Draft: Resolve "Proposal of updates with Langevin equation" by ALF-Import-Bot · Pull Request #592 · ALF-QMC/ALF

ALF-Import-Bot · 2026-02-12T10:47:14Z

In GitLab by @fassaad on Feb 5, 2026, 09:45 UTC:

Anika has done a great job here, and we should merge this into the master once tested by a third party.

Assignees: fassaad

Migrated from GitLab: https://git.physik.uni-wuerzburg.de/ALF/ALF/-/merge_requests/262

…tion

ALF-Import-Bot · 2026-02-12T10:47:18Z

In GitLab by @fassaad on Feb 5, 2026, 09:51 UTC:

added 39 commits

680df87...957545b - 38 commits from branch master
9b196a3 - Merge branch 'master' into 305-proposal-of-updates-with-langevin-equation

Compare with previous version

ALF-Import-Bot · 2026-02-12T10:47:21Z

In GitLab by @fassaad on Feb 5, 2026, 09:58 UTC:

I tested

Propose_MALA        = .true.
Delta_t_MALA_sequential = 0.1
Max_Force_MALA_sequential = 5.0

And this works just fine!

For global_tau_MALA with:

Global_tau_MALA_moves   = .true.
N_Global_tau_MALA       = 1
Delta_t_MALA_global_tau =  0.1
Max_Force_MALA_global_tau = 5.0

I get an error, the issue being that Nt_sequential_end=-1

ALF-Import-Bot · 2026-02-12T10:47:24Z

In GitLab by @fassaad on Feb 5, 2026, 15:00 UTC:

added 1 commit

27ead59 - CLeaned up issue when only Mala_global_tau and no sequential moves are carried out.

Compare with previous version

ALF-Import-Bot · 2026-02-12T10:47:27Z

In GitLab by @fassaad on Feb 8, 2026, 16:25 UTC:

added 1 commit

467053b - Worked on Documentation

Compare with previous version

ALF-Import-Bot · 2026-02-12T10:47:30Z

In GitLab by @fassaad on Feb 8, 2026, 17:03 UTC:

added 1 commit

1cd8972 - WWorking on doc

Compare with previous version

ALF-Import-Bot · 2026-02-12T10:47:33Z

In GitLab by @fassaad on Feb 8, 2026, 19:16 UTC:

added 1 commit

e8fd3e1 - Working on doc for Updating

Compare with previous version

ALF-Import-Bot · 2026-02-12T10:47:36Z

In GitLab by @fassaad on Feb 9, 2026, 06:47 UTC:

added 1 commit

e4a9dbb - Update

Compare with previous version

ALF-Import-Bot · 2026-02-12T10:47:39Z

In GitLab by @fassaad on Feb 9, 2026, 07:50 UTC:

added 1 commit

6a6d3f7 - Worked on the documentation.

Compare with previous version

ALF-Import-Bot · 2026-02-12T10:47:42Z

In GitLab by @fassaad on Feb 9, 2026, 07:57 UTC:

@anika, All seems to be working now. I have also rewritten the documentation for the updating schemes. I have also modified the Hubbard model module such that one can test all the updating schemes (at the exception of the parallel tempering) out of the box, without having to program anything. I did that on the 6 sites lattice with open boundary conditions and consistently obtained the same results with different updating schemes. It would be great if you could go through the documentation. Thanks Fakher.

…tion

joaogci · 2026-02-26T09:36:39Z

Looks good! I think that snippets of code/parameters for Langevin and HMC updates are missing. It would make it more clear on how to use/implement these methods.
Also, there is an issue if you just want to do global tau moves (either normal moves or MALA) and no sequential spin flips. For this, you have to set Nt_sequential_start = 0 and also Nt_sequential_end = 0, and then you can do your simulation with just global tau moves. In this case, the code will produce a warning: "Warning: Nt_sequential_end is smaller than Nt_sequential_start". Is this something that we want to keep?

