Surface forces #488

moritzgubler · 2024-06-12T19:57:47Z

This is my implementation of the idea I had of calculating forces with surface integrals. It works quite well and the forces are more accurate. Also, mrchem struggles to compute forces when the world precision is 1e-7 or smaller. Then it requires a lot of memory (more than 32 GB for a H2 molecule). This is not the case with my approach . Geometry optimizations seem to be reliable if the stopping criterion is chosen 10 * world_prec.

I have already put all the parameters into the parser, so it is quite simple to test and run my code (look for the section "- name: Forces" in the template.yaml input parser file).

At the moment, there is two things that need to be improved:

I did not figure out a way to get access to the exchange correlation density and potential, so I had to change the xc_operator classes a bit in order to get access to them. It works now, but I did it a bit hacky.
I need the charge density, the electrostatic potential and the exchange correlation density and potential (the total one in case of an unrestricted calculation) to calculate the forces. A lot of the time when calculating the forces is spent recalculating these objects because I did not figure out a clean way to reuse them from the last SCF iteration. The time savings will probably not impact the entire SCF calculation significantly.
At the moment the method works only for LDA (this is also why some tests are failing). Extending the method to GGA would be super easy, the XC stress for GGAs is given here: Evaluation of exchange-correlation energy, potential, and stress Phys. Rev. B 64, 165110 – Published 4 October 2001
I am not sure how to calculate the exchange stress for hartree fock exchange. It shouldn't be too complicated but in a quick search I did not find a reference that derives it.

@stigrj Do you think my changes in the xc operator classes are fine.

codecov · 2024-07-31T15:31:23Z

Codecov Report

Attention: Patch coverage is 2.54735% with 1492 lines in your changes missing coverage. Please review.

Project coverage is 62.59%. Comparing base (df39ee2) to head (13b89d8).
Report is 2 commits behind head on master.

Files	Patch %	Lines
src/surface_forces/SurfaceForce.cpp	0.00%	220 Missing ⚠️
src/surface_forces/xcStress.cpp	0.00%	125 Missing ⚠️
src/surface_forces/detail/lebedev_utils.cpp	0.00%	107 Missing ⚠️
src/surface_forces/LebedevData.cpp	0.00%	75 Missing ⚠️
src/surface_forces/lebedev.cpp	0.00%	36 Missing ⚠️
src/surface_forces/detail/Lebedev_021_0170.hpp	0.00%	33 Missing ⚠️
src/surface_forces/detail/Lebedev_023_0194.hpp	0.00%	33 Missing ⚠️
src/surface_forces/detail/Lebedev_027_0266.hpp	0.00%	33 Missing ⚠️
src/surface_forces/detail/Lebedev_035_0434.hpp	0.00%	33 Missing ⚠️
src/surface_forces/detail/Lebedev_047_0770.hpp	0.00%	33 Missing ⚠️
... and 31 more

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #488      +/-   ##
==========================================
- Coverage   68.66%   62.59%   -6.08%     
==========================================
  Files         194      233      +39     
  Lines       15287    16790    +1503     
==========================================
+ Hits        10497    10509      +12     
- Misses       4790     6281    +1491

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

moritzgubler · 2024-08-01T14:40:09Z

Hi @robertodr @ilfreddy
I implemented now everything I wanted to, cleaned up to code and added a bit of documentation. The new method to calculate the forces now works for LDA and GGA functionals for both restricted and unrestricted calculations. The pulll request is now ready for review.
Best, Moritz

moritzgubler · 2024-08-06T11:58:16Z

I previously tested GGA calculations only for H2 where the forces where the forces were correct. In more complicated systems this is not the case. (LDA is always correct). I will look into it and let you know once I have solved the problem.

moritzgubler · 2024-08-07T11:21:04Z

The GGA bug is fixed now, I forgot to add the divergence terms in the xc potential for the GGA case. They have been added now.
MPI does not work anymore, it segfaults in the calculation of the kinetic stress. I suspect, that I try to access orbitals in an orbitalvector that are stored on another rank. @gitpeterwind Do you know what could be the problem?

moritzgubler · 2024-08-07T11:21:45Z

src/surface_forces/SurfaceForce.cpp

+            n1 = nablaPhi[iOrb][0].real().evalf(pos);
+            n2 = nablaPhi[iOrb][1].real().evalf(pos);
+            n3 = nablaPhi[iOrb][2].real().evalf(pos);


