Answers to keep your research moving.

Prime is a general diffusion-NMR environment with configurable methods, calibration and R&D modules. ResinAtOnce focuses on resins, their components and reporting workflow. They share parts of the scientific core, but a resin calibration does not become universal in Prime.

This update documents Prime 0.14.1 and preserves existing distribution. It does not add a public installer or change access terms. See Software & code for availability; if you have a package, use the guide matching your executable.

The application uses WPF on 64-bit Windows. The self-contained package includes the .NET runtime. Local C# methods need no MATLAB, Python or HTTP connection. External engines and AI models have their own requirements; no unmeasured memory minimum or GPU speedup is claimed.

Yes: they are Prime 0.12.5 WPF windows captured with experimental TP194 data, ¹H/CDCl₃, experiments 10–32. Settings and library views are labelled. MwOSY shows an exploratory PS conversion, not a validated identity or absolute mass. The ResinAtOnce 0.48.1 guide uses simulated DEMO; its product page also includes three separately labelled views of an archived experimental result. Research figures are derived outputs, not Prime screenshots.

No. Conventional DOSY uses processed Bruker 1r data and compatible ramps. The ultrafast D–T₂ module has a specific raw-data route with an echo contract and spatial transform; this does not mean support for every sequence. SPEN needs its own reconstruction and is not automatically run when opening the example package.

No. Import, inspection and calculation do not rewrite original spectra. Alignment or b recalibration creates a derived copy and archives the previous state. Keep originals and projects; fingerprints allow input changes to be detected.

No. You can calculate D without an internal standard if b encoding is reviewed. Results and reports state that no standard was used. Rh and Mw may require viscosity and other conditions even when D is calculable.

Several substances can occur in that region. Confirm the added standard’s identity, region and attenuation. Rh = 3.65 Å belongs to the supplied laboratory protocol; it is not an identification by chemical shift.

Each standard has local D, expected D and a factor. Prime checks overlap and disagreement before combining compatible factors using a geometric mean. Disagreements are not hidden, and each signal is not separately forced to its expected value.

In Prepare DOSY / MwOSY review Medium / temperature or η(Tref) beside the standard. Accept fixed η at 293 K if appropriate, choose variable η or supply a manual measurement. Confirm identity and conditions. Without documented Rh, provide a sourced manual expected D; do not invent a radius.

No. Detection reads metadata; it does not measure Gmax or mixture viscosity. Check sequence, b, temperature and applicability. The shared experimental TBO calibration is not replaced with nominal gradient strength.

It must not. fReference, fTemperature and their product are stored and applied once to the D axis. Reviewing η(Tref) is distinct from enabling thermal normalisation. Check saved-result factors, not just current checkboxes.

No column passes the current attempt’s screening. Review the diagnostic: unestimable noise, too few useful points, negatives, growth, insufficient b coverage or exclusions. The previous map is not erased. Do not relax requirements merely to obtain an image.

The method has not satisfied its stopping criteria in that calculation. The iterate may be useful diagnostically but is not an approved solution. Review parameters, coverage and residuals; more iterations do not guarantee identifiability.

Not necessarily: their objectives and regularisation differ. Compare identical signed data, grid and calibration, plus prediction and residuals. This version uses a modified in-house dual-Newton MaxEnt implementation; a narrow or broad curve alone does not establish composition.

A band comes from a projection over many ppm positions. A peak calculation integrates a specific region. To compare algorithms use Local signal · TRAIn / compare with identical inputs and parameters, distinguishing measured from corrected D.

D only calculates a monoexponential fit and draws a marker. Use Calculate D / Mw and distribution for an actual distribution. A single D does not determine width or dispersity.

Restrict calculation to a relevant region or increase points per bin after reviewing lost resolution. Choose suitable method and grid, and compare a small case before running a batch. Save parameters. Do not interpret an overfitted smaller residual as a guaranteed improvement.

