Exploring results

Every page so far has looked at a result right after it finished: a single climbing curve, a final value. Once a result exists, there's more to it than that: every one of a node's own states plotted together, a timeline you can scrub or play back at any speed and jump controls for getting back to a specific sample without dragging. This page covers all three, on a 10-second offline run of the "Thermal Management" example.

A node's full result, every state together

Battery module carries two states: Temperature and State of charge. Opening its Results tab plots both across the whole run, not just the one you'd see on the canvas label.

This is the plot for one node's own states, not a cross-node comparison tray. There's no way yet to pin signals from Enclosed air and Electrical losses onto this same chart. That's a natural thing to want and it's on the project's roadmap, just not built yet.

Scrubbing the timeline

Dragging the timeline slider moves a shared time cursor: the plot's dashed marker, the node cards on the canvas and the numeric time readout all follow it together.

Nothing here is a live recomputation. The whole result already exists from the run. Scrubbing just changes which already-computed sample you're looking at, which is why it's instant regardless of how long the original run took.

Playing it back

Rather than dragging by hand, the transport can advance on its own: play/pause and a rate selector independent of however the simulation itself was paced when it ran.

That independence matters for a run that took real wall-clock time to produce (an online/paced run, from "Running a simulation"). You're never stuck reviewing it at the same speed it took to generate.

Jumping to a specific sample

Beyond dragging and playing, four buttons move by exact samples: first, previous, next and last.

Landing on the very first sample is a quick way to confirm what a run actually started from, especially useful after "Running a simulation"'s Extend simulation stitches a second run onto the first. The timeline stays one continuous sequence, and the start button always finds the true beginning of it.

Where to go from here

Everything up to here has been about a model you built and ran yourself. Two remaining pieces work differently: the model assistant proposes a model from a plain-language description, and causal inference proposes one from a CSV of data instead of either.