# Error in plotting the FEniCS variable over time to generate a plot that represents values at a particular time

**URL:** <https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234>\
**Category:** General\
**Created:** [April 12, 2024, 8:11pm UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234 "2024-04-12T20:11:03Z")\
**Posts on this page:** 16\
**Page:** 1

<div class="post-metadata">

**Author:** ![Ashini](https://avatars.discourse-cdn.com/v4/letter/a/57b2e6/32.png) [@Ashini](https://fenicsproject.discourse.group/u/Ashini)\
**Post date:** [April 12, 2024, 8:11pm UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/1 "2024-04-12T20:11:03Z")

</div>

Hii

I am trying to make a code of flow through a circular cylinder in 3d in FeniCS. I see the code " [Test problem 2: Flow past a cylinder (DFG 2D-3 benchmark) — FEniCSx tutorial](https://jsdokken.com/dolfinx-tutorial/chapter2/ns_code2.html) ". It is written in FeniCSx.

Below, I have attached a part of the syntax. Is it fine to compute the drag and lift coefficients like that? The problem I am facing is in plotting, as `c_drag` contains functions in FEniCS format, i.e., `f_1`. I don’t know how I can plot these graphs corresponding to time, as this list doesn’t contain numbers. I need a plot similar to the one in the link above.

Hope to hear from you soon.

```auto
F = (idt*dot(v,u)*dx - idt*dot(v,u_n)*dx + nu*0.5 * inner(D(u), D(v)) * dx + inner(dot(u_n,nabla_grad(u)), v) * dx - p * div(v) * dx - div(u) * q * dx - dot(f1,v)*dx -f2*q*dx )
n = FacetNormal(mesh)  
dObs = Measure("ds", domain =mesh, subdomain_data= 3)
c_drag = []
c_lift =[]
Time =[]

for z in range(num_steps):
    t = (z+1)*dt
    u_exact.t=t
    solve(F == 0, up, bc, solver_parameters={"newton_solver":{"linear_solver":'mumps'},"newton_solver":{"relative_tolerance":1e-6}})
    u, p, rho = up.split()
    u_t = inner(as_vector((-n[1],n[0],0)),u)
    drag = 2 / 0.1 * (nu * inner(grad(u_t), n) * n[1] - p * n[0]) * dObs
    lift = -2 / 0.1 * (nu * inner(grad(u_t), n) * n[0] + p * n[1]) * dObs
    c_drag.append(drag)
    c_lift.append(lift)
    Time.append(t)
    assign(u_n,u)
    assign(p_n,p)

```

---

<div class="post-metadata">

**Author:** ![dokken](https://yyz2.discourse-cdn.com/free1/user_avatar/fenicsproject.discourse.group/dokken/32/1560_2.png) [@dokken](https://fenicsproject.discourse.group/u/dokken)\
**Post date:** [April 12, 2024, 8:40pm UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/2 "2024-04-12T20:40:07Z")

</div>

> [@Ashini](#):
>
> ```auto
> dObs
> c_drag.append(drag)
> c_lift.append(lift)
> 
> ```

No, this is not sufficient.  
Please look through the rest of the example, as ir shows that you need to call `dolfinx.fem.assemble_scalar(drag)` and do relevant mpi communication if running in parallel.

```python
 # Compute physical quantities
    # For this to work in paralell, we gather contributions from all processors
    # to processor zero and sum the contributions.
    drag_coeff = mesh.comm.gather(assemble_scalar(drag), root=0)
    lift_coeff = mesh.comm.gather(assemble_scalar(lift), root=0)
    if mesh.comm.rank == 0:
        t_u[i] = t
        t_p[i] = t - dt / 2
        C_D[i] = sum(drag_coeff)
        C_L[i] = sum(lift_coeff)

```

---

<div class="post-metadata">

**Author:** ![Ashini](https://avatars.discourse-cdn.com/v4/letter/a/57b2e6/32.png) [@Ashini](https://fenicsproject.discourse.group/u/Ashini)\
**Post date:** [April 12, 2024, 9:01pm UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/3 "2024-04-12T21:01:34Z")

</div>

Sir, i didn’t understand why i need to do this. If i simply append the values and then plot what is the problem ?