…tion

…HMC.

fassaad · 2026-03-25T13:04:02Z

Looks good! I think that snippets of code/parameters for Langevin and HMC updates are missing. It would make it more clear on how to use/implement these methods. Also, there is an issue if you just want to do global tau moves (either normal moves or MALA) and no sequential spin flips. For this, you have to set Nt_sequential_start = 0 and also Nt_sequential_end = 0, and then you can do your simulation with just global tau moves. In this case, the code will produce a warning: "Warning: Nt_sequential_end is smaller than Nt_sequential_start". Is this something that we want to keep?

Hi Joao, I have added the snippets of code/parameters for the Langevin and HMC. I think that it is always good if the code issues a warning, and I would keep it like that

johanneshofmann87 · 2026-03-25T20:52:49Z

+          deallocate(forces_0)
+
+        end Subroutine Ham_Langevin_HMC_S0_single_base
+


I assumed S0(phi) = 1/2 phi^2, which is dangerous and likely incorrect as the warning message states. Now I'm thinking it would be better to flip the logic around, i.e., HMC_S0_base builds all forces_0 out of the single version, accompanied by an inefficiency warning asking the user to override it.
Here, it's probably best to terminate and given an error message instead of assuming the simplest gaussian form of S0.

johanneshofmann87 · 2026-03-25T20:55:11Z

 !--------------------------------------------------------------------
-          Subroutine Overide_global_tau_sampling_parameters_base(Nt_sequential_start,Nt_sequential_end,N_Global_tau)
+          Subroutine Overide_global_tau_sampling_parameters_base(Nt_sequential_start,Nt_sequential_end, &
+                           &        N_Global_tau, N_Global_tau_MALA)


We can probably remove this override sampling function now and use the new QMC variables functionality and set those variables according to the model parameters in the hamiltonian directly.

johanneshofmann87 · 2026-03-25T21:08:13Z

+      T0_Proposal_ratio = 1
+
+
+   End Subroutine Global_move


What is the relation between those global moves and MALA updates?

johanneshofmann87 · 2026-03-25T21:09:17Z

+      S0_ratio = S0(Flip_list(1),ntau,Flip_value(1))
+
+
+   end Subroutine Global_move_tau


How does this relate to MALA moves?

johanneshofmann87 · 2026-03-25T21:13:16Z

+      Subroutine Global_MALA_move_base(Flip_list)
+
+         Implicit none
+         Integer                   , INTENT(OUT) :: Flip_list(:,:)


Flip list essentially tags a field to be updated via MALA moves if the entry is set to 1, correct? Wouldn't it be better turning this into a logical array? And change the name to something like perform_MALA or MALA_tag or enable_MALA?

johanneshofmann87 · 2026-03-25T21:14:49Z

+           do n = 1, n1
+              if (OP_V(n,1)%type == 3 ) then
+                 do nt = 1, n2
+                    if ( flip_list(n,nt) == 1 ) then


looks like logical variable would be better here. I struggle reading the code atm.

johanneshofmann87 · 2026-03-25T21:41:51Z

+           Call Metropolis_Langevin%make(.False., .False., MALA, delta_t_MALA_global, MAX_Force_MALA_global, Leapfrog_steps)
        else
-           Call Langevin_HMC%set_Update_scheme(Langevin, HMC )
+           Call Metropolis_Langevin%set_Update_scheme(.False., .False., MALA )


Do we need to keep Langevin_HMC separate from Metropolis_Langevin? I'd understand the overall logic yet...

johanneshofmann87 · 2026-03-25T21:46:16Z

If I'm not mistaken, we planed to clean the doc.pdf outside of the repo and use artefacts of the pipeline on the website. @jonasschwab is this still the case?

johanneshofmann87

@fassaad I would really appreciate more comments in the source code. I'm trying to read the code and it's not very easy.