@gitpeterwind These are the lines that cause the segfault

In an OrbitalVector, not all orbitals are directly available to all MPI processes. You should do something like:

if (mrcpp::mpi::my_orb(phi_i)) { pos[0] = gridPos(i, 0); pos[1] = gridPos(i, 1); pos[2] = gridPos(i, 2); n1 = nablaPhi[iOrb][0].real().evalf(pos); n2 = nablaPhi[iOrb][1].real().evalf(pos); n3 = nablaPhi[iOrb][2].real().evalf(pos); stress[i](0, 0) -= occ * n1 * n1; stress[i](1, 1) -= occ * n2 * n2; stress[i](2, 2) -= occ * n3 * n3; stress[i](0, 1) -= occ * n1 * n2; stress[i](0, 2) -= occ * n1 * n3; stress[i](1, 2) -= occ * n2 * n3; } } }

And then use mrcpp::mpi::allreduce_vector to collect all the results. It is a bit difficult with the way you have defined the stress vector. But if you cannot easily define it as a DoubleVector, or DoubleMatrix , I can try to give a detailed way to do it (one way is simply to copy all the values into a DoubleVector, the do the allreduce operation, and copy back).

That makes sense, thanks for your quick answer. I will use arrays for the stress tensor.

@gitpeterwind I think it might be a bit more complicated, I need

nablaPhi[iOrb][0] nablaPhi[iOrb][1] nablaPhi[iOrb][2]

all on the same rank. How can I make sure that this is the case?

I think it is correct: the last [0],[1],[2] picks out one OrbitalVector, and iOrb only determines the rank. Or did I misunderstood?

No, nablaPhi[0][0] contains the x component of the gradient of the zeroth orbital. nablaPhi[0] is an orbital vector that contains the x, y and z component of the gradient of orbital 0.

Adding the following print to the loop

for (int i = 0; i < Phi.size(); i++) { nablaPhi.push_back(nabla(Phi[i])); nablaPhi[i].distribute(); if (mrcpp::wrk_rank == 0){ std::cout << "gradient of orbital " << i << " stored on rank: x: " << nablaPhi[i][0].getRank() << " y: " << nablaPhi[i][1].getRank() << " z: " << nablaPhi[i][2].getRank() << std::endl; } }

results in:

gradient of orbital 0 stored on rank: x: 0 y: 1 z: 2 gradient of orbital 1 stored on rank: x: 0 y: 1 z: 2 gradient of orbital 2 stored on rank: x: 0 y: 1 z: 2 gradient of orbital 3 stored on rank: x: 0 y: 1 z: 2

but I need all components of the gradient on the same rank

The OrbitalVector class expects some properties, like that the size is the "usual" number of orbitals, and that each MPI process only access its own indices. If a vector has only have 3 orbitals, you should not define it as an OrbitalVector (and "distribute them is meaningless). Can you simply define std::vector<std::vector<Orbital>> nablaPhi(3); ? or do you need some OrbitalVector specific properties? Still Phi[i] will only be defined on the MPI process with the right rank.
Anyway, if you use the if (mrcpp::mpi::my_orb(phi_i)) { , do you still get the seg fault?

This is a good idea. I thought that this wouldn't work because nabla(Phi) returns an orbitalvector. I tried your suggestion and it compiled. I will try to solve the problem like that, thanks for your help

There was a bug in the master code. A PR has been sent

moritzgubler · 2024-08-07T11:22:20Z

src/surface_forces/SurfaceForce.cpp

+    std::vector<mrchem::OrbitalVector> nablaPhi;
+    mrchem::OrbitalVector hessRho = hess(rho);
+    for (int i = 0; i < Phi.size(); i++) {
+        nablaPhi.push_back(nabla(Phi[i]));
+    }
+


Here I create the vector that contains the gradient of all orbitals

I think it is necessary (at least it won't harm) to add:

std::vector<mrchem::OrbitalVector> nablaPhi; mrchem::OrbitalVector hessRho = hess(rho); for (int i = 0; i < Phi.size(); i++) { nablaPhi.push_back(nabla(Phi[i])); } nablaPhi.distribute();

( distribute will tell each element in the vector which rank it has)

ok, i will add this

moritzgubler · 2024-08-07T11:24:47Z

I can recreate the error with this input:

world_prec = 1.0e-4
world_unit = bohr

WaveFunction {
  method = blyp
  restricted = false
}

Properties {
  geometric_derivative = true
}
Forces {
    method = "surface_integrals"
    #method = 'hellmann_feynman'
    surface_integral_precision = medium
}

Molecule {
$coords
H 0.86387418 -0.02789966 0.13501906
Li -0.44463806 -0.04858572 -0.04959119
$end
}

I start the simulation with:

mrchem --dryrun hli.inp
mpirun -np 4 mrchem.x hli.json

ilfreddy · 2024-08-07T14:43:16Z

@moritzgubler would you mind adding also some testing to your code? According to the code coverage report, your patch is only marginally covered. Tests are essential to make sure the code does not get broken by others later on.

ilfreddy · 2024-08-07T14:47:58Z

src/qmoperators/one_electron/HessianOperator.h

Could you add some class documentation?