---

<div class="post-metadata">

**Author:** ![dokken](https://yyz2.discourse-cdn.com/free1/user_avatar/fenicsproject.discourse.group/dokken/32/1560_2.png) [@dokken](https://fenicsproject.discourse.group/u/dokken)\
**Post date:** [April 12, 2024, 9:37pm UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/4 "2024-04-12T21:37:13Z")

</div>

You are not appending values. You are appending a symbolic expression. To convert the symbolic expression into a scalar value, you need to call `assemble_scalar`.

---

<div class="post-metadata">

**Author:** ![Ashini](https://avatars.discourse-cdn.com/v4/letter/a/57b2e6/32.png) [@Ashini](https://fenicsproject.discourse.group/u/Ashini)\
**Post date:** [April 13, 2024, 5:42am UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/5 "2024-04-13T05:42:58Z")

</div>

Please help me to resolve this issue. I am getting error in syntax

```auto
from dolfin import *
from petsc4py import PETSc
import numpy as np
import meshio

F = (idt*dot(v,u)*dx - idt*dot(v,u_n)*dx + nu*0.5 * inner(D(u), D(v)) * dx + inner(dot(u_n,nabla_grad(u)), v) * dx - p * div(v) * dx - div(u) * q * dx - dot(f1,v)*dx -f2*q*dx )
n = FacetNormal(mesh)
dObs = Measure("ds", domain =mesh, subdomain_data= 3)

for z in range(num_steps):
    t = (z+1)*dt
 
    solve(F == 0, up, bc, solver_parameters={"newton_solver":{"linear_solver":'mumps'},"newton_solver":{"relative_tolerance":1e-6}})
    u, p, rho = up.split()
    u_t = inner(as_vector((-n[1],n[0],0)),u)
    drag = 2 / 0.1 * (nu * inner(grad(u_t), n) * n[1] - p * n[0]) * dObs
    lift = -2 / 0.1 * (nu * inner(grad(u_t), n) * n[0] + p * n[1]) * dObs
    if MPI.rank(mesh.mpi_comm()) == 0:
       C_D = np.zeros(num_steps, dtype=PETSc.ScalarType)
       C_L = np.zeros(num_steps, dtype=PETSc.ScalarType)
       t_u = np.zeros(num_steps, dtype=np.float64)
       t_p = np.zeros(num_steps, dtype=np.float64)
    
    drag_coeff = mesh.comm.gather(assemble_scalar(drag), root=0)
    lift_coeff = mesh.comm.gather(assemble_scalar(lift), root=0)
    if mesh.comm.rank == 0:
       t_u[z] = t
       t_p[z] = t - dt / 2
       C_D[z] = sum(drag_coeff)
       C_L[z] = sum(lift_coeff)

```

Error:

```auto
   drag_coeff = mesh.comm.gather(assemble_scalar(drag), root=0)
AttributeError: 'dolfin.cpp.mesh.Mesh' object has no attribute 'comm'

```

---

<div class="post-metadata">

**Author:** ![dokken](https://yyz2.discourse-cdn.com/free1/user_avatar/fenicsproject.discourse.group/dokken/32/1560_2.png) [@dokken](https://fenicsproject.discourse.group/u/dokken)\
**Post date:** [April 13, 2024, 5:48am UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/6 "2024-04-13T05:48:00Z")

</div>

You are using legacy dolfin, not dolfinx as in the tutorial. BI would strongly advice you tol Upgrade.

