Addition of new density-density operators mentioned in the discussion… #53

LeoGaspard · 2025-04-28T07:15:27Z

Addition of NupNdn, NupNup and NdnNdn operators. Addition of the equivalent hole density-density operators handled at the compile step following the discussion #50.

awietek#50

awietek · 2025-04-28T10:42:02Z

Dear Leo,

thank you very much for the pull request. Currently, according to the automated tests, the MPI code is not compiling: https://github.com/awietek/xdiag/actions/runs/14702408841/job/41256972462?pr=53

Please have a look at it and try to fix it. I will then take a closer look at the new code.

awietek · 2025-04-29T12:15:55Z

Thank you very much, the tests now work. I would have the following two suggestions before proceeding:

As we discussed, I would not like to add all the operators you now added to XDiag, but only "NupNup", "NupNdn", "NdnNup", and "NdnNdn". The reason is that I want to replace these correlators later on with an algebra feature and correlations like these operators might then be deprecated. So I do not want to have too many operators in there and I think all the measurements you want to perform can be done with the above four. So that's why I suggested to only implement "NupNup", "NupNdn", "NdnNup" and compile "NupNdn" to "NdnNup". So I would like you to remove the lines when you compile the operators like "NhtotNhtot". For you I think it does not make much difference, you could just have one line of code defining these correlators in your application, is this correct? Sorry if I was not sufficiently clear about that.
Second, I would like the operators to be tested in a symmetrized setting as well. This is done in line 48 and following of `tests/algebra/test_apply.cpp". Could you please add the corresponding tests (there only done for "NtotNtot") for the remaining diagonal operators?

Thanks for all your efforts

LeoGaspard · 2025-04-29T13:49:04Z

Dear Alexander,

Thank you for your input. Indeed I misunderstood what you told me, I now removed all the composed operators, leaving only NupNdn, NupNup and NdnNdn. I also added the tests in the file test/algebra/test_apply.cpp for the operators NupNdn (both i j and j i since it is important to check that they can be different), NupNup and NdnNdn.

awietek · 2025-04-29T15:26:43Z

Thanks, now there is a test failing unfortunately with the MPI library. Could you have a look into what the problem is there?

awietek · 2025-04-30T08:07:58Z

Thanks, so superficially all looks good. I will have to take a closer look now which can take one or two days, then I'll come back to you. I guess you can already use the version for project.

awietek

Thanks, this is almost ready to be merged. I have two suggestions:

The identity operator implementation should be changed, see annotations
The NdnNup operator should be added in addition to the NupNdn operator, by compilation.

awietek · 2025-05-01T08:14:32Z

xdiag/basis/apply_identity_distributed.hpp

This function should be moved to apply_identity.hpp. Also, this is specific to the electron bases. I would implement this more generically, something like this:

void apply_identity(Coupling const &cpl, const coeff_t *vec_in, coeff_t *vec_out, int64_t size) { using bit_t = typename basis_t::bit_t; coeff_t s = cpl.scalar().as<coeff_t>(); for (int64_t idx=0; idx<size; ++idx) { vec_out[idx] += s * vec_in[idx]; } }

which will work for the distributed case, you only have to give the size as an argument.

awietek · 2025-05-01T08:16:16Z

xdiag/basis/electron/apply/apply_number_number.hpp

awietek · 2025-05-01T08:17:15Z

xdiag/basis/electron_distributed/apply/apply_number_number.hpp

awietek · 2025-05-01T08:18:13Z

xdiag/basis/electron_distributed/apply/apply_terms.hpp

@@ -11,6 +11,7 @@
 #include <xdiag/basis/electron_distributed/apply/apply_raise_lower.hpp>
 #include <xdiag/basis/electron_distributed/apply/apply_szsz.hpp>
 #include <xdiag/basis/electron_distributed/apply/apply_u.hpp>
+#include <xdiag/basis/apply_identity_distributed.hpp>


please follow the suggestions iin apply_identity_distributed.hpp. Move to apply_identity.hpp

awietek · 2025-05-01T08:18:26Z

xdiag/basis/electron_distributed/apply/apply_terms.hpp

+        electron_distributed::apply_ndn_ndn<coeff_t>(
+            cpl, op, basis_in, vec_in.memptr(), vec_out.memptr());
+    } else if (type == "Id") {
+        apply_identity_distributed<coeff_t>( 


