feat[FieldData]: export to ZBF #2397

bzhangflex · 2025-04-22T22:13:04Z

Adds a to_zbf() method to FieldData for exporting E fields to a ZBF file.

yaugenst-flex

Nice, looks great! Only some minor comments.

yaugenst-flex · 2025-04-23T08:27:29Z

tests/test_data/test_monitor_data.py

Nice! Two comments on the tests:

It looks like you already implemented all the logic for reading a zbf file here. Should we also add a FieldDataset.from_zbf()? That'd make it really easy to create custom field sources from zbf data.

Instead of re-implementing dummy field data generation here it's probably cleaner to set up a simulation and use the emulated run to get the field data? That'd also be a bit more representative of the actual workflow I guess.

For 2. are you referring to this for the emulated run?

tidy3d/tests/utils.py

Line 920 in b848f5d

def run_emulated(simulation: td.Simulation, path=None, **kwargs) -> td.SimulationData:

yea exactly

yaugenst-flex · 2025-04-23T08:36:51Z

tidy3d/components/data/monitor_data.py

+        fname: str,
+        background_refractive_index: float,
+        freq: Optional[float] = None,
+        mode_index: Optional[int] = 0,


Looks like this is not really Optional but would always expect an int?

lucas-flexcompute · 2025-04-23T10:00:21Z

tests/test_data/test_monitor_data.py

+
+    def read_beam_file(self, filename: str) -> tuple:
+        """Read in a Zemax Beam file
+        Adapted from https://community.zemax.com/code-exchange-10/python-class-to-open-zbf-file-4845


Should we worry about the (lack of) license in the original code? Is it assumed to be public domain because it was shared in a forum? I'm not sure if that's how it works.

Good question, I'm not sure either 😅

lucas-flexcompute · 2025-04-23T10:10:39Z

tests/test_data/test_monitor_data.py

+        # beamfilename = self.open_zbf()
+        f = open(filename, "rb")
+        # zemax version number
+        version = unpack("i", f.read(4))[0]


This use of unpack is not setting alignment and size, so it will only work if the native settings match the ZBF specs. In my original implementation I fixed the sizes and left native order (=) because the manual did not specify it, but it should probably be set little-endian (<).

Also, it could probably be a lot more efficient if it unpacked more values at once:

version, nx, ny, ispol, units = unpack("<5i", f.read(4 * 5))

tylerflex

Thanks @bzhangflex , this looks great! Will be super useful and a highly requested feature!

I had a few minor comments. I'm not familiar with zemax configuration so take what I say with a bit of a grain of salt.

tylerflex · 2025-04-23T12:55:28Z

tidy3d/components/data/monitor_data.py

+        mode_index : int
+            Mode index selection.
+        num_samples: int
+            Number of samples to use. Must be a power of 2


could you explain a bit more what is being sampled? and what does the default value correspond to?

tylerflex · 2025-04-23T12:58:39Z

tidy3d/components/data/monitor_data.py

+        e_x = e_x.sel(f=freq, drop=True)
+        e_y = e_y.sel(f=freq, drop=True)
+
+        if "mode_index" in e_x.coords:


what case does this cover? I guess if the object is a ModeSolverData?

If the mode_index is only used in this case, maybe it does actually make sense for the default value to be None? And then only do the isel here if it's not None?

@yaugenst-flex

Good question, I assumed it was for ModeSolverData but I haven't tested it yet. I'll test it out and also change the default value

if so maybe we can mention this in the changelog too

tylerflex · 2025-04-23T13:00:40Z

tidy3d/components/data/monitor_data.py

+
+        # Use the central frequency if freq is not specified
+        if freq is None:
+            freq = e_x.coords["f"].values[e_x.coords["f"].size // 2]


do we need to worry if the number of frequency coords is even? wonder if we should rather interpolate to the mean frequency?

tylerflex · 2025-04-23T13:02:11Z

tidy3d/components/data/monitor_data.py

+        # Header info
+        version = 1
+        polarized = 1
+        units = 0  # "mm": 0, "cm": 1, "in": 2, "m": 3


should this be something that the user can configure through the kwargs?

tylerflex · 2025-04-23T13:05:08Z

tidy3d/components/data/monitor_data.py

+        version = 1
+        polarized = 1
+        units = 0  # "mm": 0, "cm": 1, "in": 2, "m": 3
+        lda = C_0 / freq * 1e-3


this 1e-3 floating around here and below makes me a bit nervous. I think it would be better if we had something like

# kwarg length_unit = "mm" length_scale_map = dict(mm=1e-3, cm=1e-2, in=...) # maps kwarg to length scale factor if length_unit not in length_scale_map: raise KeyError(...) length_scale = length_scale_map[length_unit]

and then use length_scale in place of 1e-3.

