MIDAS_Main returns everything in one struct R (see API Reference for the full field list). This page is about what to actually look at once you have it, and what each diagnostic is telling you.

Did the run finish the way you expect?

Check R.stoppedEarly first. If true, R.t_final is less than R.params.t_tot (times checkFinT_Eq if used) and R.stopReason holds the error message that ended the run early - most commonly a trace-element boundary value going negative at an extreme DamA/DRG corner of parameter space (see Configuration Options). The state returned is the last fully self-consistent step, not a partial or corrupted one - safe to use, just shorter than requested.

Separately, a run can end with a hard error rather than a graceful R.stoppedEarly=true - most notably 'better stop here (crystal becomes negligible)' if the crystal resorbs below 5% of the domain. Since this happens before R is packaged, nothing is returned at all in that case; see Benchmarks for the P-T-path mechanism that can drive this and how the shipped examples avoid it.

Apparent age vs. true age: the core diagnostic

R.tALuHf1_final is the single-point apparent-age profile across phase A at the final step - what you’d calculate at every point if you (naively) assumed no diffusive resetting had occurred (see Equations). Compare it against R.t_final (the actual model time): where they agree, that part of the crystal has preserved its age record; where tALuHf1 diverges below t_final, diffusion has reset it. R.misfitApparent_final is exactly this gap, reduced to one worst-case number ($\max_x \lvert \tau(x) - t\rvert$); its full time evolution is in R.misfitApparentH if you want to plot it yourself. plot_velocity_age.m shows the same misfit spatially (vs. position and time) rather than reduced to a single curve.

If you’re using isochron ages instead (R.t_rimA_final/ R.t_coreA_final/R.t_bulkA_final/R.t_maxA_final), remember these depend on isoRefMode (which “second mineral” point they’re regressed against) - two runs with different isoRefMode are not directly comparable unless you know which reference point each used.

Is the run numerically trustworthy?

The mass-balance fields (R.dMB_max, R.dMBMn_max, R.dMBHfr_max - peak drift over the whole run for MgO/MnO/Hfr, species with no physical source or sink) are the main numerical-integrity check. See Benchmarks for what “small” means in practice.

History fields, if store_history = 1

Every <field>rec array (R.Srec, R.Vrec, R.Trec, R.Prec, R.CArec, …) is one row per recorded step, in the same order as R.trec. R.tA1/R.tB1 are the exception to the naming pattern - the full apparent-age history, phase A and phase B (not R.tALuHf1rec). Use these for anything time-resolved that isn’t already a dedicated plot_*.m function - e.g. plot(R.trec, R.Srec) for growth/resorption history, or indexing R.CArec(i,:) against R.xArec(i,:) for the composition profile at a specific recorded step i.

Which plot function answers which question

Question	Function
How does apparent age compare to true age, everywhere and always?	plot_velocity_age(R)
Is mass conserved (numerical sanity check)?	plot_massbalance_MgO(R)
What does the crystal look like now, all elements at once?	plot_all_composition_profiles(R)
How does age/composition evolve at a few fixed depths?	plot_age_at_fixed_positions(R) / plot_conc_at_fixed_positions(R)
Where’s the closure temperature?	plot_age_vs_temperature(R)

See API Reference for the complete list.