) + n_players = 1 + n^2 + var_dim = n + + # Return Jacobian + dxdr = jacobian( + r -> solve( + game, + r; + initial_guess = vcat(x, zeros(total_dim(game) - n_players * var_dim)), + verbose = false, + return_primals = false, + ).variables[1:(n_players * var_dim)], + r, + )[1] + + BlockArray(dxdr, [var_dim for _ in 1:n_players], [var_dim]) +end + +# -------------------------------------------------------------------------------------------------------------------------------- +# ------------------------------------------------------- STAGE 2 SOLVER --------------------------------------------------------- +# -------------------------------------------------------------------------------------------------------------------------------- + +""" +Build parametric game for Stage 2. One single large game. + +Inputs: + ps: prior distribution of k worlds for each signal, nx1 vector + βs: vector containing P1's cost parameters for each world. vector of nx1 vectors +Outputs: + parametric_game: ParametricGame object + fs: vector of symbolic expressions for each player's objective function + +""" +function build_stage_2(ps, βs) + n = length(ps) # assume n_signals = n_worlds + 1 + n_players = 1 + n^2 + + # Define Bayesian game player costs in Stage 2 + p_w_k_0(w_idx, θ) = (1 - θ[w_idx]) * ps[w_idx] / (1 - θ' * ps) + fs = [ + (x, θ) -> sum([ + J_1(x[Block(1)], x[Block(w_idx + n + 1)], βs[w_idx]) * p_w_k_0(w_idx, θ) for + w_idx in 1:n + ]), # u|s¹=0 IPI + [ + (x, θ) -> J_1(x[Block(w_idx + 1)], x[Block(w_idx + 2 * n + 1)], βs[w_idx]) for + w_idx in 1:n + ]..., # u|s¹={1,2,3} PI + [(x, θ) -> J_2(x[Block(1)], x[Block(w_idx + n + 1)], βs[w_idx]) for w_idx in 1:n]..., # v|s¹=0 IPI + [ + (x, θ) -> J_2(x[Block(w_idx + 1)], x[Block(w_idx + 2 * n + 1)], βs[w_idx]) for + w_idx in 1:n + ]..., # v|s¹={1,2,3} PI + ] + + # equality constraints + gs = [(x, θ) -> [sum(x[Block(i)]) - 1] for i in 1:n_players] # Everyone must attack/defend + + # inequality constraints + hs = [(x, θ) -> x[Block(i)] for i in 1:n_players] # All vars must be non-negative + + # shared constraints + g̃ = (x, θ) -> [0] + h̃ = (x, θ) -> [0] + + ParametricGame(; + objectives = fs, + equality_constraints = gs, + inequality_constraints = hs, + shared_equality_constraint = g̃, + shared_inequality_constraint = h̃, + parameter_dimension = 3, + primal_dimensions = [3 for _ in 1:n_players], + equality_dimensions = [1 for _ in 1:n_players], + inequality_dimensions = [3 for _ in 1:n_players], + shared_equality_dimension = 1, + shared_inequality_dimension = 1, + ), + fs +end + +""" +Build complete information parametric for Stage 2. Assumes 2 players, 3 signals, 3 worlds. +""" +function build_complete_info_game() + fs = [ + (x, θ) -> J_1(x[Block(1)], x[Block(2)], θ) + (x, θ) -> J_2(x[Block(1)], x[Block(2)], θ) + ] + gs = [(x, θ) -> [sum(x[Block(i)]) - 1] for i in 1:2] + hs = [(x, θ) -> x[Block(i)] for i in 1:2] + g̃ = (x, θ) -> [0] + h̃ = (x, θ) -> [0] + + ParametricGame(; + objectives = fs, + equality_constraints = gs, + inequality_constraints = hs, + shared_equality_constraint = g̃, + shared_inequality_constraint = h̃, + parameter_dimension = 3, + primal_dimensions = [3, 3], + equality_dimensions = [1, 1], + inequality_dimensions = [3, 3], + shared_equality_dimension = 1, + shared_inequality_dimension = 1, + ) +end + +""" +Build incomplete information parametric for Stage 2. Assumes 2 players, 3 signals, 3 worlds. +""" +function build_incomplete_info_game(ps, βs) + n = length(ps)# assume n_signals = n_worlds + 1 + n_players = 1 + n + + p_w_k_0(w_idx, θ) = (1 - θ[w_idx]) * ps[w_idx] / (1 - θ' * ps) + fs = [ + (x, θ) -> sum([ + p_w_k_0(w_idx, θ) * J_1(x[Block(1)], x[Block(w_idx + 1)], βs[w_idx]) for w_idx in 1:n + ]), # x^1(0, i) + [(x, θ) -> J_2(x[Block(1)], x[Block(w_idx + 1)], βs[w_idx]) for w_idx in 1:n]..., + ] + gs = [(x, θ) -> [sum(x[Block(i)]) - 1] for i in 1:n_players] + hs = [(x, θ) -> x[Block(i)] for i in 1:n_players] + g̃ = (x, θ) -> [0] + h̃ = (x, θ) -> [0] + + ParametricGame(; + objectives = fs, + equality_constraints = gs, + inequality_constraints = hs, + shared_equality_constraint = g̃, + shared_inequality_constraint = h̃, + parameter_dimension = 3, + primal_dimensions = [3 for _ in 1:n_players], + equality_dimensions = [1 for _ in 1:n_players], + inequality_dimensions = [3 for _ in 1:n_players], + shared_equality_dimension = 1, + shared_inequality_dimension = 1, + ) +end + +""" +Compute Stage 1 objective + +Inputs: + r: scout allocation + x: decision variables of Stage 2 + ps: prior distribution of k worlds, nx1 vector + βs: vector containing P2's cost parameters for each world. vector of nx1 vectors +Output: + K: Stage 1's objective function value +""" +function compute_K(r, x, ps, βs) + n = length(ps) + sum([(1 - r[j]) * ps[j] * J_1(x[Block(1)], x[Block(j + n + 1)], βs[j]) for j in 1:n]) + sum([r[j] * ps[j] * J_1(x[Block(j + 1)], x[Block(j + 2 * n + 1)], βs[j]) for j in 1:n]) +end + +""" +Compute incomplete information cost term for a single world +Inputs: + r: scout allocation + ps: prior distribution of k worlds, nx1 vector + r_i: scout allocation for world i + x_1_0: defender's decision for signal 0 + x_2_0_i: attacker's decision for signal 0 and world i + ps_i: prior distribution of world i + βs_i: P2's cost parameters for world i +Outputs: + cost_term: incomplete info cost term for world i +""" +function compute_incomplete_info_cost_term_i(r, ps, r_i, x_1_0, x_2_0_i, ps_i, βs_i) + (1 - r_i) * ps_i / (1 - r' * ps) * J_1(x_1_0, x_2_0_i, βs_i) +end + +function compute_P1_incomplete_info_cost(r, x_1_0, x_2_0s, ps, βs) + n = length(ps) + sum([ + compute_incomplete_info_cost_term_i(r, ps, r[i], x_1_0, x_2_0s[Block(i)], ps[i], βs[i]) for + i in 1:n + ]) +end + +""" +Compute Stage 2 decision variables given r using PATH + +Input: + r: scout allocation + ps: prior distribution of k worlds, nx1 vector +Output: + x: decision variables of Stage 2 given r. BlockedArray with a block per player +""" +function compute_stage_2( + r, + ps, + βs, + complete_info_game, + incomplete_info_game; + initial_guess = nothing, + verbose = false, +) + num_worlds = length(ps) # assume n_signals = n_worlds + 1 + n_players = 1 + num_worlds^2 + var_dim = num_worlds # TODO: Change this to be more general + + solution_complete = [ + solve( + complete_info_game, + β; + initial_guess = isnothing(initial_guess) ? + 1 / 3 * ones(total_dim(complete_info_game)) : initial_guess, + verbose, + return_primals = true, + ) for β in βs + ] + + solution_incomplete = solve( + incomplete_info_game, + r; + initial_guess = isnothing(initial_guess) ? + 1 / 3 * ones(total_dim(incomplete_info_game)) : initial_guess, + verbose, + return_primals = true, + ) + + return BlockArray( + vcat( + solution_incomplete.variables[1:var_dim], + [solution_complete[i].variables[1:var_dim] for i in 1:num_worlds]..., + solution_incomplete.variables[(var_dim + 1):((num_worlds + 1) * var_dim)], + [solution_complete[i].variables[(var_dim + 1):(2 * var_dim)] for i in 1:num_worlds]..., + ), + [var_dim for _ in 1:n_players], + ) +end + +struct IBRGameSolver end + +""" +Compute Stage 2 decision variables using Iterative Best Response + +Inputs: + r: scout allocation + ps: prior distribution of k worlds, nx1 vector + βs: vector containing P2's cost parameters for each world. vector of nx1 vectors + Js: vector containing P1 and P2's cost functions +Outputs: + x: decision variables of Stage 2 given r. BlockedArray with a block per player +""" +function compute_stage_2( + ::IBRGameSolver, + r, + ps, + βs, + Js; + initial_guess = nothing, + max_ibr_rounds = 100, + ibr_convergence_tolerance = 1e-3, + verbose = false, +) + num_worlds = length(ps) # assume n_signals = n_worlds + 1 + total_num_vars = num_worlds + 1 + 2 * num_worlds + if isnothing(initial_guess) + x = 1 / 3 * ones((num_worlds + 1) * num_worlds + 2 * num_worlds^2) + else + x = initial_guess + end + x = BlockArray(x, [num_worlds for _ in 1:total_num_vars]) + + # Solve complete information games + for world_idx in 1:num_worlds + β = βs[world_idx] + x_1 = x[Block(1 + world_idx)] + x_2 = x[Block(1 + 2 * num_worlds + world_idx)] + for i_ibr in 1:max_ibr_rounds + last_solution = [x_1, x_2] + x_1 = gradient_play(x_1 -> Js[1](x_1, x_2, β), x_1; verbose) + x_2 = gradient_play(x_2 -> Js[2](x_1, x_2, β), x_2; verbose) + converged = norm([x_1, x_2] - last_solution) < ibr_convergence_tolerance + if converged + verbose && + @info "World $world_idx complete info. game converged after $i_ibr IBR iterations" + break + end + end + x[Block(1 + world_idx)] = x_1 + x[Block(1 + 2 * num_worlds + world_idx)] = x_2 + end + + # Solve incomplete information game + x_1_0 = x[Block(1)] + x_2_0s = BlockArray( + x[((1 + num_worlds) * num_worlds + 1):((1 + num_worlds) * num_worlds + num_worlds * num_worlds)], # P2, s = 0 + [num_worlds for _ in 1:num_worlds], + ) + for i_ibr in 1:max_ibr_rounds + last_solution = vcat(x_1_0, x_2_0s) + x_1_0 = gradient_play( + x_1_0 -> compute_P1_incomplete_info_cost(r, x_1_0, x_2_0s, ps, βs), + x_1_0; + verbose, + ) + for world_idx in 1:num_worlds + x_2_0s[Block(world_idx)] = gradient_play( + x_2_0s -> Js[2](x_1_0, x_2_0s, βs[world_idx]), + x_2_0s[Block(world_idx)]; + verbose, + ) + end + converged = norm(vcat(x_1_0, x_2_0s) - last_solution) < ibr_convergence_tolerance + if converged + verbose && @info "Incomplete complete info. game converged after $i_ibr IBR iterations" + break + end + end + x[Block(1)] = x_1_0 + for world_idx in 1:num_worlds + x[Block(1 + num_worlds + world_idx)] = x_2_0s[Block(world_idx)] end + return x +end + +""" +Gradient descent while projecting onto the simplex + +Input: + cost_function: function to minimize + x: initial guess +Output: + x: minimizer of cost_function +""" +function gradient_play( + cost_function, + x; + max_iter = 200, + α = 0.05, + tol = 1e-3, + verbose = false, + text = nothing, +) + iter = 0 + x_prev = x + while iter < max_iter + x_prev = x + dJdx = gradient(x -> cost_function(x), x)[1] + x_temp = x - α .* dJdx + x = project_onto_simplex(x_temp) + iter += 1 + if (norm(x - x_prev) < tol) + verbose && @info " Gradient descent converged after $iter iterations" + return x + end + end + @warn " Gradient descent did not converge after $max_iter iterations" + return x +end + +# -------------------------------------------------------------------------------------------------------------------------------- +# ------------------------------------------------------- VISUALIZATION ---------------------------------------------------------- +# -------------------------------------------------------------------------------------------------------------------------------- + +""" +Visualization. Calculate and plot Stage 1 cost as a function of r for a given prior distribution and attacker preference. +""" +function run_visualization() + dr = 0.05 + ps = [1 / 3, 1 / 3, 1 / 3] + βs = [[3.0, 2.0, 2.0], [2.0, 3.0, 2.0], [2.0, 2.0, 3.0]] Ks = calculate_stage_1_costs(ps, βs; dr) fig = display_surface(ps, Ks) if save_name !