If there are free units floating around like this it can lead to some unexpected bugs.

tylerflex · 2025-04-23T13:05:58Z

tidy3d/components/data/monitor_data.py

+        r_y = 0
+
+        # Pilot beam z position w.r.t. the waist
+        z_x = 0 * 1e-3


what's with the 0 * 1e-3? are we planning to allow this to be modified eventually?

good catch! I'm not fully sure what the pilot beam is used for (seems like it helps guide the actual beam) but it sounds like a good idea to let the user define values for it.

tylerflex · 2025-04-23T13:07:03Z

tidy3d/components/data/monitor_data.py

+        # Number of samples: n_x = 2 ** a, a >= 5 and a <= 13 (32-bit), or a <= 14 (64-bit)
+        if num_samples is not None:
+            # custom user defined number of samples
+            # Raise an error if the number of samples is not a power of 2


can you explain why the num_samples needs to be a factor of 2? in principle we could interpolate using however points they provide?

it's a requirement set by zemax. My guess is it might have to do with FFTs in their beam propagation algorithm.

Should we not enforce it here and let the user choose an arbitrary number? That could be fine too if Zemax emits a clear error message, or changes behavior in the future

it's probably fine to enforce it here. maybe we can just mention in the error message that this is a requirement for zemax

tylerflex · 2025-04-23T13:07:59Z

tidy3d/components/data/monitor_data.py

+            n_x = num_samples
+            n_y = num_samples
+        else:
+            n_x = 2 ** min(13, max(5, int(np.log2(x.size) + 1)))


should these 5 and 13 be hardcoded somewhere? eg as a module level constant? or configurable? not sure exactly where they come from. is it a requirement of zemax?

hmm it was left in from @lucas-flexcompute's code, I wonder if there was a specific reason for the values. If not maybe we don't specify any default n_x or n_y?

The Zemax manual I have (dated July 8, 2011) defines that the number of samples must be a power of 2 between 32 and 8192 (inclusive)

tylerflex · 2025-04-23T13:09:53Z

CHANGELOG.md

it might be worth also adding this to the FAQ section of the docs, as we recently had some users unaware of our export to MATLAB. so probably some extra information about this capability in the docs wouldn't hurt. Also, not sure if it applies, but it might be nice also to add it to at least one notebook example.

tomflexcompute · 2025-04-23T15:24:25Z

Since this is a bit experimental and largely untested, there might be cases that it will not work properly. Should we add a warning when it's called? Something like "This is an experimental feature. If you run into any issues, contact us at flexcompute.com/tidy3d/technical-support/"

feat[FieldData]: export to ZBF

5beeb76

bzhangflex requested review from tomflexcompute, lucas-flexcompute, yaugenst-flex and tylerflex April 22, 2025 22:13

yaugenst-flex reviewed Apr 23, 2025

View reviewed changes

lucas-flexcompute approved these changes Apr 23, 2025

View reviewed changes

tylerflex requested changes Apr 23, 2025

View reviewed changes

yaugenst-flex assigned bzhangflex Apr 23, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat[FieldData]: export to ZBF #2397

feat[FieldData]: export to ZBF #2397

bzhangflex commented Apr 22, 2025

yaugenst-flex left a comment

yaugenst-flex Apr 23, 2025

bzhangflex Apr 23, 2025

yaugenst-flex Apr 24, 2025

yaugenst-flex Apr 23, 2025

lucas-flexcompute Apr 23, 2025

bzhangflex Apr 23, 2025

lucas-flexcompute Apr 23, 2025

tylerflex left a comment

tylerflex Apr 23, 2025

tylerflex Apr 23, 2025

bzhangflex Apr 23, 2025

tylerflex Apr 23, 2025

tylerflex Apr 23, 2025

tylerflex Apr 23, 2025

tylerflex Apr 23, 2025

tylerflex Apr 23, 2025

bzhangflex Apr 23, 2025

tylerflex Apr 23, 2025

bzhangflex Apr 23, 2025 •

edited

Loading

tylerflex Apr 23, 2025

tylerflex Apr 23, 2025

bzhangflex Apr 23, 2025

lucas-flexcompute Apr 23, 2025

tylerflex Apr 23, 2025

tomflexcompute commented Apr 23, 2025

feat[FieldData]: export to ZBF #2397

Are you sure you want to change the base?

feat[FieldData]: export to ZBF #2397

Conversation

bzhangflex commented Apr 22, 2025

yaugenst-flex 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

tylerflex 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

bzhangflex Apr 23, 2025 • 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

tomflexcompute commented Apr 23, 2025

bzhangflex Apr 23, 2025 •

edited

Loading