Merge branch '8-simulation-update' into 'main'

Resolve "Simulation update"

Closes #8

See merge request !11

pycall_test.jl

@@ -4,10 +4,13 @@ Pkg.activate(".")
 using Turing
 using PyCall
 using Plots
+using LinearAlgebra
+using Optim
 
 PyCall.python
 # Import the SWESolver class from the utils module
 @pyimport swe_wrapper as swe
+@pyimport swe_wrapper.utils as utils
 using HDF5 # Has to imported after my module otherwise pycall cannot find swe_wrapper
 struct observation_data
     t::Array{Float64}
@@ -28,73 +31,80 @@ function load_observation_data(file_path::String)
 end
 
 file_path = "./data/toy_measurement/simulation_data.h5"
+measurement = "./experiment_data/Data_Sensors_orig/With_Bathymetry/Heat1.txt"
+mean_bc = "./experiment_data/Data_Sensors_orig/With_Bathymetry/mean_bc.txt"
 observation = load_observation_data(file_path)
 
-xbounds = (0., 10.)
-nx =  64
-tend = 10
-timestep = 1e-3
+dataObj = utils.data(measurement)
+
+# lbc1 = swe.leftbc(measurement)
+# lbc2 = swe.leftbc(mean_bc)
+bathy = swe.rampFunc
+# gaussian = swe.gaussian_bathymetry
+# println(norm(lbc1.f(observation.x) - lbc2.f(observation.x)))
+# cost(x) = sum((bathy(observation.x) - gaussian(observation.x, x)).^2)
+# res = optimize(cost, [4., 0.1])
+# println(Optim.minimizer(res))
+# println(Optim.minimum(res))
+# plot(observation.x, observation.b)
+# plot!(observation.x, gaussian(observation.x, (4., 0.1)))
+# display(plot!(observation.x, bathy(observation.x)))
+# println(norm(gaussian(observation.x, (4., 0.15)) - bathy(observation.x)))
+
+xbounds = (1.5, 15.)
+nx =  100
+total_t = 10
+tstart = 32.
+timestep = 5e-5
+timestepstr = "5e-5"
 g = 9.81
 kappa = 0.2
 dealias = 3/2
-bathy_peak = (5,1) #rand(MvNormal(zeros(130), 1),1)
-#b = observation.b[1:2:end-2]
-b = zeros(64)
-b[1] = rand(Normal(0,0.01),1)[1]
-for i in 2:64
-    b[i] = rand(Normal(b[i-1],0.01),1)[1]
-end
-function create_exponential_decay_matrix(n, decay_factor)
-    A = zeros(n, n)
-    for i in 1:n
-        for j in 1:n
-            A[i, j] = exp(-1/decay_factor * abs(i - j))
-        end
-    end
-    return A
-end
+problemtype = "waterchannel"
+
+# Create SWE solver and calculate the solution
+solver = swe.SWESolver(xbounds, timestep, nx, total_t, tstart=tstart, g=g, kappa=kappa, dealias=dealias,  problemtype=problemtype)
+x = solver.domain.x
+b = bathy(x)
+# Use the solver to simulate the shallow water equation with the sampled peak
+@time sensor_sim, t, simulation_h, _ = solver.solve(b)
+H = simulation_h .+ b'
+t = t .+ 32
 
-# Create laplacian operator matrix with dirichlet boundary conditions
-function create_laplace_matrix_1d(N)
-    # N ist die Anzahl der Gitterpunkte
-    L = zeros(Float64, N, N)  # Erstellen einer N×N Matrix
+sensor_obs = hcat(dataObj.f[1](t), dataObj.f[2](t), dataObj.f[3](t)).+ 0.3
 
-    # Füllen der Matrix mit den Werten für den Laplace-Operator
-    for i in 1:N
-        if i > 1
-            L[i, i - 1] = 1  # Nachbar links
-        end
-        L[i, i] = -2  # Zentraler Punkt
-        if i < N
-            L[i, i + 1] = 1  # Nachbar rechts
-        end
+error = (sensor_sim .- sensor_obs)
+println(size(error))
+rel_l2_error = [norm(error[:,i])./ norm(sensor_obs[:,i]) for i in 1:size(error,2)]
+println("Relative Error: ", rel_l2_error)
+plot_path = "plots/simulation_$(problemtype)_nx=$(nx)_dt=$(timestepstr)"
+println("Save figure in $(plot_path)? (y/o/[n]): ")
+s = readline()
+if s == "y" || s == "o"
+    if s == "o"
+        println("Enter path: ")
+        plot_path = readline()
     end
-    return 1/N.*L
+    mkpath(plot_path)
+    cd(plot_path)
+    plot(x,t, H, st=:surface, colorbar_title="h [m]")
+    xlabel!("x [m]")
+    ylabel!("t [s]")
+    zlabel!("h [m]")
+    savefig("simulation_h_surface_$(problemtype).png")
+    plot(x,t, H, st=:heatmap, colorbar_title="h [m]")
+    xlabel!("x [m]")
+    ylabel!("t [s]")
+    savefig("simulation_h_heatmap_$(problemtype).png")
+    for i in 1:3
+        plot(t, sensor_obs[:,i], label="sensor $(i+1)")
+        plot!(t, sensor_sim[:,i], label="sensor $(i+1) sim")
+        title!("Relative L2-Error: $(rel_l2_error[i])")
+        savefig("sensor_$(i+1)_observation.png")
+    end
+    plot(t .+ 32, sensor_sim, label=hcat(["sensor $(i)" for i in 2:4]...), xticks=32:2:42, yticks=([0.294:0.002:0.306;],[0.294:0.002:0.306;]))
+    xlabel!("t [s]")
+    ylabel!("h [m]")
+    savefig("sensor_observation_$(problemtype).png")
 end
-
-
-
-decay_factor = 100  # Decay factor for off-diagonals
-A = create_exponential_decay_matrix(nx, decay_factor)
-b = rand(MvNormal(zeros(64), A), 1)
-
-# smoothed prior
-L = create_laplace_matrix_1d(nx)
-gamma = 0.1
-w = rand(MvNormal(zeros(nx), 1), 1)
-b = L\(gamma.*w)
-b = 0.2.* b/maximum(b)
-plot(b)
-savefig("b_rand_smoothed.png")
-# Create SWE solver and calculate the solution
-solver = swe.SWESolver(xbounds, timestep, nx, tend, g, kappa, dealias)
-solver.set_initial_conditions()
-c = 1.2e-3 + minimum(solver.initial_conditions.H)
-println(minimum(solver.initial_conditions.H))
-b=ones(64)*c
-
-solver.set_initial_conditions()
-println(minimum(solver.initial_conditions.H-b))
-# Use the solver to simulate the shallow water equation with the sampled peak
-@time simulation_h, _, t = solver.solve(b)
 println(t[end])
\ No newline at end of file