== "" @@ -170,32 +642,30 @@ function visualize_decisions(world_idx;r=[1.,0.,0.],βs =nothing, save_name="") end """ -Temp. script to calculate and plot surfaces for the terms in Stage 1's cost function +Visualization. Calculate and plot all terms in the Stage 1 cost as a function of r for a given prior distribution and attacker preference. +Assumes number of worlds and signals is 3. """ -function run_stage_1_breakout(;display_controls = 0, dr = 0.05, cost_player = 1, βs = nothing,save_prefix="") - # dr = 0.05 - ps = [1/3, 1/3, 1/3] - - if βs == nothing - βs = [ - [6.0, 2.0, 2.0], - [2.0, 3., 2.0], - [2.0, 2.0, 3.0] - ] - end - #### Choose the initial guess for Stage 2 initialization - primal_guess = (1/3)*ones(30) ## Initialization frorm primes - initial_guess = vcat(primal_guess,(1/3)*ones(42)) ## concatenate, assume duals are 1/3 - - - - if (display_controls in [1,2]) - world_1_misid_costs, world_1_misid_controls = calculate_misid_costs(ps, βs, 1; dr, return_controls=display_controls, initial_guess=initial_guess, cost_player=cost_player) - world_2_misid_costs, world_2_misid_controls = calculate_misid_costs(ps, βs, 2; dr, return_controls=display_controls, initial_guess=initial_guess, cost_player=cost_player) - world_3_misid_costs, world_3_misid_controls = calculate_misid_costs(ps, βs, 3; dr, return_controls=display_controls, initial_guess=initial_guess, cost_player=cost_player) - world_1_id_costs, world_1_id_controls = calculate_id_costs(ps, βs, 1; dr, return_controls=display_controls, initial_guess=initial_guess, cost_player=cost_player) - world_2_id_costs, world_2_id_controls = calculate_id_costs(ps, βs, 2; dr, return_controls=display_controls, initial_guess=initial_guess, cost_player=cost_player) - world_3_id_costs, world_3_id_controls = calculate_id_costs(ps, βs, 3; dr, return_controls=display_controls, initial_guess=initial_guess, cost_player=cost_player) +function run_stage_1_breakout(; + display_controls = 0, + dr = 0.05, + βs = [[3.0, 2.0, 2.0], [2.0, 3.0, 2.0], [2.0, 2.0, 3.0]], + ps = [1 / 3, 1 / 3, 1 / 3], +) + if (display_controls in [1, 2]) + println("Calculating misid. costs") + world_1_misid_costs, world_1_misid_controls = + calculate_misid_costs(ps, βs, 1; dr, return_controls = display_controls) + world_2_misid_costs, world_2_misid_controls = + calculate_misid_costs(ps, βs, 2; dr, return_controls = display_controls) + world_3_misid_costs, world_3_misid_controls = + calculate_misid_costs(ps, βs, 3; dr, return_controls = display_controls) + println("Calculating id costs") + world_1_id_costs, world_1_id_controls = + calculate_id_costs(ps, βs, 1; dr, return_controls = display_controls) + world_2_id_costs, world_2_id_controls = + calculate_id_costs(ps, βs, 2; dr, return_controls = display_controls) + world_3_id_costs, world_3_id_controls = + calculate_id_costs(ps, βs, 3; dr, return_controls = display_controls) else world_1_misid_costs = calculate_misid_costs(ps, βs, 1; dr, initial_guess=initial_guess, cost_player=cost_player) world_2_misid_costs = calculate_misid_costs(ps, βs, 2; dr, initial_guess=initial_guess, cost_player=cost_player) @@ -205,24 +675,19 @@ function run_stage_1_breakout(;display_controls = 0, dr = 0.05, cost_player = 1, world_3_id_costs = calculate_id_costs(ps, βs, 3; dr, initial_guess=initial_guess, cost_player=cost_player) end # Normalize using maximum value across all worlds - - maxormin = cost_player == 2 ? minimum : maximum - - max_value = - maxormin( - filter( - !isnan, - vcat( - world_1_misid_costs, - world_2_misid_costs, - world_3_misid_costs, - world_1_id_costs, - world_2_id_costs, - world_3_id_costs, - ), + max_value = maximum( + filter( + !isnan, + vcat( + world_1_misid_costs, + world_2_misid_costs, + world_3_misid_costs, + world_1_id_costs, + world_2_id_costs, + world_3_id_costs, ), - ) - max_value = (-1)^(cost_player+1)*max_value + ), + ) world_1_misid_costs = [isnan(c) ? NaN : c / max_value for c in world_1_misid_costs] world_2_misid_costs = [isnan(c) ? NaN : c / max_value for c in world_2_misid_costs] world_3_misid_costs = [isnan(c) ? NaN : c / max_value for c in world_3_misid_costs] @@ -230,8 +695,9 @@ function run_stage_1_breakout(;display_controls = 0, dr = 0.05, cost_player = 1, world_2_id_costs = [isnan(c) ? NaN : c / max_value for c in world_2_id_costs] world_3_id_costs = [isnan(c) ? NaN : c / max_value for c in world_3_id_costs] - if (display_controls in [1,2]) - display_stage_1_costs_controls( + fig = nothing + if (display_controls in [1, 2]) + fig = display_stage_1_costs_controls( [ world_1_id_costs, world_2_id_costs, @@ -248,12 +714,10 @@ function run_stage_1_breakout(;display_controls = 0, dr = 0.05, cost_player = 1, world_2_misid_controls, world_3_misid_controls, ], - ps, - save_file=save_prefix*"P"*string(display_controls)*"_", - cost_player=cost_player + ps; ) else - display_stage_1_costs( + fig = display_stage_1_costs( [ world_1_id_costs, world_2_id_costs, @@ -262,65 +726,89 @@ function run_stage_1_breakout(;display_controls = 0, dr = 0.05, cost_player = 1, world_2_misid_costs, world_3_misid_costs, ], - ps, - ) - end - -end - -function run_residuals() - dr = 0.01 - ps = [1/3, 1/3, 1/3] - βs = [ - [4.0, 2.0, 2.0], - [2.0, 4.0, 2.0], - [2.0, 2.0, 4.0] - ] - world_1_residuals = calculate_residuals(ps, βs, 1; dr) - world_2_residuals = calculate_residuals(ps, βs, 2; dr) - world_3_residuals = calculate_residuals(ps, βs, 3; dr) - - display_residuals( - [ - world_1_residuals, - world_2_residuals, - world_3_residuals, - ], - ps, - ) -end - -function calculate_residuals(ps, βs, world_idx; dr = 0.05) - @assert sum(ps) ≈ 1.0 "Prior distribution ps must be a probability distribution" - game, _ = build_stage_2(ps, βs) - rs = 0:dr:1 - num_worlds = length(ps) - residuals = NaN * ones(Float64, Int(1 / dr + 1), Int(1 / dr + 1)) - for (i, r1) in enumerate(rs) - for (j, r2) in enumerate(rs) - if r1 + r2 > 1 - continue - end - r3 = 1 - r1 - r2 - r = [r1, r2, r3] - _, residual = compute_stage_2(r, ps, βs, game; return_residual = true) - residuals[i, j] = residual - end + ps, + ) end + return fig +end + +""" +Visualization. Run sweep over a set of perturbations for the attacker's cost functions. +""" +function run_sweep(perturbations, k, perturbation_type; dr = 0.05) + ps = [1 / 3, 1 / 3, 1 / 3] + fig = Figure(size = (1300, 800)) + for perturbation in perturbations + βs = [[3.0 + perturbation, 2.0, 2.0], [2.0, 3.0, 2.0], [2.0, 2.0, 3.0]] + # βs = [ + # [2.0 + perturbation, 2.0, 2.0], + # [2.0, 2.0 + perturbation, 2.0], + # [2.0, 2.0, 2.0 + perturbation] + # ] + Ks = calculate_stage_1_costs(ps, βs; dr) + + # Nasty but gets the job done + fig = Figure(size = (1300, 800)) + + run_stage_1_breakout(display_controls = 1, dr = dr, βs = βs, ps = ps) + defender_controls = load("figures/stage_1_controls.png") + image( + fig[1, 1], + rotr90(defender_controls), + axis = (aspect = DataAspect(), title = "defender"), + ) + hidedecorations!(fig.content[1]) + + run_stage_1_breakout(display_controls = 2, dr = dr, βs = βs, ps = ps) + attacker_controls = load("figures/stage_1_controls.png") + image( + fig[1, 2], + rotr90(attacker_controls), + axis = (aspect = DataAspect(), title = "attacker"), + ) + hidedecorations!(fig.content[2]) + + display_surface(ps, Ks) + stage_1_surface = load("figures/stage_1_surface.png") + image(fig[2, 2], rotr90(stage_1_surface), axis = (aspect = DataAspect(), title = "stage 1")) + hidedecorations!(fig.content[3]) + + Axis( + fig[2, 1], + aspect = DataAspect(), + title = perturbation_type * " \n perturbation: $perturbation \n k = $k", + backgroundcolor = :gray50, + ) + # hidedecorations!(fig.content[4]) + + save("figures/sweep/sweep_$(perturbation_type)_s$(perturbation)_k$(k).png", fig) - return residuals + # Show the figure + fig + end end -function calculate_id_costs(ps, βs, world_idx; dr = 0.05, return_controls=0, initial_guess=nothing, cost_player = 1) +""" +Calculate costs corresponding to a perfect info term in the Stage 1 cost function with index world_idx. + +Input: + ps: prior distribution of k worlds for each signal, nx1 vector + βs: vector containing P2's cost parameters for each world. vector of nx1 vectors + world_idx: index of the world for which the cost is calculated +Output: + id_costs: 2D Matrix of costs for each r in the simplex +""" +function calculate_id_costs(ps, βs, world_idx; dr = 0.05, return_controls = 0) @assert sum(ps) ≈ 1.0 "Prior distribution ps must be a probability distribution" - game, _ = build_stage_2(ps, βs) + # complete_info_game = build_complete_info_game() + # incomplete_info_game = build_incomplete_info_game(ps, βs) rs = 0:dr:1 num_worlds = length(ps) id_costs = NaN * ones(Float64, Int(1 / dr + 1), Int(1 / dr + 1)) - if(return_controls>0) ## ideally, it should be 1 or 2 for P1 or P2 - if(return_controls <= 2) + if (return_controls > 0) ## ideally, it should be 1 or 2 for P1 or P2 + if (return_controls <= 2) controls = NaN * ones(Float64, Int(1 / dr + 1), Int(1 / dr + 1), 3) - else + else println("Invalid return_controls option.") return_controls = 0 end @@ -334,7 +822,8 @@ function calculate_id_costs(ps, βs, world_idx; dr = 0.05, return_controls=0, in end r3 = 1 - r1 - r2 r = [r1, r2, r3] - x = compute_stage_2(r, ps, βs, game, initial_guess=initial_guess) + # x = compute_stage_2(r, ps, βs, complete_info_game, incomplete_info_game) + x = compute_stage_2(IBRGameSolver(), r, ps, βs, [J_1, J_2]) id_cost = r[world_idx] * ps[world_idx] * @@ -351,23 +840,34 @@ function calculate_id_costs(ps, βs, world_idx; dr = 0.05, return_controls=0, in end end - if(return_controls>0) + if (return_controls > 0) return id_costs, controls else return id_costs end end -function calculate_misid_costs(ps, βs, world_idx; dr = 0.05, return_controls = 0, initial_guess=nothing, cost_player = 1) +""" +Calculate costs corresponding to a imperfect info term in the Stage 1 cost function with index world_idx. + +Input: + ps: prior distribution of k worlds for each signal, nx1 vector + βs: vector containing P2's cost parameters for each world. vector of nx1 vectors + world_idx: index of the world for which the cost is calculated +Output: + id_costs: 2D Matrix of costs for each r in the simplex +""" +function calculate_misid_costs(ps, βs, world_idx; dr = 0.05, return_controls = 0) @assert sum(ps) ≈ 1.0 "Prior distribution ps must be a probability distribution" - game, _ = build_stage_2(ps, βs) + # complete_info_game = build_complete_info_game() + # incomplete_info_game = build_incomplete_info_game(ps, βs) rs = 0:dr:1 num_worlds = length(ps) misid_costs = NaN * ones(Float64, Int(1 / dr + 1), Int(1 / dr + 1)) - if(return_controls>0) ## ideally, it should be 1 or 2 for P1 or P2 - if(return_controls <= 2) + if (return_controls > 0) ## ideally, it should be 1 or 2 for P1 or P2 + if (return_controls <= 2) controls = NaN * ones(Float64, Int(1 / dr + 1), Int(1 / dr + 1), 3) - else + else println("Invalid return_controls option.") return_controls = 0 end @@ -381,7 +881,8 @@ function calculate_misid_costs(ps, βs, world_idx; dr = 0.05, return_controls = end r3 = 1 - r1 - r2 r = [r1, r2, r3] - x = compute_stage_2(r, ps, βs, game, initial_guess=initial_guess) + # x = compute_stage_2(r, ps, βs, complete_info_game, incomplete_info_game) + x = compute_stage_2(IBRGameSolver(), r, ps, βs, [J_1, J_2]) defender_signal_0 = x[Block(1)] attacker_signal_0_world_idx = x[Block(world_idx + num_worlds + 1)] misid_cost = J(defender_signal_0, attacker_signal_0_world_idx, βs[world_idx]) @@ -394,12 +895,11 @@ function calculate_misid_costs(ps, βs, world_idx; dr = 0.05, return_controls = end end - if(return_controls>0) + if (return_controls > 0) return misid_costs, controls else return misid_costs end - end """ @@ -473,7 +973,7 @@ Output: function display_stage_1_costs(costs, ps) rs = 0:(1 / (size(costs[1])[1] - 1)):1 num_worlds = length(ps) - fig = Figure(size = (1500, 800), title = "test") + fig = Figure(size = (900, 700), title = "test", fontsize = 22) max_value = 1.0 axs = [ [ @@ -481,15 +981,14 @@ function display_stage_1_costs(costs, ps) fig[1, world_idx], aspect = (1, 1, 1), perspectiveness = 0.5, - elevation = pi / 5, + elevation = pi / 9, azimuth = -π * (1 / 2 + 1 / 4), zgridcolor = :grey, ygridcolor = :grey, xgridcolor = :grey; xlabel = "r₁", ylabel = "r₂", - zlabel = "Cost", - title = "World $world_idx", + title = L"\mathbf{r}_{%$world_idx} p(\omega_{%$world_idx})J^1(...)", limits = (nothing, nothing, (0.01, max_value)), ) for world_idx in 1:num_worlds ], @@ -505,8 +1004,7 @@ function display_stage_1_costs(costs, ps) xgridcolor = :grey; xlabel = "r₁", ylabel = "r₂", - zlabel = "Cost", - title = "World $world_idx", + title = L"(1 - \mathbf{r}_{%$world_idx}) p(\omega_{%$world_idx})J^1(...)", limits = (nothing, nothing, (0.01, max_value)), ) for world_idx in 1:num_worlds ], @@ -520,10 +1018,7 @@ function display_stage_1_costs(costs, ps) colormap = :viridis, colorrange = (0, max_value), ) - # text!