Your data / exact continuous search
How far to flip the result?
Enter 3–21 readings and a chosen target. Odd samples use their middle reading; even samples average the central pair. Set the permitted movement range, including signed values. The lab checks every real position using exact fractions, then exports a certificate for independent verification.
01 / Model assumptions
Set the readings and bounds
The target and allowable range are stipulated by you. The result establishes a conditional mathematical statement, not which estimate should be trusted for real measurements.
The calculation runs in this browser. Values are not put into a URL or sent to an application server. The certificate contains every input; share it only if those values may be disclosed.
02 / Exact result
Every one-reading route
The starting estimate and exact routes will appear here.
03 / Reproduce
Take the evidence with you
The partition includes every median transition and exact factor root. Each open interval has a constant winner; the printed positions of the bars are approximate, while the exported fractions are exact. A nearest threshold is a tie and is an infimum of movement, usually not an attained smallest strict reversal.
Check a downloaded certificate independently, browse your local Research Archive, or run the Python checker from the repository root:
python3 tools/verify_odd_sample_flip.py odd-sample-flip-certificate.jsonSaving rechecks the certificate with a separate browser verifier and keeps it in this device's local archive. A passing checker verifies the mathematical claim under the entered inputs. The download itself does not authenticate measurements or prove an empirical theory. Explore the checked even-sample report, the signed example, or read the proof argument and limits.