Baseline objective
—Waiting for accepted baseline shots
Cost —FERMI3 · EXPERIMENT MONITOR
M-LOOP progress
Read onlySaved results · no experiment controls
Connecting to the local monitor…
Best search objective
—No accepted search point
Cost —Search improvement
—Relative to the accepted baseline
No accepted comparison yetConfirmation objective
—No confirmation result yet
Search results remain unconfirmed0 submitted shotsSession budget unavailable
- Completed
- 0
- Accepted fits
- 0
- Rejected fits
- 0
- Pending
- 0
- Search points
- 0
Complete search points in this session
Best accepted search point
An interrupted run can retain a best measured point without completing confirmation. This is not proof of a global optimum.
| Variable | Best measured value | Units |
|---|
Images at a glance
Baseline, best point and latest image
Display colors clip sOD below 0 and above 3; saved arrays are unchanged. A representative shot is not an average of repeated images. Check its calibration and analysis below.
Every recorded shot
Objective history
Accepted fitRejected / failedPendingRepeat mean ± uncertainty
Choose a plotted shot to inspect its image, fit and variables. Missing values appear in a separate status row.
Selected shot
Choose a shot
No image selected
sOD uses the shared 0–3 display scale.
Explore the search
Variables and objective
Color shows the recorded objective. A correlation in this view does not establish an independent variable effect.
Exact saved values
Variables and bounds
Select a shot to see the recorded parameter source.
| Variable | Recorded value | Grid | Allowed range |
|---|
Image calibration and timing
Recorded stage durations
M-LOOP default views
Parameter history and measured cost
Normalized parameter vs run number
Cost vs normalized parameter
Normalized value = (recorded value − lower bound) / (upper bound − lower bound). Zero is the lower bound; one is the upper bound. Hover or keyboard-focus a marker for the exact value; select it to inspect the shot. Cost plots include accepted measurements only and do not establish an independent effect of each parameter.
Saved Gaussian-process learner
Model landscapes and fit history
Model plots appear only when a compatible trained learner archive is available.
Log₁₀ length scale vs fit
Noise variance vs fit
Landscapes are model predictions along one-dimensional slices, not measured scans or independent-variable effects. Their reference point, uncertainty convention and archive age are shown with the model. A stopped run may have useful measurements even if the GP never trained.
Parameter values across shots
Position within each allowed range
Lower boundUpper boundGray = missing · red outline = outside bounds
Each column uses its own bounds. Hover a cell for its exact value; click or keyboard-select a row to inspect all of that shot's values.
Audit trail
Recorded shots
| Shot | Phase / repeat | Status | Objective | Cost | Completed · local time | Image |
|---|