(axs[world_idx], "$(round(ps[1], digits=2))", position = (0.9, 0.4, cost_min), font = "Bold") - # text!(axs[world_idx], "$(round(ps[2], digits=2))", position = (0.1, 0.95, cost_min), font = "Bold") - # text!(axs[world_idx], "$(round(ps[3], digits=2))", position = (0.2, 0.1, cost_min), font = "Bold") - + hidezdecorations!(axs[1][world_idx]; ticklabels = false, ticks = false, grid = false) end for world_idx in 1:num_worlds hmap = surface!( @@ -534,57 +1029,10 @@ function display_stage_1_costs(costs, ps) colormap = :viridis, colorrange = (0, max_value), ) - # text!(axs[world_idx], "$(round(ps[1], digits=2))", position = (0.9, 0.4, cost_min), font = "Bold") - # text!(axs[world_idx], "$(round(ps[2], digits=2))", position = (0.1, 0.95, cost_min), font = "Bold") - # text!(axs[world_idx], "$(round(ps[3], digits=2))", position = (0.2, 0.1, cost_min), font = "Bold") - - if world_idx == num_worlds - Colorbar( - fig[1:2, num_worlds + 1], - hmap; - label = "Cost", - width = 15, - ticksize = 15, - tickalign = 1, - ) - end + hidezdecorations!(axs[2][world_idx]; ticklabels = false, ticks = false, grid = false) end - fig -end - - -function display_residuals(costs, ps) - rs = 0:(1 / (size(costs[1])[1] - 1)):1 - num_worlds = length(ps) - fig = Figure(size = (1500, 500), title = "test") - axs = [ - Axis3( - fig[1, world_idx], - aspect = (1, 1, 1), - perspectiveness = 0.5, - elevation = pi / 5, - azimuth = -π * (1 / 2 + 1 / 4), - zgridcolor = :grey, - ygridcolor = :grey, - xgridcolor = :grey; - xlabel = "r₁", - ylabel = "r₂", - zlabel = "Residual", - title = "World $world_idx", - # limits = (nothing, nothing, (0.01, 1)), - ) for world_idx in 1:num_worlds - ] - for world_idx in 1:num_worlds - hmap = surface!( - axs[world_idx], - rs, - rs, - costs[world_idx], - colormap = :viridis, - # colorrange = (0, 1), - ) - end + save("figures/stage_1_costs.png", fig) fig end @@ -599,98 +1047,57 @@ Output: function display_stage_1_costs_controls(costs, controls, ps; save_file = "", cost_player=1) rs = 0:(1 / (size(costs[1])[1] - 1)):1 num_worlds = length(ps) - fig = Figure(size = (1500, 1000), title = "test") - ylims = cost_player == 2 ? (-1.0,0.0) : (0.01, 1.0) ## either graph from y=0,1 (for normalized cost for P1), or else y=-1,0 (for P2) + fig = Figure(size = (600, 400), title = "test") axs = [ [ - Axis3( + Axis( fig[1, world_idx], - aspect = (1, 1, 1), - perspectiveness = 0.5, - elevation = pi / 5, - azimuth = -π * (1 / 2 + 1 / 4), - zgridcolor = :grey, + aspect = 1, ygridcolor = :grey, xgridcolor = :grey; xlabel = "r₁", ylabel = "r₂", - zlabel = "Cost", - title = "W$world_idx, S$world_idx", - limits = (nothing, nothing, ylims), + title = "World $world_idx, Signal $world_idx", + limits = (nothing, nothing), ) for world_idx in 1:num_worlds ], [ - Axis3( + Axis( fig[2, world_idx], - aspect = (1, 1, 1), - perspectiveness = 0.5, - elevation = pi / 5, - azimuth = -π * (1 / 2 + 1 / 4), - zgridcolor = :grey, + aspect = 1, ygridcolor = :grey, xgridcolor = :grey; xlabel = "r₁", ylabel = "r₂", - zlabel = "Cost", - title = "W$world_idx, S0", - limits = (nothing, nothing, ylims), + title = "World $world_idx, Signal 0", + limits = (nothing, nothing), ) for world_idx in 1:num_worlds ], ] + for world_idx in 1:num_worlds colors = get_RGB_vect(controls[world_idx]) for ii in 1:size(costs[world_idx])[1] for jj in 1:size(costs[world_idx])[2] - hmap = scatter!( - axs[1][world_idx], - rs[ii], - rs[jj], - costs[world_idx][ii,jj], - color = colors[ii,jj], - # colormap = :viridis, - # colorrange = (0, max_value), - ) + if rs[ii] + rs[jj] > 1 + continue + end + hmap = scatter!(axs[1][world_idx], rs[ii], rs[jj], color = colors[ii, jj]) end end - - # text!(axs[world_idx], "$(round(ps[1], digits=2))", position = (0.9, 0.4, cost_min), font = "Bold") - # text!(axs[world_idx], "$(round(ps[2], digits=2))", position = (0.1, 0.95, cost_min), font = "Bold") - # text!(axs[world_idx], "$(round(ps[3], digits=2))", position = (0.2, 0.1, cost_min), font = "Bold") - end for world_idx in 1:num_worlds - colors = get_RGB_vect(controls[world_idx+num_worlds]) - for ii in 1:size(costs[world_idx+num_worlds])[1] - for jj in 1:size(costs[world_idx+num_worlds])[2] - hmap = scatter!( - axs[2][world_idx], - rs[ii], - rs[jj], - costs[world_idx+num_worlds][ii,jj], - color = colors[ii,jj], - # colormap = :viridis, - # colorrange = (0, max_value), - ) + colors = get_RGB_vect(controls[world_idx + num_worlds]) + for ii in 1:size(costs[world_idx + num_worlds])[1] + for jj in 1:size(costs[world_idx + num_worlds])[2] + if rs[ii] + rs[jj] > 1 + continue + end + hmap = scatter!(axs[2][world_idx], rs[ii], rs[jj], color = colors[ii, jj]) end end - # text!(axs[world_idx], "$(round(ps[1], digits=2))", position = (0.9, 0.4, cost_min), font = "Bold") - # text!(axs[world_idx], "$(round(ps[2], digits=2))", position = (0.1, 0.95, cost_min), font = "Bold") - # text!(axs[world_idx], "$(round(ps[3], digits=2))", position = (0.2, 0.1, cost_min), font = "Bold") - - # if world_idx == num_worlds - # Colorbar( - # fig[1:2, num_worlds + 1], - # hmap; - # label = "Cost", - # width = 15, - # ticksize = 15, - # tickalign = 1, - # ) - # end end - - filename = "figures/"*save_file*"stage_1_controls.png" - save(filename, fig) + save("figures/stage_2_controls.png", fig) fig end @@ -698,9 +1105,9 @@ end This is quick function to turn my controls into RGB vectors """ function get_RGB_vect(controls) - R = controls[:,:,1] - G = controls[:,:,2] - B = controls[:,:,3] + R = controls[:, :, 1] + G = controls[:, :, 2] + B = controls[:, :, 3] if size(R) == size(G) == size(B) n = size(R)[1] m = size(R)[2] @@ -712,12 +1119,10 @@ function get_RGB_vect(controls) end return RGB_values else - return(RGB(1,0,0)) + return (RGB(1, 0, 0)) end - end - """ Calculate Stage 1's objective function for all possible values of r. @@ -730,7 +1135,6 @@ Outputs: """ function calculate_stage_1_costs(ps, βs; dr = 0.05, normalize = true) @assert sum(ps) ≈ 1.0 "Prior distribution ps must be a probability distribution" - game, _ = build_stage_2(ps, βs) rs = 0:dr:1 Ks = NaN * ones(Float64, Int(1 / dr + 1), Int(1 / dr + 1)) for (i, r1) in enumerate(rs) @@ -740,7 +1144,7 @@ function calculate_stage_1_costs(ps, βs; dr = 0.05, normalize = true) end r3 = 1 - r1 - r2 r = [r1, r2, r3] - x = compute_stage_2(r, ps, βs, game) + x = compute_stage_2(IBRGameSolver(), r, ps, βs, [J_1, J_2]) K = compute_K(r, x, ps, βs) Ks[i, j] = K end @@ -766,7 +1170,7 @@ Output: """ function display_surface(ps, Ks) rs = 0:(1 / (size(Ks)[1] - 1)):1 - fig = Figure(size = (600, 400)) + fig = Figure(size = (450, 375)) ax = Axis3( fig[1, 1], aspect = (1, 1, 1), @@ -778,212 +1182,13 @@ function display_surface(ps, Ks) xgridcolor = :grey; xlabel = "r₁", ylabel = "r₂", - zlabel = "K", - title = "Normalized stage 1 cost\n priors = $(round.(ps, digits=2))", + zlabel = "|K|", limits = (nothing, nothing, (0.01, 1)), ) Ks_min = minimum(filter(!isnan, Ks)) hmap = surface!(ax, rs, rs, Ks, colorrange = (0, 1)) - Colorbar(fig[1, 2], hmap; label = "K", width = 15, ticksize = 15, tickalign = 1) - text!(ax, "$(round(ps[1], digits=2))", position = (0.9, 0.2, 0.01), font = "Bold") - text!(ax, "$(round(ps[2], digits=2))", position = (0.1, 0.95, 0.01), font = "Bold") - text!(ax, "$(round(ps[3], digits=2))", position = (0.1, 0.2, 0.01), font = "Bold") - fig -end - -""" -Project onto simplex using Fig. 1 Duchi 2008 -""" -function project_onto_simplex(v; z = 1.0) - μ = sort(v, rev = true) - ρ = findfirst([μ[j] - 1 / j * (sum(μ[1:j]) - z) <= 0 for j in eachindex(v)]) - ρ = isnothing(ρ) ? length(v) : ρ - 1 - θ = 1 / ρ * (sum(μ[1:ρ]) - z) - return [maximum([v[i] - θ, 0]) for i in eachindex(v)] -end - -"Defender cost function" -function J_1(u, v, β) - -J_2(u, v, β) -end - -""" -Attacker cost function -β: vector containing P2's (attacker) preference parameters for each world. -""" -function J_2(u, v, β) - # -sum([β[ii]^(v[ii] - u[ii]) for ii in eachindex(β)]) - δ = [β[ii]*v[ii] - u[ii] for ii in eachindex(β)] - -sum([activate(δ[j])*(β[j]*v[j]-u[j])^2 for j in eachindex(β)]) - -end - -function activate(δ; k=10.0) - return 1/(1 + exp(-2 * δ * k)) -end - -""" -Build parametric game for Stage 2. - -Inputs: - ps: prior distribution of k worlds for each signal, nx1 vector - βs: vector containing P1's cost parameters for each world. vector of nx1 vectors -Outputs: - parametric_game: ParametricGame object - fs: vector of symbolic expressions for each player's objective function - -""" -function build_stage_2(ps, βs) - n = length(ps) # assume n_signals = n_worlds + 1 - n_players = 1 + n^2 + Colorbar(fig[1, 2], hmap; label = "|K|", width = 15, ticksize = 15, tickalign = 1) - # Define Bayesian game player costs in Stage 2 - p_w_k_0(w_idx, θ) = (1 - θ[w_idx]) * ps[w_idx] / (1 - θ' * ps) - fs = [ - (x, θ) -> sum([ - J_1(x[Block(1)], x[Block(w_idx + n + 1)], βs[w_idx]) * p_w_k_0(w_idx, θ) for - w_idx in 1:n - ]), # u|s¹=0 IPI - [ - (x, θ) -> J_1(x[Block(w_idx + 1)], x[Block(w_idx + 2 * n + 1)], βs[w_idx]) for - w_idx in 1:n - ]..., # u|s¹={1,2,3} PI - [(x, θ) -> J_2(x[Block(1)], x[Block(w_idx + n + 1)], βs[w_idx]) for w_idx in 1:n]..., # v|s¹=0 IPI - [ - (x, θ) -> J_2(x[Block(w_idx + 1)], x[Block(w_idx + 2 * n + 1)], βs[w_idx]) for - w_idx in 1:n - ]..., # v|s¹={1,2,3} PI - ] - - # equality constraints - gs = [(x, θ) -> [sum(x[Block(i)]) - 1] for i in 1:n_players] # Everyone must attack/defend - - # inequality constraints - hs = [(x, θ) -> x[Block(i)] for i in 1:n_players] # All vars must be non-negative - - # shared constraints - g̃ = (x, θ) -> [0] - h̃ = (x, θ) -> [0] - - ParametricGame(; - objectives = fs, - equality_constraints = gs, - inequality_constraints = hs, - shared_equality_constraint = g̃, - shared_inequality_constraint = h̃, - parameter_dimension = 3, - primal_dimensions = [3 for _ in 1:n_players], - equality_dimensions = [1 for _ in 1:n_players], - inequality_dimensions = [3 for _ in 1:n_players], - shared_equality_dimension = 1, - shared_inequality_dimension = 1, - ), - fs -end - -""" -Compute objective at Stage 1 -""" -function compute_K(r, x, ps, βs) - n = length(ps) - sum([(1 - r[j]) * ps[j] * J_1(x[Block(1)], x[Block(j + n + 1)], βs[j]) for j in 1:n]) + sum([r[j] * ps[j] * J_1(x[Block(j + 1)], x[Block(j + 2 * n + 1)], βs[j]) for j in 1:n]) -end - -""" -Compute derivative