Anika Goetz and others added 30 commits January 15, 2025 16:59

Switch for Langevin proposal and function for calculating force

a12ae38

Proposal with Langevin equation for local moves

0c3b65a

Merge branch 'master' into 305-proposal-of-updates-with-langevin-equa…

a2a4ba2

…tion

First version of Langevin proposal on one time slice

380815d

Warning message if fields are not type 3

12227d0

Global Metropolis-adjusted Langevin algorithm (MALA)

f2db2d8

Same format for global_tau and global_langevin_tau

cf13f54

Choice of the fields in the Global_MALA

1bed453

Single S0 instead of the whole thing

b72cbca

Variable renaming

6229b8e

Override functions Include N_Global_Langevin_tau

227489d

Bug in Wrapgr_Langevin_update; phase not updated

dce7e11

Global MALA: don't check precision before move is acceptec or rejected

49acb4d

Draft version of table with updating schemes

b5962a7

Control forces and max force for global MALA

86f8fdb

Control forces and max force for global tau MALA

9019de4

Control forces and max force for sequential MALA

f345af8

check forces before and after update

aebf65e

Merge branch 'master' into 305-proposal-of-updates-with-langevin-equa…

63693fc

…tion

Merge branch 'master' into 305-proposal-of-updates-with-langevin-equa…

87f98e6

…tion

Update doc.pdf

e28ab3d

Update

a103ab8

base implementation of Ham_Langevin_HMC_S0_single

7a8a58f

Handle non type 3 fields for Global tau MALA updates

69f6aef

Merge branch 'master' into 305-proposal-of-updates-with-langevin-equa…

08fe884

…tion

Update

1cdbc91

Merge branch 'master' into 305-proposal-of-updates-with-langevin-equa…

eb7d66e

…tion

Merge branch 'master' into 305-proposal-of-updates-with-langevin-equa…

a180e0b

…tion

Update

ca76770

Update

4362739

fassaad added 4 commits February 8, 2026 18:03

WWorking on doc

1cd8972

Working on doc for Updating

e8fd3e1

Update

e4a9dbb

Worked on the documentation.

6a6d3f7

ALF-Import-Bot added the gitlab merge request label Feb 12, 2026

Merge branch 'master' into 305-proposal-of-updates-with-langevin-equa…

d6f1e9d

…tion

fassaad added 2 commits March 25, 2026 11:37

Merge branch 'master' into 305-proposal-of-updates-with-langevin-equa…

2c05d25

…tion

Worked on the documentation, and added code snipets for Langevin ahd …

55a5ead

…HMC.

fassaad mentioned this pull request Mar 25, 2026

Resolve "implement Hamiltonian variables check" #566

Merged

fassaad requested a review from johanneshofmann87 March 25, 2026 15:59

johanneshofmann87 reviewed Mar 25, 2026

View reviewed changes

		deallocate(forces_0)

		end Subroutine Ham_Langevin_HMC_S0_single_base

		S0_ratio = S0(Flip_list(1),ntau,Flip_value(1))


		end Subroutine Global_move_tau

Conversation

ALF-Import-Bot commented Feb 12, 2026

Uh oh!

ALF-Import-Bot commented Feb 12, 2026

Uh oh!

ALF-Import-Bot commented Feb 12, 2026

Uh oh!

ALF-Import-Bot commented Feb 12, 2026

Uh oh!

ALF-Import-Bot commented Feb 12, 2026

Uh oh!

ALF-Import-Bot commented Feb 12, 2026

Uh oh!

ALF-Import-Bot commented Feb 12, 2026

Uh oh!

ALF-Import-Bot commented Feb 12, 2026

Uh oh!

ALF-Import-Bot commented Feb 12, 2026

Uh oh!

ALF-Import-Bot commented Feb 12, 2026

Uh oh!

joaogci commented Feb 26, 2026

Uh oh!

fassaad commented Mar 25, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

johanneshofmann87 left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants