User guide

Here you can find the user guides for different loaders (in future for other aspects of SED as well).

Flash Loader

If you face a problem loading your run/file, the first thing to troubleshoot is to check the raw H5 file.

A few common issues:

The file doesn't contain the defined dataset (channel)
- Fix: Remove that channel
The file contains the dataset but it is empty (We plan to introduce a fix for this)
- Fix: Remove that channel
Numba can not bin
- Reason: Numba can not infer the datatype
- Fix: Change histrogram mode to numpy
```
binning:
  hist_mode: numpy
```

To setup metadata collection, please define the metadata section in the config file

metadata:
  scicat_url: <URL>
  scicat_token: <TOKEN>

After metadata is collected and you want to export to Nexus with all relevant data, put this section in the config file

nexus:
  reader: "flash"
  definition: "NXmpes"
  input_files: ["NXmpes_config_HEXTOF_light.json"]

When reading files, using SedProcessor(), there are multiple options that allow different behavior

force_recreate: Recreates the buffer files that were already generated
save_parquet: Saves the whole run or multiple runs into a single parquet file
load_parquet: Reads the whole run or multiple runs that were saved
- If you saved while defining runs=[1234, 5678], then you have to define the same runs to load the parquet file
detector: Allows you to add a suffix to your buffer files
- Example use case: detector = "4Q" and detector = "1Q" to distinguish the buffer files of two different detectors
converted: Set to True if you augment the data by adding additional columns externally and save it into converted folder (subfolder under the parquet directory).
debug: Set to True if you want to troubleshoot a problem by loading data serially.