Add extended mag field map

[tvami]

I am updating ldmx-sw, here are the details.

What are the issues that this addresses?

Resolves #1562
(technically, but I'm not sure I like the solution)

I extended the grid to

    x_range = np.arange(-500.0, 500.0, 5.0)
    y_range = np.arange(-500.0, 500.0, 5.0)
    z_range = np.arange(-1500.0, 2500.0, 5.0)

and added zeros everywhere where the is no value already.
I also tried to just add zeros to the edges of the cude: this didnt work, since the bin size is stepping and it's not available.

Check List

• I successfully compiled ldmx-sw with my developments
• I ran my developments and the following shows that they are successful.

The ACTS msg about propagation failure is gone.

[tvami]

This lead to a lot of differences all over
https://github.com/LDMX-Software/ldmx-sw/actions/runs/13469851573?pr=1632

[tvami]

I'm pretty confused about a few things I see:

I'm looking at signal now, since that should be more affected.
[plots are normalized to unit area]

The incident electron pT is wildly different:

But even the recoil goes higher:

The recoil ele at the ECAL face is now deflected in pos X

But the y distribution looks more smooth

The effect on the BDT is good too, we gain about 20% signal (because the electron becomes fiducial in more cases)

Things in HCAL changed a lot, but it all goes down to seeing less stuff in the side HCAL

Which then improves the HCAL veto performance (a tiny bit)

None of the above depends on the ACTS tracking, it's all truth.

Now ACTS tracking

Recoil d0:

Pulls at the ECAL improve:

I guess we lost more hits in the end of the recoil

The number of tracks is not going into the direction I wanted in CKF... but I guess given how off the d0 etc is this is not surprising.

Now the same in GAS looks improved

The truth phi is shifted, so it's not just ACTS settings

[tvami]

I should also say that in ECAL PN gold the text "PROP ERROR Step failed with MagneticFieldError:1: Interpolation out of bounds was requested" appears 129 times, in the logs after this PR all that is gone... so at least the direct problem is gone haha, but may have introduced other problems. I'm not sure what to take from the plots above.

[EBerzin]

@tvami I think the issue might be the order the Bfield values are listed in file. From looking at how the field is read in MagneticFieldMap3D.cxx, it looks like it assumes that the points will be provided in either strict increasing or decreasing order. The interpolation in GetFieldValue() is then based on the grid index. In the extended field map, this is violated here, for example, where the x and y positions jump from 495 to -250 and -70. I think for things to make sense, the existing field values need to be filled in to preserve position ordering, rather than at the end of the file, where they are now.

495.0 495.0 2485.0 1e-11 1e-11 1e-11
495.0 495.0 2490.0 1e-11 1e-11 1e-11
495.0 495.0 2495.0 1e-11 1e-11 1e-11
-250.0 -70.0 -1495.0 2.509e-07 -4.39e-06 1.031e-06
-250.0 -70.0 -1490.0 2.571e-07 -4.457e-06 1.076e-06
-250.0 -70.0 -1485.0 2.633e-07 -4.525e-06 1.121e-06

[tomeichlersmith]

(I would love for you to commit whatever code you are using to evolve the map into MagFieldMap/utils or something as well. I have a feeling we will need to do something like this again...)