Change call as suggested

awietek · 2025-05-01T08:22:01Z

xdiag/operators/logic/types.hpp


 inline const std::vector<std::string> real_types = {
    "Id",     "SdotS",    "Exchange",   "SzSz",   "Sz",     "S+",
    "S-",     "Hop",      "Hopup",      "Hopdn",  "Cdagup", "Cup",
    "Cdagdn", "Cdn",      "HubbardU",   "Ntot",   "Nup",    "Ndn",
-    "Nupdn",  "NtotNtot", "NupdnNupdn", "tJSzSz", "tJSdotS"};
+    "Nupdn",  "NtotNtot", "NupdnNupdn", "tJSzSz", "tJSdotS",
+    "NupNdn", "NupNup", "NdnNdn"};


Add NdnNup here

awietek · 2025-05-01T08:22:15Z

xdiag/operators/logic/types.hpp

+    {"NupNdn", 2},
+    {"NupNup", 2},
+    {"NdnNdn", 2},
+};


Add NdnNup here

awietek · 2025-05-01T08:22:31Z

xdiag/operators/logic/valid.cpp

@@ -25,7 +25,8 @@ void check_valid(Op const &op) try {
    } else if ((type == "Hop") || (type == "Hopup") || (type == "Hopdn") ||
               (type == "SzSz") || (type == "SdotS") || (type == "Exchange") ||
               (type == "NtotNtot") || (type == "NupdnNupdn") ||
-               (type == "tJSzSz") || (type == "tJSdotS")) {
+               (type == "tJSzSz") || (type == "tJSdotS") || (type == "NupNdn") ||
+               (type == "NupNup") || (type == "NdnNdn")) {


Add NdnNup here

awietek · 2025-05-01T08:23:17Z

xdiag/operators/logic/order.cpp

We should take care of the case that an NdnNup operator gets properly ordered to NupNdn to compare it.

awietek · 2025-05-01T08:24:33Z

docs/documentation/operators/operator_types.md

@@ -19,6 +19,7 @@ Generic operators in XDiag are represented as [OpSum](opsum.md) objects made up
 | `Nup`             | A number operator for an $\uparrow$ spin $n_{i\uparrow}$                                                                                                                   | 1            | tJ, Electron, tJDistributed, ElectronDistributed                                |
 | `Ndn`             | A number operator for an $\downarrow$ spin $n_{i\downarrow}$                                                                                                               | 1            | tJ, Electron, tJDistributed, ElectronDistributed                                |
 | `Ntot`            | A number operator $n_i = n_{i\uparrow} + n_{i\downarrow}$                                                                                                                  | 1            | tJ, Electron, tJDistributed, ElectronDistributed                                |
+| `Nhtot`           | A number operator for holes $n^H_i = 2 - n_{i\uparrow} - n_{i\downarrow}$                                                                                                  | 1            | tJ, Electron, tJDistributed, ElectronDistributed                                |


Remove Nhtot here

Addition of new density-density operators mentioned in the discussion :

b4fa0a1

awietek#50

Electron_distributed fix for NupNdn, NupNup and NdnNdn

9133bf3

Removed some composed operators

15d920f

Fixed wrong NupNup and NdnNdn in electron_distributed block

30198af

awietek requested changes May 1, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Addition of new density-density operators mentioned in the discussion… #53

Addition of new density-density operators mentioned in the discussion… #53

LeoGaspard commented Apr 28, 2025

awietek commented Apr 28, 2025

awietek commented Apr 29, 2025 •

edited

Loading

LeoGaspard commented Apr 29, 2025

awietek commented Apr 29, 2025

awietek commented Apr 30, 2025

awietek left a comment

awietek May 1, 2025

awietek May 1, 2025

awietek May 1, 2025

awietek May 1, 2025

awietek May 1, 2025

awietek May 1, 2025

awietek May 1, 2025

awietek May 1, 2025

awietek May 1, 2025

awietek May 1, 2025

Addition of new density-density operators mentioned in the discussion… #53

Are you sure you want to change the base?

Addition of new density-density operators mentioned in the discussion… #53

Conversation

LeoGaspard commented Apr 28, 2025

awietek commented Apr 28, 2025

awietek commented Apr 29, 2025 • edited Loading

LeoGaspard commented Apr 29, 2025

awietek commented Apr 29, 2025

awietek commented Apr 30, 2025

awietek left a 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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

awietek commented Apr 29, 2025 •

edited

Loading