---

<div class="post-metadata">

**Author:** ![Ashini](https://avatars.discourse-cdn.com/v4/letter/a/57b2e6/32.png) [@Ashini](https://fenicsproject.discourse.group/u/Ashini)\
**Post date:** [April 13, 2024, 5:50am UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/7 "2024-04-13T05:50:00Z")

</div>

Is it possible to write in dolfin legacy without upgrading it to dolfinx? I didn’t use dolfinx

---

<div class="post-metadata">

**Author:** ![dokken](https://yyz2.discourse-cdn.com/free1/user_avatar/fenicsproject.discourse.group/dokken/32/1560_2.png) [@dokken](https://fenicsproject.discourse.group/u/dokken)\
**Post date:** [April 13, 2024, 7:22am UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/8 "2024-04-13T07:22:59Z")

</div>

Yes, you can use legacy dolfin, just note that you are following the tutorial for dolfinx.

'assemble(drag)`and`assemble(lift)` would do the trick.

---

<div class="post-metadata">

**Author:** ![Ashini](https://avatars.discourse-cdn.com/v4/letter/a/57b2e6/32.png) [@Ashini](https://fenicsproject.discourse.group/u/Ashini)\
**Post date:** [April 13, 2024, 10:20am UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/9 "2024-04-13T10:20:04Z")

</div>

Could you refer some link so that I am able to plot drag, lift using dolfin legacy ?

The trick you suggested show an error:

```auto
for z in range(num_steps):
    t = (z+1)*dt
  
    solve(F == 0, up, bc, solver_parameters={"newton_solver":{"linear_solver":'mumps'},"newton_solver":{"relative_tolerance":1e-6}})
    u, p, rho = up.split()
    u_t = inner(as_vector((-n[1],n[0],0)),u)
    drag = 2 / 0.1 * (nu * inner(grad(u_t), n) * n[1] - p * n[0]) * dObs
    lift = -2 / 0.1 * (nu * inner(grad(u_t), n) * n[0] + p * n[1]) * dObs
    if MPI.rank(mesh.mpi_comm()) == 0:
       C_D = np.zeros(num_steps, dtype=PETSc.ScalarType)
       C_L = np.zeros(num_steps, dtype=PETSc.ScalarType)
       t_u = np.zeros(num_steps, dtype=np.float64)
       t_p = np.zeros(num_steps, dtype=np.float64)
    drag_coeff = assemble(drag)
    lift_coeff = assemble(lift)
    if mesh.comm.rank == 0:
       t_u[z] = t
       t_p[z] = t - dt / 2
       C_D[z] = sum(drag_coeff)
       C_L[z] = sum(lift_coeff)

```

error:

```auto
 drag_coeff = assemble(drag)
  File "/usr/lib/petsc/lib/python3/dist-packages/dolfin/fem/assembling.py", line 202, in assemble
    dolfin_form = _create_dolfin_form(form, form_compiler_parameters)
  File "/usr/lib/petsc/lib/python3/dist-packages/dolfin/fem/assembling.py", line 60, in _create_dolfin_form
    return Form(form,
  File "/usr/lib/petsc/lib/python3/dist-packages/dolfin/fem/form.py", line 85, in __init__
    self.set_exterior_facet_domains(subdomains)
TypeError: set_exterior_facet_domains(): incompatible function arguments. The following argument types are supported:
    1. (self: dolfin.cpp.fem.Form, arg0: dolfin.cpp.mesh.MeshFunctionSizet) -> None

Invoked with: <dolfin.fem.form.Form object at 0x7fdd8e074680>, 3

```

---

<div class="post-metadata">

**Author:** ![dokken](https://yyz2.discourse-cdn.com/free1/user_avatar/fenicsproject.discourse.group/dokken/32/1560_2.png) [@dokken](https://fenicsproject.discourse.group/u/dokken)\
**Post date:** [April 13, 2024, 10:31am UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/10 "2024-04-13T10:31:33Z")

</div>

> [@Ashini](#):
>
> `dObs = Measure("ds", domain =mesh, subdomain_data= 3)`

Subdomain data is wrong. Please consult the legacy documentation, or the FEniCS book (2011) or FEniCS tutorial (2016) for help with legacy dolfin.