The logarithm is not calculated for invalid points or I≤0. They remain in the table with a reason and are not replaced by zero. Linear-tab masks do not clip the signed data used by inversions.

No. Selecting points by R² can bias the result and hide an inadequate model. Prime limits exclusions, retains comparison with all valid points and records the mask. Review physical reasons, not just the R² number.

No. PS and PPG are two high-dilution calibrations. There is also one empirical concentration model developed with PS. Applying it to other families requires justification; no universal validation is claimed.

First solute; second solute + solvent. Solvent mass is the difference. Quantities with units are read from the title of the first experiment in the ramp. Missing or ambiguous quantities remain empty and are requested when using concentration.

Not in general. Since Rh is proportional to 1/D, mean radius differs from the radius of mean D. Prime distinguishes monoexponential effective Rh from distribution mean Rh, with the viscosity and temperature used.

Not automatically. Moments are signal-weighted; proton response, relaxation, suppression, overlap and regularisation can affect areas. Interpreting an index as dispersity requires the appropriate model and weighting. Quantitative NMF unmixing is a research topic, not an already validated feature.

It is a shape clue, not an identification. Overlap, exchange, heterogeneity and processing can also broaden signals. Review the spectrum, other resonances, attenuation and chemical evidence before assigning.

The model retains positive unfitted signal and declares visually added area. Projection follows the active representation. Full range and Fit signals only reframe; they must not disable modelling or erase results. Review residual, threshold and original signal.

No: it proposes documentary hypotheses for the current selection, with sources and checks. A phthalic candidate can be compatible without establishing a free or bound species or concentration. No catalogue match does not rule out a substance either. In ResinAtOnce local inference requires a matching model, corpus and manifest: checksum mismatch blocks it without removing rules and documentation. In Prime, GGUF CPU requires an absolute path, verified SHA-256 and a model with a compatible chat template. No newly validated fine-tuning is claimed.

Current parameters or acquisition are not those that produced the saved result. Recover map parameters to retain it, or recalculate with the edits. Beatrix must not mix historical values with the current selection.

No. Alignment compensates reviewable spectral shifts; it does not remove material transport or alter kernel physics. Convection requires its own diagnosis and evidence.

If conditions, ramps or evidence are missing, a binary answer would be invented. Inconclusive preserves that uncertainty. Only a compatible, accepted review can provide a corrected final D for the signal and be documented in the report.

Not as general validated workflows. Prime includes an exploratory C# reconstruction for ultrafast D–T₂ example 1502, with TRAIn2D and FISTA. SPEN-DOSY remains a published inspection package; any external reproduction is labelled accordingly. The user’s new ultrafast DOSY sequence remains untested, and UF-DEXSY does not validate it.

No. The supplied train2d.py program was ported to C# with retained provenance. TRAIn, 2D inversion schemes and FISTA have published antecedents, distinguished from the local file. References explain the method; they do not attribute that file to a specific paper or certify this implementation.

Its sequence combines spatial diffusion encoding with CPMG readout. The map represents D and effective T₂. It is not a chemical ppm–D map or a test of the new DOSY sequence. A measured coil-reference profile is missing; spatial window and b require review before quantitative scale interpretation.

Reproducible cases were fixed for local state after errors/cancellation, invalid component-separation inputs, molecular inversion near the κ(M) segment change and interrupted JSON saving. Prime 0.12.5 and ResinAtOnce 0.48.1 share these fixes and regression tests. This does not mean every algorithm or sample is experimentally validated.

The empirical equation is piecewise: κ(M) changes at 117,000 Da. It can leave a D interval with no mass represented by that formula. The current version retains valid endpoints and does not invent mass by interpolating across the gap. Check units, C, viscosity and domain; use an alternative calibration only if justified.

Check the final status; do not use an earlier graph as output from the cancelled attempt. Atomic review-JSON saving retains the previous file until the new output completes and leaves a .bak copy of the replaced file. A multi-file export requires checking its manifest and completion marker; do not confuse a partial batch with a complete report.

