How can I use this package to solve a finite horizon <a href="https://benjaminmoll.com

There is a Julia translation of Ben Moll's Matlab code to Julia <a href="https://githu

Finite horizon consumption-saving problem about econpdes.jl HOT 2 CLOSED

matthieugomez commented on May 13, 2024

Finite horizon consumption-saving problem

from econpdes.jl.

Comments (2)

ptoche commented on May 13, 2024 1

There is a Julia translation of Ben Moll's Matlab code to Julia here. It does not use EconPDE.jl, but can at least get you started.

from econpdes.jl.

matthieugomez commented on May 13, 2024 1

Yes, it is possible: you simply need to add a fourth argument in pdesolve for the time grid you want to use. For instance

m = AchdouHanLasryLionsMollModel()
distribution = Gamma(2 * m.κy * m.ybar / m.σy^2, m.σy^2 / (2 * m.κy))
stategrid = OrderedDict(
        :y => range(quantile(distribution, 0.001), quantile(distribution, 0.999), length = 10), 
        :a =>  range(m.amin, m.amax, length = 100)
)
# value function at terminal time
yend = OrderedDict(:v => [max(a + y)^(1-m.γ)/(1-m.γ) for y in stategrid[:y], a in stategrid[:a]]) 
# time grid with terminal value at 100 years
τs = range(0, stop = 100, step = 1)
# solve pde
ys, results, distances = pdesolve(m, stategrid, yend, τs)