Firdrake run time errors

[jared-rescale]

Hi, I'm trying to run firedrake on a sample problem

Here is the problem, I can supply input files if needed

############################################ from firedrake import from firedrake.petsc import PETSc from firedrake_adjoint import

import math import numpy as np import time as ti from mpi4py import MPI from scipy import optimize

import solvers from myutils import io, utils

""" This is an example of how to use Spyro to perform Full-waveform inversion"""

-----------------------------------------------------------------------

P R O C E D U R E

-----------------------------------------------------------------------

1. Run forward problem for exact data (this is only done if you don't have real data)

2. Run forward problem for guess data.

3. Compute the gradient of the functional w.r.t. to the control (vp_guess)

4. Call SciPy optimization routine (here we're using LBFGS-B)

-----------------------------------------------------------------------

I N I T I A L I Z E

-----------------------------------------------------------------------

Import the options for your model

from inputdata import model

Create the computational environment

comm = utils.mpi_init(model)

-----------------------------------------------------------------------

F O R W A R D

-----------------------------------------------------------------------

Configure source positions based on the input dictionary.

sources = utils.create_sources(model)

Read in an external mesh and project P-wave velocity onto it

mesh,V = io.ReadMesh(model,comm) vp_exact = io.Interp(model,mesh,guess=False) vp_guess = io.Interp(model,mesh,guess=True)

Compute forward solution for all shots

p_exact = solvers.Leapfrog(model,mesh,comm,vp_exact,sources)

Compute the weights for fast interp to all receiver points

map1,map2 = utils.build_maps(model,mesh,comm)

Compute the interpolation matrix

interp_mat = utils.build_interp_matrix(model,mesh,comm,map1,map2)

vp_iterates = File("vp_iter.pvd")

dj_iterates = File("dj_iter.pvd")

Define a callback function that returns the gradient and functional

temp_vp = Function(V) temp_dj = Function(V) def shots(xi, stops): """ Callback function that returns gradient of control and functional to be optimized using scipy """ stops[0] = COMM_WORLD.bcast(stop[0], root=0) if stops[0] == 0: xi = COMM_WORLD.bcast(xi, root=0) temp_vp.dat.data[:] = xi

lets look at what's going on here

    # vp_iterates.write(temp_vp)
    #
    p_guess = solvers.Leapfrog(model,mesh,comm,temp_vp,sources)
    #
    residual = utils.evaluate_misfit(model,comm,p_exact,p_guess,map1,map2)
    #
    dJ_masked= solvers.Leapfrog_adjoint(model,mesh,comm,temp_vp,p_guess,residual,interp_mat)
    # mask out the water layer
    #dJ_masked = utils.mask_water_layer(model,mesh,dJ)
    temp_dj.dat.data[:]=dJ_masked
    # dj_iterates.write(temp_dj)
    #
    J= utils.compute_functional(model,comm,residual)
else:
    J=0.0
    dJ_masked=0.0
return J, dJ_masked

Initialize a file to visualize the control each iteration

if COMM_WORLD.rank == 0: viz_m_file = File("control.pvd", comm=COMM_SELF) viz_m_file.write(vp_guess)

def derivative_cb(m): """ Plots intermidate control file to disk""" vp_guess.dat.data[:] = m viz_m_file.write(vp_guess)

10. Call the optimization routine

if COMM_WORLD.rank==0: stop = [0] xi = vp_guess.dat.data res = optimize.minimize(shots, xi, args = (stop, ), method = "L-BFGS-B", jac = True, bounds = [(1.0, 10.0) for x in range(len(xi))], callback = derivative_cb, options = {"disp": True,"maxiter": 20}) stop = [1] shots(xi, stop) vp_guess.dat.data[:] = res.x File("final.pvd", comm=COMM_SELF).write(vp_guess) else: stop = [0] xi = vp_guess.dat.data while stop[0] == 0: shots(xi,stop) ############################################ When I run the problem it seems to pass through one iteration fine then errors out, I have included the output below, any suggestions? ############################################# process_output2.log

[jared-rescale]

Out put log didn't paste here is the message

process_output2.log

[jared-rescale]

resolved= needed intelmpi2018 and newer gnu libraries for Microsoft cloud machine images.