of Stage 1's objective function w.r.t. x -""" -function compute_dKdx(r, x, ps, βs) - gradient(x -> compute_K(r, x, ps, βs), x)[1] -end - -""" -Compute full derivative of Stage 1's objective function w.r.t. r - -Inputs: - x: decision variables of Stage 2 - ps: prior distribution of k worlds, nx1 vector - -Outputs: - djdq: Jacobian of Stage 1's objective function w.r.t. r -""" -function compute_dKdr(r, x, ps, βs, game) - dKdx = compute_dKdx(r, x, ps, βs) - dKdr = gradient(r -> compute_K(r, x, ps, βs), r)[1] - dxdr = compute_dxdr(r, x, ps, βs, game) - n = length(ps) - for idx in 1:(1 + n^2) - dKdr += (dKdx[Block(idx)]' * dxdr[Block(idx)])' - end - dKdr -end - -""" -Solve stage 2 and return full derivative of objective function w.r.t. r - -Inputs: - r: scout allocation - ps: prior distribution of k worlds, nx1 vector - βs: vector containing P2's cost parameters for each world. vector of nx1 vectors - -Outputs: - dxdr: Blocked Jacobian of Stage 2's decision variables w.r.t. Stage 1's decision variable -""" -function compute_dxdr(r, x, ps, βs, game; verbose = false) - n = length(ps) - n_players = 1 + n^2 - var_dim = n # TODO: Change this to be more general - - # Return Jacobian - dxdr = jacobian( - r -> solve( - game, - r; - initial_guess = vcat(x, zeros(total_dim(game) - n_players * var_dim)), - verbose = false, - return_primals = false, - ).variables[1:(n_players * var_dim)], - r, - )[1] - - BlockArray(dxdr, [var_dim for _ in 1:n_players], [var_dim]) -end - -""" -Return Stage 2 decision variables given scout allocation r - -Input: - r: scout allocation - ps: prior distribution of k worlds, nx1 vector - βs: vector containing P2's cost parameters for each world. Vector of nx1 vectors -Output: - x: decision variables of Stage 2 given r. BlockedArray with a block per player -""" -function compute_stage_2(r, ps, βs, game; initial_guess = nothing, return_residual = false,verbose=false) - n = length(ps) # assume n_signals = n_worlds + 1 - n_players = 1 + n^2 - var_dim = n # TODO: Change this to be more general - - solution = solve( - game, - r; - initial_guess = isnothing(initial_guess) ? 1/3 * ones(total_dim(game)) : initial_guess, ### gives smooth cost surfaces - # initial_guess = isnothing(initial_guess) ? repeat([1.0,0.0,0.0],24) : initial_guess, - # initial_guess = isnothing(initial_guess) ? vcat(repeat([1.0,0.0,0.0],10),zeros(14*3)) : initial_guess, - # initial_guess = isnothing(initial_guess) ? vcat(repeat([1.0,0.0,0.0],10),(1/3) * ones(14*3)) : initial_guess, - # initial_guess = isnothing(initial_guess) ? vcat((1/3)*ones(30),(1/3)*ones(10),zeros(32)) : initial_guess, - # initial_guess = isnothing(initial_guess) ? vcat((1/3)*ones(30),(0.0)*ones(10),(1/3)*ones(30),zeros(2)) : initial_guess, ### gives smooth cost surfaces - # initial_guess = isnothing(initial_guess) ? vcat((1/3)*ones(30),(0.0)*ones(10),repeat([0.0, 0.5,0.5],10),zeros(2)) : initial_guess, ## also smooth - # initial_guess = isnothing(initial_guess) ? vcat(repeat([1.0, 0.0,0.0],10),(0.0)*ones(10),repeat([0.0, 0.5,0.5],10),zeros(2)) : initial_guess, - # initial_guess = isnothing(initial_guess) ? vcat(repeat([0.9,0.05,0.05],4),repeat([0.1,0.5,0.4],6),(1/3)*ones(14*3)) : initial_guess, - # initial_guess = initial_guess, - verbose = verbose, - return_primals = false, - ) - - if return_residual - return BlockArray(solution.variables[1:(n_players * var_dim)], [n for _ in 1:n_players]), - solution.info.residual - else - return BlockArray(solution.variables[1:(n_players * var_dim)], [n for _ in 1:n_players]) - end + save("figures/stage_1_surface.png", fig) + fig end \ No newline at end of file diff --git a/experiments/tower_defense_exponential.jl b/experiments/tower_defense_exponential.jl deleted file mode 100644 index 7b7172e..0000000 --- a/experiments/tower_defense_exponential.jl +++ /dev/null @@ -1,249 +0,0 @@ -using GamesVoI -using BlockArrays -using LinearAlgebra: norm_sqr -using Zygote - -""" Nomenclature - n : Number of worlds (=3) - pws = [P(w₁),..., P(wₙ)] : prior distribution of k worlds for each signal, nx1 vector - ws : vector containing P2's cost parameters for each world. vector of nx1 vectors - x[Block(1)] : u(0), P1's action given signal s¹=0 depends on r - x[Block(2)] : u(1), P1's action given signal s¹=1 - x[Block(3)] : u(2), P1's action given signal s¹=2 - x[Block(4)] : u(3), P1's action given signal s¹=3 - x[Block(5)] ~ x[Block(7)] : v(wₖ, 0), P2's action for each worlds given signal s¹=0 depends on r - x[Block(8)] : v(wₖ, 1), P2's action for world 1 given signal s¹=1 - x[Block(9)] : v(wₖ, 2), P2's action for world 2 given signal s¹=2 - x[Block(10)] : v(wₖ, 3), P2's action for world 3 given signal s¹=3 - θ = rₖ = [r₁, ... , rₙ] : r, Scout allocation in each direction - J : Stage 1's objective function -""" - - -""" -Solve Stage 1 to find optimal scout allocation r. - -Inputs: - pws: prior distribution of k worlds, nx1 vector - r_init: initial guess scout allocation -Outputs: - r: optimal scout allocation -""" -function solve_r(pws, ws; r_init = [1/3, 1/3, 1/3], iter_limit=50, target_error=.00001, α=1, return_states = false) - cur_iter = 0 - n = length(pws) - n_players = 1 + n^2 - var_dim = n # TODO: Change this to be more general - if return_states - x_list = [] - r_list = [] - end - - game, _ = build_stage_2(pws, ws) - r = r_init - println("0: r = $r") - x = compute_stage_2(r, pws, ws, game) - dJdr = zeros(Float64, n) - while cur_iter < iter_limit # TODO: Break if change from last iteration is small - dJdr = compute_dJdr(r, x, pws, ws, game) - r_temp = r - α .* dJdr - r = project_onto_simplex(r_temp) - x = compute_stage_2( - r, pws, ws, game; - initial_guess=vcat(x, zeros(total_dim(game) - n_players * var_dim)) - ) - if return_states - push!(x_list,x) - push!(r_list,r) - end - cur_iter += 1 - println("$cur_iter: r = $r") - # println("x = $x \n") - # print_state(x) - end - println("$cur_iter: r = $r") - if return_states - r_matrix = reduce(hcat, r_list) - x_matrix = reduce(hcat, x_list) - out = Dict("r"=>r, "x"=>x, "r_matrix"=>r_matrix, "x_matrix"=>x_matrix) - return out - end - return r -end - -""" -Project onto simplex using Fig. 1 Duchi 2008 -""" -function project_onto_simplex(v; z=1.0) - μ = sort(v, rev=true) - ρ = findfirst([μ[j] - 1/j * (sum(μ[1:j]) - z) <= 0 for j in eachindex(v)]) - ρ = isnothing(ρ) ? length(v) : ρ - 1 - θ = 1/ρ * (sum(μ[1:ρ]) - z) - return [maximum([v[i] - θ, 0]) for i in eachindex(v)] -end - - -function print_state(x) - out = reshape(x,3,10) - println("x = $out") -end - -"Defender cost function" -function J_1(u, v) - norm_sqr(u - v) -end - -"Attacker cost function" -function J_2(u, v, w) - m = length(w) - sum([w[ii]^(v[ii]-u[ii]) for ii=1:m]) - # (u[w] - v[w]) # P2 only cares about a SINGLE direction. -end - - -""" -Build parametric game for Stage 2. - -Inputs: - pws: prior distribution of k worlds for each signal, nx1 vector - ws: vector containing P1's cost parameters for each world. vector of nx1 vectors -Outputs: - parametric_game: ParametricGame object - fs: vector of symbolic expressions for each player's objective function - -""" -function build_stage_2(pws, ws) - - n = length(pws) # assume n_signals = n_worlds + 1 - n_players = 1 + n^2 - - # Define Bayesian game player costs in Stage 2 - p_w_k_0(w_idx, θ) = (1 - θ[w_idx]) * pws[w_idx] / (1 - θ' * pws) - fs = [ - (x, θ) -> sum([J_1(x[Block(1)], x[Block(w_idx + n + 1)]) * p_w_k_0(w_idx, θ) for w_idx in 1:n]), # u|s¹=0 IPI - [(x, θ) -> J_1(x[Block(w_idx + 1)], x[Block(w_idx + 2 * n + 1)]) for w_idx in 1:n]..., # u|s¹={1,2,3} PI - [(x, θ) -> J_2(x[Block(1)], x[Block(w_idx + n + 1)], ws[w_idx]) for w_idx in 1:n]..., # v|s¹=0 IPI - [(x, θ) -> J_2(x[Block(w_idx + 1)], x[Block(w_idx + 2 * n + 1)], ws[w_idx]) for w_idx in 1:n]... # v|s¹={1,2,3} PI - ] - - # equality constraints - gs = [(x, θ) -> [sum(x[Block(i)]) - 1] for i in 1:n_players] # Everyone must attack/defend - - # inequality constraints - hs = [(x, θ) -> x[Block(i)] for i in 1:n_players] # All vars must be non-negative - - # shared constraints - g̃ = (x, θ) -> [0] - h̃ = (x, θ) -> [0] - - ParametricGame(; - objectives=fs, - equality_constraints=gs, - inequality_constraints=hs, - shared_equality_constraint=g̃, - shared_inequality_constraint=h̃, - parameter_dimension=3, - primal_dimensions=[3 for _ in 1:n_players], - equality_dimensions=[1 for _ in 1:n_players], - inequality_dimensions=[3 for _ in 1:n_players], - shared_equality_dimension=1, - shared_inequality_dimension=1 - ), fs -end - -""" -Compute objective at Stage 1 -""" -function compute_J(r, x, pws, ws) - n = length(pws) - sum([(1 - r[w_idx]) * pws[w_idx] * J_1(x[Block(1)], x[Block(w_idx + n + 1)]) for w_idx in 1:n]) + sum([r[w_idx] * pws[w_idx] * J_1(x[Block(w_idx + 1)], x[Block(w_idx + 2 * n + 1)]) for w_idx in 1:n]) -end - -""" -Compute derivative of Stage 1's objective function w.r.t. x -""" -function compute_dJdx(r, x, pws, ws) - gradient(x -> compute_J(r, x, pws, ws), x)[1] -end - -""" -Compute full derivative of Stage 1's objective function w.r.t. r - -Inputs: - x: decision variables of Stage 2 - pws: prior distribution of k worlds, nx1 vector - -Outputs: - djdq: Jacobian of Stage 1's objective function w.r.t. r -""" -function compute_dJdr(r, x, pws, ws, game) - dJdx = compute_dJdx(r, x, pws, ws) - dJdr = gradient(r -> compute_J(r, x, pws, ws), r)[1] - dxdr = compute_dxdr(r, x, pws, ws, game) - - dJdr_norm = norm_sqr(dJdx) - dxdr_norm = norm_sqr(dxdr) - - println("dJdr = $dJdr_norm") - println("dxdr = $dxdr_norm") - - n = length(pws) - for idx in 1:(1 + n^2) - dJdr += (dJdx[Block(idx)]' * dxdr[Block(idx)])' - end - dJdr -end - -""" -Solve stage 2 and return full derivative of objective function w.r.t. r - -Inputs: - r: scout allocation - pws: prior distribution of k worlds, nx1 vector - ws: vector containing P2's cost parameters for each world. vector of nx1 vectors - -Outputs: - dxdr: Blocked Jacobian of Stage 2's decision variables w.r.t. Stage 1's decision variable -""" -function compute_dxdr(r, x, pws, ws, game; verbose=false) - n = length(pws) - n_players = 1 + n^2 - var_dim = n # TODO: Change this to be more general - - # Return Jacobian - dxdr = jacobian(r -> solve( - game, - r; - initial_guess=vcat(x, zeros(total_dim(game) - n_players * var_dim)), - verbose=false, - return_primals=false - ).variables[1:n_players*var_dim], r)[1] - - BlockArray(dxdr, [var_dim for _ in 1:n_players], [var_dim]) -end - -""" -Return Stage 2 decision variables given scout allocation r - -Input: - r: scout allocation - pws: prior distribution of k worlds, nx1 vector - ws: vector containing P2's cost parameters for each world. Vector of nx1 vectors -Output: - x: decision variables of Stage 2 given r. BlockedArray with a block per player -""" -function compute_stage_2(r, pws, ws, game; initial_guess = nothing, verbose=false) - n = length(pws) # assume n_signals = n_worlds + 1 - n_players = 1 + n^2 - var_dim = n # TODO: Change this to be more general - - solution = solve( - game, - r; - initial_guess=isnothing(initial_guess) ? zeros(total_dim(game)) : initial_guess, - verbose=verbose, - return_primals=false - ) - - BlockArray(solution.variables[1:n_players * var_dim], [n for _ in 1:n_players]) -end \ No newline at end of file diff --git a/figures/all.png b/figures/all.png new file mode 100644 index 0000000..a96c4fa Binary files /dev/null and b/figures/all.png differ diff --git a/figures/stage_1_attack.png