---

<div class="post-metadata">

**Author:** ![Ashini](https://avatars.discourse-cdn.com/v4/letter/a/57b2e6/32.png) [@Ashini](https://fenicsproject.discourse.group/u/Ashini)\
**Post date:** [April 13, 2024, 11:31am UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/11 "2024-04-13T11:31:04Z")

</div>

I am getting an error regarding MPI. I just want to confirm the way I am computing is fine ?

```auto
from dolfin import *
from ufl import JacobianInverse, indices
from dolfin import project
from petsc4py import PETSc
import numpy as np
import meshio
from mpi4py import MPI

# Initialize MPI
MPI.Init()

# Get the MPI communicator
comm = MPI.COMM_WORLD

# Your code using MPI goes here

# Finalize MPI
MPI.Finalize()

dim = 3        
msh = meshio.read("test.msh")
.
.
.
.
c_drag = []
c_lift =[]
Time =[]

for z in range(num_steps):
    t = (z+1)*dt
    solve(F == 0, up, bc, solver_parameters={"newton_solver":{"linear_solver":'mumps'},"newton_solver":{"relative_tolerance":1e-6}})
    u, p, rho = up.split()
    u_t = inner(as_vector((-n[1],n[0],0)),u)
    drag = 2 / 0.1 * (nu * inner(grad(u_t), n) * n[1] - p * n[0]) * ds(3)
    lift = -2 / 0.1 * (nu * inner(grad(u_t), n) * n[0] + p * n[1]) * ds(3)
    if MPI.rank(mesh.mpi_comm()) == 0:
       C_D = np.zeros(num_steps, dtype=PETSc.ScalarType)
       C_L = np.zeros(num_steps, dtype=PETSc.ScalarType)
       t_u = np.zeros(num_steps, dtype=np.float64)
       t_p = np.zeros(num_steps, dtype=np.float64)
    drag_coeff = assemble(drag)
    lift_coeff = assemble(lift)   
    comm = MPI.COMM_WORLD
    if comm.rank == 0:
       t_u[z] = t
       t_p[z] = t - dt / 2
       C_D[z] = sum(drag_coeff)
       C_L[z] = sum(lift_coeff)

```

error

```auto
hwloc/linux: Ignoring PCI device with non-16bit domain.
Pass --enable-32bits-pci-domain to configure to support such devices
(warning: it would break the library ABI, don't enable unless really needed).
hwloc/linux: Ignoring PCI device with non-16bit domain.
Pass --enable-32bits-pci-domain to configure to support such devices
(warning: it would break the library ABI, don't enable unless really needed).
--------------------------------------------------------------------------
Open MPI has detected that this process has attempted to initialize
MPI (via MPI_INIT or MPI_INIT_THREAD) more than once. This is
erroneous.
--------------------------------------------------------------------------
[ashini-Precision-5820-Tower:36038] *** An error occurred in MPI_Init
[ashini-Precision-5820-Tower:36038] *** reported by process [2019164161,0]
[ashini-Precision-5820-Tower:36038] *** on a NULL communicator
[ashini-Precision-5820-Tower:36038] *** Unknown error
[ashini-Precision-5820-Tower:36038] *** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort,
[ashini-Precision-5820-Tower:36038] *** and potentially your MPI job)

```

---

<div class="post-metadata">

**Author:** ![dokken](https://yyz2.discourse-cdn.com/free1/user_avatar/fenicsproject.discourse.group/dokken/32/1560_2.png) [@dokken](https://fenicsproject.discourse.group/u/dokken)\
**Post date:** [April 13, 2024, 11:39am UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/12 "2024-04-13T11:39:26Z")

</div>

> [@Ashini](#):
>
> ```auto
> # Initialize MPI
> MPI.Init()
> 
> # Get the MPI communicator
> comm = MPI.COMM_WORLD
> 
> # Your code using MPI goes here
> 
> # Finalize MPI
> MPI.Finalize()
> 
> ```

Why are you calling mpi finalize in your code? You shouldn’t do that.

---

<div class="post-metadata">

**Author:** ![Ashini](https://avatars.discourse-cdn.com/v4/letter/a/57b2e6/32.png) [@Ashini](https://fenicsproject.discourse.group/u/Ashini)\
**Post date:** [April 13, 2024, 12:04pm UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/13 "2024-04-13T12:04:58Z")

</div>

PLease help me …How can I do ?

---

<div class="post-metadata">

**Author:** ![dokken](https://yyz2.discourse-cdn.com/free1/user_avatar/fenicsproject.discourse.group/dokken/32/1560_2.png) [@dokken](https://fenicsproject.discourse.group/u/dokken)\
**Post date:** [April 13, 2024, 12:07pm UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/14 "2024-04-13T12:07:01Z")

</div>

Remove the bit of the code that i quoted from your script.

---

<div class="post-metadata">

**Author:** ![Ashini](https://avatars.discourse-cdn.com/v4/letter/a/57b2e6/32.png) [@Ashini](https://fenicsproject.discourse.group/u/Ashini)\
**Post date:** [April 13, 2024, 1:48pm UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/15 "2024-04-13T13:48:42Z")

</div>

It show error in sum

```auto
    u_t = inner(as_vector((-n[1],n[0],0)),u)
    drag = 2 / 0.1 * (nu * inner(grad(u_t), n) * n[1] - p * n[0]) * ds(3)
    lift = -2 / 0.1 * (nu * inner(grad(u_t), n) * n[0] + p * n[1]) * ds(3)
    if MPI.rank(mesh.mpi_comm()) == 0:
       C_D = np.zeros(num_steps, dtype=PETSc.ScalarType)
       C_L = np.zeros(num_steps, dtype=PETSc.ScalarType)
       t_u = np.zeros(num_steps, dtype=np.float64)
       t_p = np.zeros(num_steps, dtype=np.float64)
    drag_coeff = assemble(drag)
    lift_coeff = assemble(lift)   
    if MPI.rank(mesh.mpi_comm()) == 0:
       t_u[z] = t
       t_p[z] = t - dt / 2
       C_D[z] = sum(drag_coeff)
       C_L[z] = sum(lift_coeff)
       assign(p_n,p)
       assign(u_n,u)

```

```auto
Solving nonlinear variational problem.
  Newton iteration 0: r (abs) = 8.808e+01 (tol = 1.000e-10) r (rel) = 1.000e+00 (tol = 1.000e-06)
  Newton iteration 1: r (abs) = 8.754e-14 (tol = 1.000e-10) r (rel) = 9.938e-16 (tol = 1.000e-06)
  Newton solver finished in 1 iterations and 1 linear solver iterations.
Traceback (most recent call last):
  File "/home/ayush/Downloads/aparna/flow_past_cylinder_noslip.py", line 140, in <module>
    C_D[z] = sum(drag_coeff)
TypeError: 'float' object is not iterable

```

---

<div class="post-metadata">

**Author:** ![dokken](https://yyz2.discourse-cdn.com/free1/user_avatar/fenicsproject.discourse.group/dokken/32/1560_2.png) [@dokken](https://fenicsproject.discourse.group/u/dokken)\
**Post date:** [April 13, 2024, 2:55pm UTC](https://fenicsproject.discourse.group/t/error-in-plotting-the-fenics-variable-over-time-to-generate-a-plot-that-represents-values-at-a-particular-time/14234/16 "2024-04-13T14:55:30Z")

</div>

Please read the error messages.  
As drag\_coeff is a float there is no need to call `sum(…)`