@ilfreddy I added some documentation

ilfreddy

Just a general comment from my side: could you add some extra testing:

There is no unit-testing as far as I can see
There is only one regression test

Some classes could also benefit from some additional documentation.

As for the details of the code, I guess it is best if Stig or Peter comment on those.

moritzgubler · 2024-08-13T11:43:46Z

I added a test, is there a way to look at the codecov report?

moritzgubler added 30 commits April 16, 2024 09:28

density wrong

b498633

better plots

ac6dd6d

add new directory

ba79c2c

add docs

c341527

add header

0c65c1d

move stuff around

d3fec08

it compiles

dba6fd8

maxwell test running

2eee72b

maxwell stress correct

45a809b

xc extraction working

5182024

add hessian operator

25a412c

add test for kinetic stress

84b1d0f

add other template

944ff7f

change types

69a3990

working for tiny radius

6a6b6d9

surface force working for one orbital

139649f

cleanup surface force function

dac3be0

multiple orbitals fix

ba24584

delete unused stuff

9115c52

prepare for tests

ff7a1ce

more points

e402bbc

switch to surface forces and debug info

0cf22e0

fix initialization of nabla

ea6e9f8

add check

90ad7c7

reduce number of grid points

9dfc7ba

average over different radii

2978a21

more averaging

7b0f913

pass operators

098b59c

add timer

41fc2d3

better radii

f52ed1e

moritzgubler added 3 commits August 1, 2024 12:22

remove averaging feature

be22297

remove averaging parameters

642c71b

update input parser

6f5d1c2

moritzgubler marked this pull request as ready for review August 1, 2024 14:36

output more digits in xyz file

51417d4

moritzgubler added 5 commits August 7, 2024 09:51

make xc potentials visible

210cff8

make xc operator to get wavelet represantation of xc potential

cf2fcc6

use wavelet represenatation for xc potential in gga case

6c75108

add new radius_factor keyword.

4992efd

remove prints

6fa276a

moritzgubler commented Aug 7, 2024

View reviewed changes

ilfreddy reviewed Aug 7, 2024

View reviewed changes

moritzgubler added 9 commits August 8, 2024 15:01

remove orbitalvectors

64bf02b

more prints

f3552f2

turn on xc stress again

e2f1395

Merge branch 'master' into surface_forces_fix_MPI

a32c309

debugging

34c9bf0

remove prints

2f44450

mpi working

1dca98f

add test

fc49bac

reorganize calculation of nuclear gradients

9408238

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Surface forces #488

Surface forces #488

moritzgubler commented Jun 12, 2024 •

edited

Loading

codecov bot commented Jul 31, 2024

moritzgubler commented Aug 1, 2024

moritzgubler commented Aug 6, 2024

moritzgubler commented Aug 7, 2024

moritzgubler Aug 7, 2024

gitpeterwind Aug 7, 2024 •

edited

Loading

moritzgubler Aug 7, 2024

moritzgubler Aug 7, 2024

gitpeterwind Aug 7, 2024

moritzgubler Aug 8, 2024

moritzgubler Aug 8, 2024

gitpeterwind Aug 8, 2024

moritzgubler Aug 8, 2024

gitpeterwind Aug 11, 2024

moritzgubler Aug 7, 2024

gitpeterwind Aug 7, 2024 •

edited

Loading

moritzgubler Aug 7, 2024

moritzgubler commented Aug 7, 2024

ilfreddy commented Aug 7, 2024

ilfreddy Aug 7, 2024

moritzgubler Aug 12, 2024

ilfreddy left a comment

moritzgubler commented Aug 13, 2024

Surface forces #488

Are you sure you want to change the base?

Surface forces #488

Conversation

moritzgubler commented Jun 12, 2024 • edited Loading

codecov bot commented Jul 31, 2024

Codecov Report

moritzgubler commented Aug 1, 2024

moritzgubler commented Aug 6, 2024

moritzgubler commented Aug 7, 2024

Choose a reason for hiding this comment

gitpeterwind Aug 7, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

gitpeterwind Aug 7, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

moritzgubler commented Aug 7, 2024

ilfreddy commented Aug 7, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ilfreddy left a comment

Choose a reason for hiding this comment

moritzgubler commented Aug 13, 2024

moritzgubler commented Jun 12, 2024 •

edited

Loading

gitpeterwind Aug 7, 2024 •

edited

Loading

gitpeterwind Aug 7, 2024 •

edited

Loading