b/figures/stage_1_attack.png new file mode 100644 index 0000000..2b28b90 Binary files /dev/null and b/figures/stage_1_attack.png differ diff --git a/figures/stage_1_controls.png b/figures/stage_1_controls.png new file mode 100644 index 0000000..b947dd6 Binary files /dev/null and b/figures/stage_1_controls.png differ diff --git a/figures/stage_1_costs.png b/figures/stage_1_costs.png new file mode 100644 index 0000000..d46ea42 Binary files /dev/null and b/figures/stage_1_costs.png differ diff --git a/figures/stage_1_surface.png b/figures/stage_1_surface.png new file mode 100644 index 0000000..20ac976 Binary files /dev/null and b/figures/stage_1_surface.png differ diff --git a/figures/stage_1_terms.jpg b/figures/stage_1_terms.jpg new file mode 100644 index 0000000..1e05cf8 Binary files /dev/null and b/figures/stage_1_terms.jpg differ diff --git a/figures/sweep/Old/sweep_assymetric_s0.0_k10.png b/figures/sweep/Old/sweep_assymetric_s0.0_k10.png new file mode 100644 index 0000000..5939fc1 Binary files /dev/null and b/figures/sweep/Old/sweep_assymetric_s0.0_k10.png differ diff --git a/figures/sweep/Old/sweep_assymetric_s3.0_k10.png b/figures/sweep/Old/sweep_assymetric_s3.0_k10.png new file mode 100644 index 0000000..31f0d83 Binary files /dev/null and b/figures/sweep/Old/sweep_assymetric_s3.0_k10.png differ diff --git a/figures/sweep/Old/sweep_assymetric_s5.0_k10.png b/figures/sweep/Old/sweep_assymetric_s5.0_k10.png new file mode 100644 index 0000000..8af97c8 Binary files /dev/null and b/figures/sweep/Old/sweep_assymetric_s5.0_k10.png differ diff --git a/figures/sweep/Old/sweep_asymetric_new_s0.0_k10.png b/figures/sweep/Old/sweep_asymetric_new_s0.0_k10.png new file mode 100644 index 0000000..eda50b5 Binary files /dev/null and b/figures/sweep/Old/sweep_asymetric_new_s0.0_k10.png differ diff --git a/figures/sweep/Old/sweep_asymetric_new_s0.5_k10.png b/figures/sweep/Old/sweep_asymetric_new_s0.5_k10.png new file mode 100644 index 0000000..8ec48c7 Binary files /dev/null and b/figures/sweep/Old/sweep_asymetric_new_s0.5_k10.png differ diff --git a/figures/sweep/Old/sweep_asymetric_new_s1.0_k10.png b/figures/sweep/Old/sweep_asymetric_new_s1.0_k10.png new file mode 100644 index 0000000..b3b98cb Binary files /dev/null and b/figures/sweep/Old/sweep_asymetric_new_s1.0_k10.png differ diff --git a/figures/sweep/Old/sweep_asymetric_new_s1.5_k10.png b/figures/sweep/Old/sweep_asymetric_new_s1.5_k10.png new file mode 100644 index 0000000..a125472 Binary files /dev/null and b/figures/sweep/Old/sweep_asymetric_new_s1.5_k10.png differ diff --git a/figures/sweep/Old/sweep_asymetric_new_s2.0_k10.png b/figures/sweep/Old/sweep_asymetric_new_s2.0_k10.png new file mode 100644 index 0000000..5649fac Binary files /dev/null and b/figures/sweep/Old/sweep_asymetric_new_s2.0_k10.png differ diff --git a/figures/sweep/Old/sweep_asymetric_new_s2.5_k10.png b/figures/sweep/Old/sweep_asymetric_new_s2.5_k10.png new file mode 100644 index 0000000..075e22f Binary files /dev/null and b/figures/sweep/Old/sweep_asymetric_new_s2.5_k10.png differ diff --git a/figures/sweep/Old/sweep_asymetric_new_s3.0_k10.png b/figures/sweep/Old/sweep_asymetric_new_s3.0_k10.png new file mode 100644 index 0000000..ed0bb09 Binary files /dev/null and b/figures/sweep/Old/sweep_asymetric_new_s3.0_k10.png differ diff --git a/figures/sweep/Old/sweep_asymetric_new_s3.5_k10.png b/figures/sweep/Old/sweep_asymetric_new_s3.5_k10.png new file mode 100644 index 0000000..0ceddc8 Binary files /dev/null and b/figures/sweep/Old/sweep_asymetric_new_s3.5_k10.png differ diff --git a/figures/sweep/Old/sweep_asymetric_new_s4.0_k10.png b/figures/sweep/Old/sweep_asymetric_new_s4.0_k10.png new file mode 100644 index 0000000..4b22c51 Binary files /dev/null and b/figures/sweep/Old/sweep_asymetric_new_s4.0_k10.png differ diff --git a/figures/sweep/Old/sweep_asymetric_new_s4.5_k10.png b/figures/sweep/Old/sweep_asymetric_new_s4.5_k10.png new file mode 100644 index 0000000..3ca564e Binary files /dev/null and b/figures/sweep/Old/sweep_asymetric_new_s4.5_k10.png differ diff --git a/figures/sweep/Old/sweep_asymetric_new_s5.0_k10.png b/figures/sweep/Old/sweep_asymetric_new_s5.0_k10.png new file mode 100644 index 0000000..5eb7638 Binary files /dev/null and b/figures/sweep/Old/sweep_asymetric_new_s5.0_k10.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s0.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s0.0_k1.0.png new file mode 100644 index 0000000..ca5ea11 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s0.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s1.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s1.0_k1.0.png new file mode 100644 index 0000000..0485298 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s1.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s10.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s10.0_k1.0.png new file mode 100644 index 0000000..60f4a97 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s10.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s100.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s100.0_k1.0.png new file mode 100644 index 0000000..8d01adc Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s100.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s15.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s15.0_k1.0.png new file mode 100644 index 0000000..fdeb5dc Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s15.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s2.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s2.0_k1.0.png new file mode 100644 index 0000000..4304373 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s2.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s20.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s20.0_k1.0.png new file mode 100644 index 0000000..438fb7c Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s20.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s25.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s25.0_k1.0.png new file mode 100644 index 0000000..db7f9c7 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s25.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s3.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s3.0_k1.0.png new file mode 100644 index 0000000..5e05050 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s3.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s30.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s30.0_k1.0.png new file mode 100644 index 0000000..1516acb Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s30.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s35.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s35.0_k1.0.png new file mode 100644 index 0000000..5897945 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s35.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s4.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s4.0_k1.0.png new file mode 100644 index 0000000..e21448b Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s4.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s40.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s40.0_k1.0.png new file mode 100644 index 0000000..dfaf6cc Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s40.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s45.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s45.0_k1.0.png new file mode 100644 index 0000000..5aa3ab2 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s45.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s5.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s5.0_k1.0.png new file mode 100644 index 0000000..65e3b2a Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s5.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s50.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s50.0_k1.0.png new file mode 100644 index 0000000..b4b0437 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s50.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s55.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s55.0_k1.0.png new file mode 100644 index 0000000..2b6a9d0 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s55.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s6.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s6.0_k1.0.png new file mode 100644 index 0000000..5dae181 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s6.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s60.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s60.0_k1.0.png new file mode 100644 index 0000000..dfb0eb1 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s60.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s65.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s65.0_k1.0.png new file mode 100644 index 0000000..b66dccb Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s65.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s7.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s7.0_k1.0.png new file mode 100644 index 0000000..9e2757d Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s7.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s70.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s70.0_k1.0.png new file mode 100644 index 0000000..89c800e Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s70.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s75.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s75.0_k1.0.png new file mode 100644 index 0000000..803e68f Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s75.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s8.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s8.0_k1.0.png new file mode 100644 index 0000000..58d3792 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s8.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s80.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s80.0_k1.0.png new file mode 100644 index 0000000..ad1cc8d Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s80.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s85.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s85.0_k1.0.png new file mode 100644 index 0000000..513e28b Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s85.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s9.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s9.0_k1.0.png new file mode 100644 index 0000000..d922187 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s9.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s90.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s90.0_k1.0.png new file mode 100644 index 0000000..8414fc9 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s90.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_single_perturbation_s95.0_k1.0.png b/figures/sweep/Old/sweep_single_perturbation_s95.0_k1.0.png new file mode 100644 index 0000000..90b5e25 Binary files /dev/null and b/figures/sweep/Old/sweep_single_perturbation_s95.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s0.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s0.0_k1.0.png new file mode 100644 index 0000000..4e65d8b Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s0.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s10.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s10.0_k1.0.png new file mode 100644 index 0000000..3374a1b Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s10.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s100.