Five verified procedures

PNNL, RefIMP, a comparison across formats, and restoring a saved view.

1. View PNNL ethanol directly from the Library

  1. Install the NIST/PNNL managed library. Open IR Spectroscopy from Viewers, even without an active session.
  2. In IR References → Records, search for ethanol-k-pnnl-2017. Select the record and check wavenumber, cm⁻¹, absorption_index, unit 1, and 15,347 samples.
  3. Use Compare → Add selected IR reference. In Method, choose overlay and none and prepare the plot.
The PNNL reference is displayed in Plot without a session Dataset.
The PNNL reference is displayed in Plot without a session Dataset.

2. Compare two grids for the same optical constant

  1. Generate and reopen the session with nist_pnnl_ir_faculty. Add ir.nist.ethanol.standard_grid and ir.nist.ethanol.legacy_grid to Sources in that order, using wavenumber and signal.
  2. Prepare overlay. The two curves overlap almost entirely. Choose difference to inspect the residual from realignment and ratio to check the common domain.
  3. In ratio mode, try setting Denominator floor to 0.01 and leave Display normalization at none. Check the missing sample count in Source and result.
The ratio with an explicit threshold only shows regions where the denominator is accepted.
The ratio with an explicit threshold only shows regions where the denominator is accepted.

3. Verify the same ItemID in HDF5 and CSV

  1. Generate and reopen the Session with refimp_full_training_ir. Add ir.refimp.item13 and ir.refimp.csv_compat using wavenumber and signal.
  2. Prepare overlay and difference with none. The 637 pairs agree. The difference confirms that both formats yield equivalent readings of the same ItemID.

RefIMP data by Jana Weisser and Teresa Pohl, 2022, CC BY-SA 4.0. The comparison uses two conversions of the same ItemID. mediaTUM source.

Difference between HDF5 and CSV for the same RefIMP ItemID, showing agreement. Weisser and Pohl, CC BY-SA 4.0.
Difference between HDF5 and CSV for the same RefIMP ItemID, showing agreement. Weisser and Pohl, CC BY-SA 4.0.

4. Run retrieval against the full RefIMP corpus

  1. In Analysis → Input, select ir.refimp.item13 and discover its variables. Set wavenumber, signal, relative_transmission_signal, and the tum.refimp.ftir.training.2022 Library.
  2. In Processing, use none for baseline and smoothing, derivative order 0, mean_center_unit_vector, and a step of 4 cm⁻¹. In Matching, choose cosine and set the band to 1300–1800 cm⁻¹ with a weight of 0.35. Use Run Analysis.
  3. When the results are published, use Read ranked references. The first row is refimp-train-00013 with a score of 1.0. This retrieves ItemID 13 from the corpus that contains it.
  4. Keep one measured trace in Sources. In Analysis → Results, choose rank 1 and use Add selected ranked reference. Prepare top_bottom to view both traces with independent Y axes.
The ranking shows the RefIMP candidate in first place. Weisser and Pohl, CC BY-SA 4.0.
The ranking shows the RefIMP candidate in first place. Weisser and Pohl, CC BY-SA 4.0.
Session ItemID 13 and the reference selected from the ranking. Weisser and Pohl, CC BY-SA 4.0.
Session ItemID 13 and the reference selected from the ranking. Weisser and Pohl, CC BY-SA 4.0.

5. Reuse a formatted plot

  1. With both NIST grids in top_bottom, use Plot → Properties… to set a shared X range of 650–3500 cm⁻¹ and a title for the first panel.
  2. In Series style…, set the first curve to color #087C8F, line style dashed, and width 2.5. Apply the changes and save the NIST k due pannelli variant with Save Plot Variant….
  3. Use Export PNG… to save the result as an image. If you change the view later, use Load Plot Variant… to restore the saved presentation and source selection.
PNG export dialog over the NIST/PNNL view with titles, limits, and styles applied.
PNG export dialog over the NIST/PNNL view with titles, limits, and styles applied.
The NIST variant is restored with its layout and source identities verified.
The NIST variant is restored with its layout and source identities verified.