DiffAtOnce · Getting started · Windows x64

Continuum Preview

Work with signed attenuations, review the D range and inspect DOSY/MwOSY maps in separate revisions. Start with the synthetic example: no real experiment is needed to learn the workflow.

Continuum 0.3.12 Preview 1 for Windows x64. Both interfaces are distributed through the Preview channel.

Preview downloads · 0.3.12-preview.1

Choose one interface; you do not need to install both. These are per-user Windows x64 MSI packages. The packages are not Authenticode-signed. Check the SHA-256 before installation. Both are independent of the stable catalog.

1. Choose an interface

  1. For the desktop interface, install the per-user Continuum Preview MSI on Windows x64 and open Continuum Preview from Start. The installer includes its own .NET Desktop runtime and molecular engine.
  2. For the local web interface, install its MSI and open “Continuum Web Local Preview” from Start. The installer includes ASP.NET Core and the molecular engine. Open http://127.0.0.1:5274/ in a current browser on the same computer. If the port is in use, check which application is using it.
  3. Use local web on a personal workstation, with one working session and non-confidential data. There are no accounts or isolation between clients on the same computer. Do not forward the port, proxy the service or share it on a network. To finish, save first, return to the server console and press Ctrl+C. Closing the tab does not stop the process.

Predict D from a structure

  1. Open “Molecules · diffusion and libraries” on desktop or “Molecules and libraries” in Web Local. Select one of the 253 records or enter its SMILES, choose a solvent and enter the temperature in kelvin.
  2. With the supplied catalogue, naphthalene in THF-d₈ at 298.15 K and 0.507 mPa·s gives D = 1.81061 × 10⁻⁹ m²/s. The result identifies the catalogue by its SHA-256. All 253 records and 15 solvents work offline without Python.
  3. For a new SMILES, paste it in the structure field and run the prediction. The installer includes the RDKit geometry engine and its dependencies. The calculation runs on this computer, even offline. Select a molecule or enter a SMILES to see its structure beside the prediction. Archived entries also provide their available experimental ¹H traces, including when the database path exceeds 260 characters.
  4. The viewer opens in 3D with the calculated conformer. Select 2D for the structure drawing and 3D to return to the conformer. Drag to rotate, use the zoom buttons and select Reset to return to the initial orientation. Rotation and zoom change only the view.

2. Try a revision with the example

  1. Choose English and “Load synthetic example”. Review the signals, 23 b values, units and provenance supplied with the example. Synthetic input tests the workflow; it does not qualify a sample.
  2. Assisted and Expert use the same core. Expert lets you edit the method, grid and D limits. Distinguish the editor draft, the plan applied to the next calculation and the visible revision: editing the draft does not change a saved result.
  3. Apply the plan, then calculate a new revision. Choose a signal and examine its attenuation, fit, residual and distribution. Save a copy under a new name. Select the revision and language to export a frozen HTML report; the report should not require another inversion.

3. Move to a Bruker copy or a batch

  1. Use a copy of the 1r/2rr data. In WPF open “Bruker 1r/2rr · batch”; on the web use “Bruker · batch” and a ZIP of the copy. Check the pdata processing, units, pulse sequence/gradients, diffusion list where applicable, and actual acquisition count. Never invent missing or ambiguous b encoding.
  2. When using TTMS, a proposed peak is not a confirmed reference. Review identity and the local fit; transferring a factor requires an accepted compatible source and explicit consent. Each ramp keeps its own temperature. Do not copy a factor to hide a failed fit.
  3. The .continuum-batch.json batch and .continuum-maps.json map archive are different documents. To run a new extraction, attach the matching Bruker copy again. Opening a saved result alone does not supply its original spectra.

4. Read the map without treating a warning as a result

  1. Open DOSY / MwOSY 2D maps. Automatic D support uses the ramp’s b values and regressions across several signal windows; neither TTMS nor a desired mass determines the range. You can review and apply manual limits for the next calculation.
  2. Keep the previous map. Examine residual, convergence and edge signal separately. A residual above 3% remains flagged. If a distribution reaches an edge, the tail beyond its support was not measured. Do not force lobes or expand the range merely to pass.
  3. MwOSY shows apparent mass under the declared calibration. It does not establish absolute masses, Mn/Mw/PDI or chemical identity. NNLS and Tikhonov are the visible baseline methods. Command-line research selectors are not desktop or web controls. Changing language or viewport does not change observations.

Export a MwOSY projection without repeating the inversion

Use this workflow in Continuum WPF or Web Local. Web Local runs on your computer; it is not the laboratory hosted at diffatonce.com. You need a saved map archive and a reviewed mass calibration for that map.

  1. In 2D maps, select “Open map archive” and choose the .continuum-maps.json copy. A .continuum-batch.json holds the batch; it is not a substitute for the map archive. Do not select Calculate.
  2. Choose the language, revision, ramp or average, and the “MwOSY · apparent M / Da” axis. Note them so that you compare the same figure in both languages. Zoom changes the view, not the cells used for the projection.
  3. Under “MwOSY report content”, choose “Global only” to summarise the observed signal, or “Global + signal components” to describe the fit on the archived cells. The second option does fit the 1D projection; neither repeats the 2D map inversion.
  4. In WPF choose “Visible-view report…”; in Web Local choose “Report of visible figure”. Use a new filename. Check the revision, ramp or average, and calibration in the HTML. Changing language does not require a new calculation.

Why a component can have no Mn, Mw or dispersity

The fitted centre, signal-weighted mean and signal SD are not Mw, Mn or measurement uncertainty. A Gaussian without an unambiguous maximum of its own does not establish a separate species.

The mass-response and number-response scenarios are different assumptions. Mn*, Mw* and Đ* are conditional apparent moments only. A term’s percentage area is relative to the fitted sum, not a measured mass fraction.

A dash does not mean zero. A cut or unresolved term, or one whose endpoint reaches 1% of its maximum, is not extrapolated. The sum may be blocked if any term fails or the observed projection does not close, even when the fitted curve looks closed.