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s100.0_k1.0.png new file mode 100644 index 0000000..0b0c506 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s100.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s15.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s15.0_k1.0.png new file mode 100644 index 0000000..1cae2fc Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s15.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s20.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s20.0_k1.0.png new file mode 100644 index 0000000..295b163 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s20.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s25.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s25.0_k1.0.png new file mode 100644 index 0000000..1b3814a Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s25.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s30.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s30.0_k1.0.png new file mode 100644 index 0000000..91ef8e8 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s30.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s35.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s35.0_k1.0.png new file mode 100644 index 0000000..e805ac2 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s35.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s40.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s40.0_k1.0.png new file mode 100644 index 0000000..1ad6dd5 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s40.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s45.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s45.0_k1.0.png new file mode 100644 index 0000000..4d08869 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s45.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s5.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s5.0_k1.0.png new file mode 100644 index 0000000..d7de4f1 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s5.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s50.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s50.0_k1.0.png new file mode 100644 index 0000000..ce6d9a0 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s50.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s55.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s55.0_k1.0.png new file mode 100644 index 0000000..6d503a2 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s55.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s60.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s60.0_k1.0.png new file mode 100644 index 0000000..d8457af Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s60.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s65.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s65.0_k1.0.png new file mode 100644 index 0000000..97c0997 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s65.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s70.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s70.0_k1.0.png new file mode 100644 index 0000000..cf83bc8 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s70.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s75.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s75.0_k1.0.png new file mode 100644 index 0000000..d7632b9 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s75.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s80.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s80.0_k1.0.png new file mode 100644 index 0000000..2618dd8 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s80.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s85.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s85.0_k1.0.png new file mode 100644 index 0000000..afaaf17 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s85.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s90.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s90.0_k1.0.png new file mode 100644 index 0000000..f62c04e Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s90.0_k1.0.png differ diff --git a/figures/sweep/Old/sweep_uniform perturbation_s95.0_k1.0.png b/figures/sweep/Old/sweep_uniform perturbation_s95.0_k1.0.png new file mode 100644 index 0000000..dc54af8 Binary files /dev/null and b/figures/sweep/Old/sweep_uniform perturbation_s95.0_k1.0.png differ diff --git a/figures/sweep/sweep_ibr_balance_s0.0_k10.png b/figures/sweep/sweep_ibr_balance_s0.0_k10.png new file mode 100644 index 0000000..f0da90c Binary files /dev/null and b/figures/sweep/sweep_ibr_balance_s0.0_k10.png differ diff --git a/figures/sweep/sweep_ibr_balance_s1.0_k10.png b/figures/sweep/sweep_ibr_balance_s1.0_k10.png new file mode 100644 index 0000000..3d07df2 Binary files /dev/null and b/figures/sweep/sweep_ibr_balance_s1.0_k10.png differ diff --git a/figures/sweep/sweep_ibr_balance_s10.0_k10.png b/figures/sweep/sweep_ibr_balance_s10.0_k10.png new file mode 100644 index 0000000..8297d65 Binary files /dev/null and b/figures/sweep/sweep_ibr_balance_s10.0_k10.png differ diff --git a/figures/sweep/sweep_ibr_balance_s2.0_k10.png b/figures/sweep/sweep_ibr_balance_s2.0_k10.png new file mode 100644 index 0000000..0d424a0 Binary files /dev/null and b/figures/sweep/sweep_ibr_balance_s2.0_k10.png differ diff --git a/figures/sweep/sweep_ibr_balance_s3.0_k10.png b/figures/sweep/sweep_ibr_balance_s3.0_k10.png new file mode 100644 index 0000000..0a52eb5 Binary files /dev/null and b/figures/sweep/sweep_ibr_balance_s3.0_k10.png differ diff --git a/figures/sweep/sweep_ibr_balance_s4.0_k10.png b/figures/sweep/sweep_ibr_balance_s4.0_k10.png new file mode 100644 index 0000000..e459afd Binary files /dev/null and b/figures/sweep/sweep_ibr_balance_s4.0_k10.png differ diff --git a/figures/sweep/sweep_ibr_balance_s5.0_k10.png b/figures/sweep/sweep_ibr_balance_s5.0_k10.png new file mode 100644 index 0000000..cf4fba4 Binary files /dev/null and b/figures/sweep/sweep_ibr_balance_s5.0_k10.png differ diff --git a/figures/sweep/sweep_ibr_balance_s6.0_k10.png b/figures/sweep/sweep_ibr_balance_s6.0_k10.png new file mode 100644 index 0000000..e24ca91 Binary files /dev/null and b/figures/sweep/sweep_ibr_balance_s6.0_k10.png differ diff --git a/figures/sweep/sweep_ibr_balance_s7.0_k10.png b/figures/sweep/sweep_ibr_balance_s7.0_k10.png new file mode 100644 index 0000000..eb55ff0 Binary files /dev/null and b/figures/sweep/sweep_ibr_balance_s7.0_k10.png differ diff --git a/figures/sweep/sweep_ibr_balance_s8.0_k10.png b/figures/sweep/sweep_ibr_balance_s8.0_k10.png new file mode 100644 index 0000000..1a99b96 Binary files /dev/null and b/figures/sweep/sweep_ibr_balance_s8.0_k10.png differ diff --git a/figures/sweep/sweep_ibr_balance_s9.0_k10.png b/figures/sweep/sweep_ibr_balance_s9.0_k10.png new file mode 100644 index 0000000..1eadbb2 Binary files /dev/null and b/figures/sweep/sweep_ibr_balance_s9.0_k10.png differ diff --git a/gradient_descent_plotting.ipynb b/gradient_descent_plotting.ipynb deleted file mode 100644 index 335e78e..0000000 --- a/gradient_descent_plotting.ipynb +++ /dev/null @@ -1,1665 +0,0 @@ -{ - "cells": [ - { - "cell_type": "code", - "execution_count": 8, - "id": "9b261715", - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "compute_stage_2" - ] - }, - "metadata": {}, - "output_type": "display_data" - } - ], - "source": [ - "using Symbolics\n", - "using GamesVoI\n", - "using Plots\n", - "using DataFrames\n", - "# include(\"experiments/tower_defense_exponential.jl\")\n", - "include(\"experiments/tower_defense.jl\")" - ] - }, - { - "cell_type": "code", - "execution_count": 9, - "id": "53dd5c17", - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "0: r = [0.5, 0.25, 0.25]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "r = [0.556, 0.222, 0.222]\n", - "r = [0.63, 0.185, 0.185]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "r = [0.728, 0.136, 0.136]\n", - "r = [0.86, 0.07, 0.07]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n", - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "data": { - "text/plain": [ - "Dict{String, AbstractArray{Float64}} with 4 entries:\n", - " \"x_matrix\" => [0.222222 0.185185 … 5.44426e-8 5.44426e-8; 0.388889 0.407407 ……\n", - " \"x\" => [5.44426e-8, 0.5, 0.5, 1.0, -2.16126e-10, -2.16126e-10, -2.1612…\n", - " \"r\" => [1.0, 0.0, 0.0]\n", - " \"r_matrix\" => [0.555556 0.62963 … 1.0 1.0; 0.222222 0.185185 … 0.0 0.0; 0.222…" - ] - }, - "metadata": {}, - "output_type": "display_data" - } - ], - "source": [ - "prior = [1/3, 1/3,1/3]\n", - "omega_params = [[2, 1, 1], [1, 2, 1], [1, 1, 2]]\n", - "r_init =[0.5,0.25,0.25]\n", - "\n", - "out = solve_r(prior,omega_params,r_init=r_init, return_states=true)" - ] - }, - { - "cell_type": "markdown", - "id": "d689953e", - "metadata": {}, - "source": [ - "## Plot Decision Variables over GD" - ] - }, - { - "cell_type": "markdown", - "id": "6ab6fce7", - "metadata": {}, - "source": [ - "This code will take my `out` object from the previous code block, and use it to generate a plot showing the evolution of $r$ and $x$ over time." - ] - }, - { - "cell_type": "code", - "execution_count": 10, - "id": "062b3399", - "metadata": {}, - "outputs": [ - { - "data": { - "image/png": "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", - "image/svg+xml": [ - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n" - ], - "text/html": [ - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n" - ] - }, - "metadata": {}, - "output_type": "display_data" - } - ], - "source": [ - "x_matrix = out[\"x_matrix\"]\n", - "r_matrix = out[\"r_matrix\"]\n", - "titles = [\"r (Stage 1)\",\"P1, s1=0\",\"P1, s1=1\",\"P1, s1=2\",\"P1, s1=3\",\n", - " \"P2, (s1,w)=(0,1)\",\"P2, (s1,w)=(0,2)\",\"P2, (s1,w)=(0,3)\",\n", - " \"P2, (s1,w)=(1,1)\",\"P2, (s1,w)=(2,2)\",\"P2, (s1,w)=(3,3)\"]\n", - "index_map = [1,2,6,7,8,3,9,4,10,5,11] ## reorder states/titles in graph\n", - "# Assuming `x_matrix` is your n x m matrix\n", - "n, m = size(x_matrix)\n", - "num_vectors = 10 # Number of 3-vectors\n", - "vector_length = 3 # Length of each vector\n", - "@assert n == num_vectors*vector_length \"Provided dimensions do not match\";\n", - "\n", - "data = []\n", - " \n", - "push!(data, hcat(1:m,transpose(r_matrix)))\n", - "\n", - "for i in 1:num_vectors ## add Stage 2 decision variables\n", - " # Extract each 3-vector and reshape it\n", - " start_row = (i - 1) * vector_length + 1\n", - " end_row = start_row + vector_length - 1\n", - " vector_matrix = x_matrix[start_row:end_row, :]\n", - "\n", - " # Convert to DataFrame\n", - " new_matrix = hcat(1:m,transpose(vector_matrix))\n", - " \n", - " push!(data, new_matrix)\n", - "end\n", - "\n", - "\n", - "### PLOTTING\n", - "\n", - "using Plots\n", - "using DataFrames\n", - "\n", - "# Assuming `data` is your reshaped data suitable for plotting\n", - "# Each element in `data` is a DataFrame with columns representing the components of a 3-vector and rows representing timesteps\n", - "\n", - "plots = []\n", - "\n", - "\n", - "\n", - "for i in 1:11 # Assuming 10 such 3-vectors\n", - " j = index_map[i]\n", - " mat = data[j] # Your DataFrame for each 3-vector\n", - " if i>1\n", - " p = plot(mat[:,1],mat[:,2:4], xlabel=\"Time\", ylabel=\"Probability\", title=titles[j],ylimits=(0,1),xlimits=(0,50))\n", - " else\n", - " p = plot(mat[:,1],mat[:,2:4], xlabel=\"Time\", ylabel=\"Scouts\", title=titles[j],ylimits=(0,1),xlimits=(0,50))\n", - " end\n", - " push!(plots, p)\n", - " if i==1\n", - " push!