Record executable version, ramp, ppm region, method, parameters, full message and use of standards/temperature. Keep an example project and manifest; remove sensitive data before sharing. Use this website’s Support and suggestions and specify Prime or ResinAtOnce. A local-method error is not the same as an HTTP connection failure.

In Prime and ResinAtOnce, pending approval means this copy’s registration was received but is not yet authorised. The administrator reviews it in Copies and installations. Retain the ID and select Check approval after review; Continue is enabled only with active authorisation. You do not need a web account, do not repeat registration, and no deadline or automatic email is guaranteed. Before approval you can use public Support.

No. Registration requests authorisation for a desktop copy; a support enquiry asks for help and does not change that permission. Integrated support needs a previously approved copy, even if access is subsequently blocked. If you are pending or cannot open the application, use Support on this website and include the product, version, status and copy ID. The visible ID is not a password: never send identity.dpapi, secrets or credentials.

No. Reopening restores the saved state and checks approval when it can connect; for an already-pending copy it does not create another registration or approve it automatically. Offline access needs a signed active authorisation that has not expired; pending does not enable offline access. Moving folders or changing Windows users may create another copy. Retain your identity and projects, check approval and request help without deleting or cloning the identity.

It is an integrated assistant: it uses the active signal or band, engine facts and available sources to guide preparation, calibration, assignment, fit and next-step checks. Prime distinguishes local rules from generated text; in ResinAtOnce the CPU LLM selects supplied documentation and candidates remain tied to the spectrum. It is neither an alternative numerical engine nor automatic identification.

In Prime use rules without an LLM or configure a GGUF CPU model with path and SHA-256. Small, medium and large adjust the response budget and support models you have configured; they do not automatically download or select a model. In ResinAtOnce, Light and Advanced limit manual queries to up to 4 and 8 available CPU threads respectively, with the installed local package. CPU, free RAM and model size affect speed; more resources do not improve chemical or calibration validation.

No. Integrated GGUF execution uses CPU. In Prime a GPU can be used only through a compatible server you configure; you can also choose an API provider with access to the specified model. Configure provider, URL, model and credentials where required. A remote endpoint requires HTTPS and explicit context-transfer permission; localhost and offline execution are different routes. ResinAtOnce’s Beatrix assistant is offline CPU and does not provide Prime’s GPU/API integration.

No. It can explain a warning and suggest a concrete check. In Prime possible misalignment leads to Review / align, where you compare and confirm a derived copy; it is not applied by a model decision and does not correct convection. For residuals or non-convergence, review the same attenuation, coverage and settings. Accepted reviews require your justification, remain separate and retain provenance. The five help examples are illustrative, not results or experimental validation.

They are different decompositions. Atoms reconstruct the spatial field; several ppm positions can share a lateral lobe. lateral-modes-v1 detects lobes supported by prominence and local support and preserves unassigned signal. It neither fixes three nor counts atoms as species.

No. Use Recalculate lateral-projection components for an explicit review. Reviewing the projection neither reinverts nor reapplies corrections. Keep the earlier project and check the exported protocol.

Prepare view, zoom, theme and standards; use PNG on the completed plot. The dialog proposes 4800 pixels and 300 dpi and preserves proportions. More dpi adds no data; also retain project, parameters and review states.

Yes. The English gallery plays six movies with English synthetic narration, their own interface and timeline; the Spanish gallery retains six Spanish-narrated movies. CC changes captions, not the voice. Videos are hosted here and also included in the corresponding WPF packages.

No. These are different quantities. Review weighting, support and units. With unknown response Mn/Mw/PDI are unavailable; with an explicitly assumed mass weighting, descriptors are apparent, not validated absolute moments. A historical 1% threshold can differ from full workbench support.

Product and version, steps, sample/ramp/region, method and settings, applied factors, complete message and screenshot. Keep a minimal authorized project copy and its provenance. Do not share keys, tokens or confidential data; use Support and suggestions.