Corrections to lsm

Author: aleCombi

src/pricing_methods/least_squares_montecarlo.jl

@@ -39,11 +39,12 @@ end
 Extracts the simulated spot paths from a `Vector` of state vectors. Returns a matrix of size (nsteps, npaths).
 Each column corresponds to a single simulation path.
 """
-function extract_spot_grid(sol)
-    # Each path is a Vector of state vectors; we extract first component at each time step
-    return hcat([getindex.(s.u, 1) for s in sol.u]...)  # size: (nsteps, npaths)
+function extract_spot_grid(sol::CustomEnsembleSolution)
+    # Each s is a solution in sol.solutions, where s.u is a vector of states
+    return hcat([getindex.(s.u, 1) for s in sol.solutions]...)  # size: (nsteps, npaths)
 end
 
+
 function solve(
     prob::PricingProblem{VanillaOption{American, C, Spot}, I},
     method::LSM

src/pricing_methods/montecarlo.jl

@@ -33,10 +33,11 @@ struct BlackScholesExact <: SimulationStrategy
     trajectories
     steps
     kwargs::NamedTuple
+    seeds::Union{Nothing, Vector{Int}}
 end
 
 EulerMaruyama(trajectories, steps; kwargs...) = EulerMaruyama(trajectories, steps, (; kwargs...))
-BlackScholesExact(trajectories, steps=1; kwargs...) = BlackScholesExact(trajectories, steps, (; kwargs...))
+BlackScholesExact(trajectories, steps=1,seeds=nothing; kwargs...) = BlackScholesExact(trajectories, steps, (; kwargs...), seeds)
 
 # ------------------ SDE Problem Builders ------------------
 
@@ -77,7 +78,11 @@ end
 
 function montecarlo_solution(problem::Union{NoiseProblem, SDEProblem}, strategy::S) where {S <: SimulationStrategy}
     dt = problem.tspan[end] / strategy.steps
-    seeds = Base.rand(1:10^9, strategy.trajectories) #generate seeds
+    N = strategy.trajectories
+
+    seeds = strategy isa BlackScholesExact && strategy.seeds !== nothing ? strategy.seeds :
+            Base.rand(1_000_000_000:2_000_000_000, N)
+
     modify = (p, seed, _) -> remake(p; seed=seed)
     custom_prob = CustomEnsembleProblem(problem, collect(seeds), modify)
     return solve_custom_ensemble(custom_prob; dt=dt)

test/least_squares_montecarlo.jl

@@ -15,7 +15,8 @@
     dynamics = LognormalDynamics()
     trajectories = 1
     steps = 100
-    strategy = BlackScholesExact(trajectories, steps; seed=42)  # Deterministic seed
+    seeds = [42]
+    strategy = BlackScholesExact(trajectories, steps, seeds)  # Deterministic seed
     degree = 3
     method = LSM(dynamics, strategy, degree)