(plots,plot(legend=false,grid=false,foreground_color_subplot=:white)) \n", - " end\n", - " if i>2 && i<5\n", - " push!(plots,plot(legend=false,grid=false,foreground_color_subplot=:white)) \n", - " end\n", - "end\n", - "\n", - "# Combine all subplots into one figure\n", - "plot(plots..., layout=(8, 2), legend=true) # Adjust layout as needed\n", - "plot!(size = (800, 1400))" - ] - }, - { - "cell_type": "markdown", - "id": "e64b9820", - "metadata": {}, - "source": [ - "## Plot Decisions/Costs over GD (under construction)" - ] - }, - { - "cell_type": "markdown", - "id": "2dae4f9d", - "metadata": {}, - "source": [ - "This section is still under construction. I wanted a visualization that included the costs, over time, as well. This would be useful for debugging." - ] - }, - { - "cell_type": "code", - "execution_count": null, - "id": "31b2fd0c", - "metadata": {}, - "outputs": [], - "source": [ - "game,fs = build_stage_2(prior,omega_params)" - ] - }, - { - "cell_type": "code", - "execution_count": 92, - "id": "b5753d38", - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "(1, 10)[2.3930842546822988e-32, 3.000000000000001, 3.0, 3.0, 0.6666666666730381, 0.6666666700460344, 0.6666666666392559, 2.2281153200039965, 2.697318212713702, 2.9428082063440497][6.958731016193841e-32, 3.000000000000001, 3.0, 3.0, 0.6666666927151942, 0.6666666734970208, 0.6666670438042436, 2.2281153181224544, 2.697318211960431, 2.9428081736570926][2.2177776286658088e-31, 3.000000000000001, 3.0, 3.0, 0.6666666672499614, 0.666666667057768, 0.6666666666712378, 2.2281153199624795, 2.697318213276322, 2.942808206342194]" - ] - }, - { - "data": { - "text/plain": [ - "3×10 Matrix{Float64}:\n", - " 2.39308e-32 3.0 3.0 3.0 0.666667 … 0.666667 2.22812 2.69732 2.94281\n", - " 6.95873e-32 3.0 3.0 3.0 0.666667 0.666667 2.22812 2.69732 2.94281\n", - " 2.21778e-31 3.0 3.0 3.0 0.666667 0.666667 2.22812 2.69732 2.94281" - ] - }, - "execution_count": 92, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "# game,fs = build_stage_2(prior,omega_params)\n", - "r_mat = out[\"r_matrix\"]\n", - "x_mat = out[\"x_matrix\"]\n", - "\n", - "cost_matrix = 1:10\n", - "cost_matrix = cost_matrix'\n", - "print(size(cost_matrix))\n", - "\n", - "m = size(r_mat)[2]\n", - "\n", - "for tt in 1:3\n", - " current_costs = [ff(BlockArray(x_mat[:,tt], repeat([3], outer = 10)),r_mat[:,tt]) for ff in fs]\n", - " cost_matrix = vcat(cost_matrix, current_costs')\n", - " print(current_costs)\n", - "end\n", - "cost_matrix = cost_matrix[2:end,:]" - ] - }, - { - "cell_type": "code", - "execution_count": 91, - "id": "07502637", - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "3×10 Matrix{Float64}:\n", - " 2.39308e-32 3.0 3.0 3.0 0.666667 … 0.666667 2.22812 2.69732 2.94281\n", - " 6.95873e-32 3.0 3.0 3.0 0.666667 0.666667 2.22812 2.69732 2.94281\n", - " 2.21778e-31 3.0 3.0 3.0 0.666667 0.666667 2.22812 2.69732 2.94281" - ] - }, - "execution_count": 91, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [] - }, - { - "cell_type": "code", - "execution_count": null, - "id": "8a7819b9", - "metadata": {}, - "outputs": [], - "source": [ - "matrix = out[\"x_matrix\"]\n", - "titles = [\"P1, s1=0\",\"P1, s1=1\",\"P1, s1=2\",\"P1, s1=3\",\n", - " \"P2, (s1,w)=(0,1)\",\"P2, (s1,w)=(0,2)\",\"P2, (s1,w)=(0,3)\",\n", - " \"P2, (s1,w)=(1,1)\",\"P2, (s1,w)=(2,2)\",\"P2, (s1,w)=(3,3)\"]\n", - "\n", - "# Assuming `matrix` is your n x m matrix\n", - "n, m = size(matrix)\n", - "num_vectors = 10 # Number of 3-vectors\n", - "vector_length = 3 # Length of each vector\n", - "@assert n == num_vectors*vector_length \"Provided dimensions do not match\";\n", - "\n", - "data_states = []\n", - "for i in 1:num_vectors\n", - " # Extract each 3-vector and reshape it\n", - " start_row = (i - 1) * vector_length + 1\n", - " end_row = start_row + vector_length - 1\n", - " vector_matrix = matrix[start_row:end_row, :]\n", - "\n", - " # Convert to DataFrame\n", - " new_matrix = hcat(1:m,transpose(vector_matrix))\n", - " \n", - " push!(data_states, new_matrix)\n", - "end\n", - "\n", - "\n", - "### PLOTTING\n", - "\n", - "using Plots\n", - "using DataFrames\n", - "\n", - "# Assuming `data` is your reshaped data suitable for plotting\n", - "# Each element in `data` is a DataFrame with columns representing the components of a 3-vector and rows representing timesteps\n", - "\n", - "plots = []\n", - "for i in 1:10 # Assuming 10 such 3-vectors\n", - " mat = data[i] # Your DataFrame for each 3-vector\n", - " p = plot(mat[:,1],mat[:,2:4], xlabel=\"Time\", ylabel=\"Probability\", title=titles[i],ylimits=(0,1),xlimits=(0,50))\n", - " push!(plots, p)\n", - "end\n", - "\n", - "# Combine all subplots into one figure\n", - "plot(plots..., layout=(5, 2), legend=true) # Adjust layout as needed\n", - "plot!(size = (800, 1000))" - ] - }, - { - "cell_type": "code", - "execution_count": 39, - "id": "1efc8dde", - "metadata": {}, - "outputs": [ - { - "ename": "LoadError", - "evalue": "KeyError: key \"xmatrix\" not found", - "output_type": "error", - "traceback": [ - "KeyError: key \"xmatrix\" not found", - "", - "Stacktrace:", - " [1] getindex(h::Dict{String, AbstractArray{Float64}}, key::String)", - " @ Base ./dict.jl:484", - " [2] top-level scope", - " @ In[39]:3" - ] - } - ], - "source": [ - "\n", - "\n" - ] - }, - { - "cell_type": "code", - "execution_count": 38, - "id": "f4d73914", - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "10-element Vector{Function}:\n", - " #419 (generic function with 1 method)\n", - " #422 (generic function with 1 method)\n", - " #422 (generic function with 1 method)\n", - " #422 (generic function with 1 method)\n", - " #424 (generic function with 1 method)\n", - " #424 (generic function with 1 method)\n", - " #424 (generic function with 1 method)\n", - " #426 (generic function with 1 method)\n", - " #426 (generic function with 1 method)\n", - " #426 (generic function with 1 method)" - ] - }, - "execution_count": 38, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "fs" - ] - }, - { - "cell_type": "code", - "execution_count": 73, - "id": "ddd5b034", - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "[2.3930842546822988e-32, 3.000000000000001, 3.0, 3.0, 0.6666666666730381, 0.6666666700460344, 0.6666666666392559, 2.2281153200039965, 2.697318212713702, 2.9428082063440497][6.958731016193841e-32, 3.000000000000001, 3.0, 3.0, 0.6666666927151942, 0.6666666734970208, 0.6666670438042436, 2.2281153181224544, 2.697318211960431, 2.9428081736570926][2.2177776286658088e-31, 3.000000000000001, 3.0, 3.0, 0.6666666672499614, 0.666666667057768, 0.6666666666712378, 2.2281153199624795, 2.697318213276322, 2.942808206342194]" - ] - } - ], - "source": [ - "r_mat = out[\"r_matrix\"]\n", - "x_mat = out[\"x_matrix\"]\n", - "\n", - "m = size(r_mat)[2]\n", - "\n", - "for tt in 1:3\n", - " current_costs = [ff(BlockArray(x_mat[:,tt], repeat([3], outer = 10)),r_mat[:,tt]) for ff in fs]\n", - " print(current_costs)\n", - "end" - ] - }, - { - "cell_type": "code", - "execution_count": 71, - "id": "430fbb92", - "metadata": {}, - "outputs": [ - { - "ename": "LoadError", - "evalue": "UndefVarError: `current_costs` not defined", - "output_type": "error", - "traceback": [ - "UndefVarError: `current_costs` not defined", - "" - ] - } - ], - "source": [ - "current_costs" - ] - } - ], - "metadata": { - "kernelspec": { - "display_name": "Julia 1.9.3", - "language": "julia", - "name": "julia-1.9" - }, - "language_info": { - "file_extension": ".jl", - "mimetype": "application/julia", - "name": "julia", - "version": "1.9.3" - } - }, - "nbformat": 4, - "nbformat_minor": 5 -} diff --git a/plotting.ipynb b/plotting.ipynb deleted file mode 100644 index 0eb637e..0000000 --- a/plotting.ipynb +++ /dev/null @@ -1,1607 +0,0 @@ -{ - "cells": [ - { - "cell_type": "code", - "execution_count": 1, - "id": "125e70da", - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "compute_stage_2" - ] - }, - "execution_count": 1, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "using Symbolics\n", - "using GamesVoI\n", - "using Plots\n", - "using DataFrames\n", - "include(\"experiments/tower_defense_exponential.jl\")" - ] - }, - { - "cell_type": "code", - "execution_count": 2, - "id": "da53cbab", - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "0: r = [0.5, 0.25, 0.25]\n", - "dJdr = 0.37666475789932047\n", - "dxdr = 1.7145557146085864\n", - "1: r = [0.6891313024452577, 0.15543434872991768, 0.1554343488248246]\n", - "dJdr = 0.36598149728969936\n", - "dxdr = 1.2697983630187428\n", - "2: r = [0.814790826608705, 0.09260458662946511, 0.09260458676182978]\n", - "dJdr = 0.38332028361487397\n", - "dxdr = 1.0783897236914317\n", - "3: r = [0.9087966461474695, 0.04560167683577082, 0.04560167701675982]\n", - "dJdr = 0.40904826583428666\n", - "dxdr = 0.9680014374258179\n", - "4: r = [0.9830964654744928, 0.008451767140653071, 0.008451767384854286]\n", - "dJdr = 0.43710827761923615\n", - "dxdr = 0.8951162975661063\n", - "5: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444414\n", - "dxdr = 0.8799943913826468\n", - "6: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826473\n", - "7: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "8: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "9: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "10: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "11: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "12: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "13: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "14: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "15: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "16: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "17: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "18: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "19: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "20: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "21: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "22: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "23: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "24: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "25: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "26: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "27: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "28: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "29: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "30: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "31: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "32: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "33: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "34: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "35: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "36: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "37: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "38: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "39: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "40: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "41: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "42: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "43: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "44: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "45: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "46: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "47: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "48: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826461\n", - "49: r = [1.0, 0.0, 0.0]\n", - "dJdr = 0.4444444444444416\n", - "dxdr = 0.8799943913826437\n", - "50: r = [1.0, 0.0, 0.0]\n", - "50: r = [1.0, 0.0, 0.0]\n" - ] - }, - { - "data": { - "text/plain": [ - "Dict{String, AbstractArray{Float64}} with 4 entries:\n", - " \"x_matrix\" => [0.131 0.072643 … 4.0522e-14 4.0522e-14; 0.4345 0.463678 … 0.5 …\n", - " \"x\" => [4.0522e-14, 0.5, 0.5, 1.0, 3.90652e-12, 3.90652e-12, 0.0, 1.0,…\n", - " \"r\" => [1.0, 0.0, 0.0]\n", - " \"r_matrix\" => [0.689131 0.814791 … 1.0 1.0; 0.155434 0.0926046 … 0.0 0.0; 0.1…" - ] - }, - "execution_count": 2, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "prior = [1/3, 1/3,1/3]\n", - "omega_params = [[.02,0.8,0.8],[0.8,0.4,0.8],[0.8,0.8,0.7]]\n", - "r_init =[0.5,0.25,0.25]\n", - "\n", - "out = solve_r(prior,omega_params,r_init=r_init, return_states=true)" - ] - }, - { - "cell_type": "markdown", - "id": "5384684f", - "metadata": {}, - "source": [ - "## Plot Decision Variables over GD" - ] - }, - { - "cell_type": "markdown", - "id": "df4a61d4", - "metadata": {}, - "source": [ - "This code will take my `out` object from the previous code block, and use it to generate a plot showing the evolution of $r$ and $x$ over time." - ] - }, - { - "cell_type": "code", - "execution_count": 3, - "id": "fb57b110", - "metadata": {}, - "outputs": [ - { - "data": { - "image/png": "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", - "image/svg+xml": [ - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n" - ], - "text/html": [ - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - " \n", - " \n", - " \n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n" - ] - }, - "execution_count": 3, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "x_matrix = out[\"x_matrix\"]\n", - "r_matrix = out[\"r_matrix\"]\n", - "titles = [\"r (Stage 1)\",\"P1, s1=0\",\"P1, s1=1\",\"P1, s1=2\",\"P1, s1=3\",\n", - " \"P2, (s1,w)=(0,1)\",\"P2, (s1,w)=(0,2)\",\"P2, (s1,w)=(0,3)\",\n", - " \"P2, (s1,w)=(1,1)\",\"P2, (s1,w)=(2,2)\",\"P2, (s1,w)=(3,3)\"]\n", - "index_map = [1,2,6,7,8,3,9,4,10,5,11] ## reorder states/titles in graph\n", - "# Assuming `x_matrix` is your n x m matrix\n", - "n, m = size(x_matrix)\n", - "num_vectors = 10 # Number of 3-vectors\n", - "vector_length = 3 # Length of each vector\n", - "@assert n == num_vectors*vector_length \"Provided dimensions do not match\";\n", - "\n", - "data = []\n", - " \n", - "push!(data, hcat(1:m,transpose(r_matrix)))\n", - "\n", - "for i in 1:num_vectors ## add Stage 2 decision variables\n", - " # Extract each 3-vector and reshape it\n", - " start_row = (i - 1) * vector_length + 1\n", - " end_row = start_row + vector_length - 1\n", - " vector_matrix = x_matrix[start_row:end_row, :]\n", - "\n", - " # Convert to DataFrame\n", - " new_matrix = hcat(1:m,transpose(vector_matrix))\n", - " \n", - " push!(data, new_matrix)\n", - "end\n", - "\n", - "\n", - "### PLOTTING\n", - "\n", - "using Plots\n", - "using DataFrames\n", - "\n", - "# Assuming `data` is your reshaped data suitable for plotting\n", - "# Each element in `data` is a DataFrame with columns representing the components of a 3-vector and rows representing timesteps\n", - "\n", - "plots = []\n", - "\n", - "\n", - "\n", - "for i in 1:11 # Assuming 10 such 3-vectors\n", - " j = index_map[i]\n", - " mat = data[j] # Your DataFrame for each 3-vector\n", - " if i>1\n", - " p = plot(mat[:,1],mat[:,2:4], xlabel=\"Time\", ylabel=\"Probability\", title=titles[j],ylimits=(0,1),xlimits=(0,50))\n", - " else\n", - " p = plot(mat[:,1],mat[:,2:4], xlabel=\"Time\", ylabel=\"Scouts\", title=titles[j],ylimits=(0,1),xlimits=(0,50))\n", - " end\n", - " push!(plots, p)\n", - " if i==1\n", - " push!(plots,plot(legend=false,grid=false,foreground_color_subplot=:white)) \n", - " end\n", - " if i>2 && i<5\n", - " push!(plots,plot(legend=false,grid=false,foreground_color_subplot=:white)) \n", - " end\n", - "end\n", - "\n", - "# Combine all subplots into one figure\n", - "plot(plots..., layout=(8, 2), legend=true) # Adjust layout as needed\n", - "plot!(size = (800, 1400))" - ] - }, - { - "cell_type": "markdown", - "id": "4c1fe1c7", - "metadata": {}, - "source": [ - "## Plot Decisions/Costs over GD (under construction)" - ] - }, - { - "cell_type": "markdown", - "id": "1abda50f", - "metadata": {}, - "source": [ - "This section is still under construction. I wanted a visualization that included the costs, over time, as well. This would be useful for debugging." - ] - }, - { - "cell_type": "code", - "execution_count": null, - "id": "257b8741", - "metadata": {}, - "outputs": [], - "source": [ - "game,fs = build_stage_2(prior,omega_params)" - ] - }, - { - "cell_type": "code", - "execution_count": 92, - "id": "e883a39d", - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "(1, 10)[2.3930842546822988e-32, 3.000000000000001, 3.0, 3.0, 0.6666666666730381, 0.6666666700460344, 0.6666666666392559, 2.2281153200039965, 2.697318212713702, 2.9428082063440497][6.958731016193841e-32, 3.000000000000001, 3.0, 3.0, 0.6666666927151942, 0.6666666734970208, 0.6666670438042436, 2.2281153181224544, 2.697318211960431, 2.9428081736570926][2.2177776286658088e-31, 3.000000000000001, 3.0, 3.0, 0.6666666672499614, 0.666666667057768, 0.6666666666712378, 2.2281153199624795, 2.697318213276322, 2.942808206342194]" - ] - }, - { - "data": { - "text/plain": [ - "3×10 Matrix{Float64}:\n", - " 2.39308e-32 3.0 3.0 3.0 0.666667 … 0.666667 2.22812 2.69732 2.94281\n", - " 6.95873e-32 3.0 3.0 3.0 0.666667 0.666667 2.22812 2.69732 2.94281\n", - " 2.21778e-31 3.0 3.0 3.0 0.666667 0.666667 2.22812 2.69732 2.94281" - ] - }, - "execution_count": 92, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "# game,fs = build_stage_2(prior,omega_params)\n", - "r_mat = out[\"r_matrix\"]\n", - "x_mat = out[\"x_matrix\"]\n", - "\n", - "cost_matrix = 1:10\n", - "cost_matrix = cost_matrix'\n", - "print(size(cost_matrix))\n", - "\n", - "m = size(r_mat)[2]\n", - "\n", - "for tt in 1:3\n", - " current_costs = [ff(BlockArray(x_mat[:,tt], repeat([3], outer = 10)),r_mat[:,tt]) for ff in fs]\n", - " cost_matrix = vcat(cost_matrix, current_costs')\n", - " print(current_costs)\n", - "end\n", - "cost_matrix = cost_matrix[2:end,:]" - ] - }, - { - "cell_type": "code", - "execution_count": 91, - "id": "9a31ab1c", - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "3×10 Matrix{Float64}:\n", - " 2.39308e-32 3.0 3.0 3.0 0.666667 … 0.666667 2.22812 2.69732 2.94281\n", - " 6.95873e-32 3.0 3.0 3.0 0.666667 0.666667 2.22812 2.69732 2.94281\n", - " 2.21778e-31 3.0 3.0 3.0 0.666667 0.666667 2.22812 2.69732 2.94281" - ] - }, - "execution_count": 91, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [] - }, - { - "cell_type": "code", - "execution_count": null, - "id": "e0da5670", - "metadata": {}, - "outputs": [], - "source": [ - "matrix = out[\"x_matrix\"]\n", - "titles = [\"P1, s1=0\",\"P1, s1=1\",\"P1, s1=2\",\"P1, s1=3\",\n", - " \"P2, (s1,w)=(0,1)\",\"P2, (s1,w)=(0,2)\",\"P2, (s1,w)=(0,3)\",\n", - " \"P2, (s1,w)=(1,1)\",\"P2, (s1,w)=(2,2)\",\"P2, (s1,w)=(3,3)\"]\n", - "\n", - "# Assuming `matrix` is your n x m matrix\n", - "n, m = size(matrix)\n", - "num_vectors = 10 # Number of 3-vectors\n", - "vector_length = 3 # Length of each vector\n", - "@assert n == num_vectors*vector_length \"Provided dimensions do not match\";\n", - "\n", - "data_states = []\n", - "for i in 1:num_vectors\n", - " # Extract each 3-vector and reshape it\n", - " start_row = (i - 1) * vector_length + 1\n", - " end_row = start_row + vector_length - 1\n", - " vector_matrix = matrix[start_row:end_row, :]\n", - "\n", - " # Convert to DataFrame\n", - " new_matrix = hcat(1:m,transpose(vector_matrix))\n", - " \n", - " push!(data_states, new_matrix)\n", - "end\n", - "\n", - "\n", - "### PLOTTING\n", - "\n", - "using Plots\n", - "using DataFrames\n", - "\n", - "# Assuming `data` is your reshaped data suitable for plotting\n", - "# Each element in `data` is a DataFrame with columns representing the components of a 3-vector and rows representing timesteps\n", - "\n", - "plots = []\n", - "for i in 1:10 # Assuming 10 such 3-vectors\n", - " mat = data[i] # Your DataFrame for each 3-vector\n", - " p = plot(mat[:,1],mat[:,2:4], xlabel=\"Time\", ylabel=\"Probability\", title=titles[i],ylimits=(0,1),xlimits=(0,50))\n", - " push!(plots, p)\n", - "end\n", - "\n", - "# Combine all subplots into one figure\n", - "plot(plots..., layout=(5, 2), legend=true) # Adjust layout as needed\n", - "plot!(size = (800, 1000))" - ] - }, - { - "cell_type": "code", - "execution_count": 39, - "id": "9dfdc276", - "metadata": {}, - "outputs": [ - { - "ename": "LoadError", - "evalue": "KeyError: key \"xmatrix\" not found", - "output_type": "error", - "traceback": [ - "KeyError: key \"xmatrix\" not found", - "", - "Stacktrace:", - " [1] getindex(h::Dict{String, AbstractArray{Float64}}, key::String)", - " @ Base ./dict.jl:484", - " [2] top-level scope", - " @ In[39]:3" - ] - } - ], - "source": [ - "\n", - "\n" - ] - }, - { - "cell_type": "code", - "execution_count": 38, - "id": "2f8846a0", - "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "10-element Vector{Function}:\n", - " #419 (generic function with 1 method)\n", - " #422 (generic function with 1 method)\n", - " #422 (generic function with 1 method)\n", - " #422 (generic function with 1 method)\n", - " #424 (generic function with 1 method)\n", - " #424 (generic function with 1 method)\n", - " #424 (generic function with 1 method)\n", - " #426 (generic function with 1 method)\n", - " #426 (generic function with 1 method)\n", - " #426 (generic function with 1 method)" - ] - }, - "execution_count": 38, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "fs" - ] - }, - { - "cell_type": "code", - "execution_count": 73, - "id": "42c76442", - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "[2.3930842546822988e-32, 3.000000000000001, 3.0, 3.0, 0.6666666666730381, 0.6666666700460344, 0.6666666666392559, 2.2281153200039965, 2.697318212713702, 2.9428082063440497][6.958731016193841e-32, 3.000000000000001, 3.0, 3.0, 0.6666666927151942, 0.6666666734970208, 0.6666670438042436, 2.2281153181224544, 2.697318211960431, 2.9428081736570926][2.2177776286658088e-31, 3.000000000000001, 3.0, 3.0, 0.6666666672499614, 0.666666667057768, 0.6666666666712378, 2.2281153199624795, 2.697318213276322, 2.942808206342194]" - ] - } - ], - "source": [ - "r_mat = out[\"r_matrix\"]\n", - "x_mat = out[\"x_matrix\"]\n", - "\n", - "m = size(r_mat)[2]\n", - "\n", - "for tt in 1:3\n", - " current_costs = [ff(BlockArray(x_mat[:,tt], repeat([3], outer = 10)),r_mat[:,tt]) for ff in fs]\n", - " print(current_costs)\n", - "end" - ] - }, - { - "cell_type": "code", - "execution_count": 71, - "id": "a9844ca3", - "metadata": {}, - "outputs": [ - { - "ename": "LoadError", - "evalue": "UndefVarError: `current_costs` not defined", - "output_type": "error", - "traceback": [ - "UndefVarError: `current_costs` not defined", - "" - ] - } - ], - "source": [ - "current_costs" - ] - } - ], - "metadata": { - "kernelspec": { - "display_name": "Julia 1.9.2", - "language": "julia", - "name": "julia-1.9" - }, - "language_info": { - "file_extension": ".jl", - "mimetype": "application/julia", - "name": "julia", - "version": "1.9.2" - } - }, - "nbformat": 4, - "nbformat_minor": 5 -}