Four-hour pairs separate two fixed GJ 3543 alias models in a 40-visit Gaussian-noise test

We test two known GJ 3543 velocity periods with public HARPS times and a projection method. Under a stable 2.73 m/s sinusoid and independent Gaussian errors, 20 four-hour pairs give the correct fixed-period choice in 97.0 to 97.4 percent of 6000 trials per branch. Forty visits at one sidereal phase give about 50 percent. The paired result remains near 97 percent with both published central periods. Actual RVBank times retain about 14 percent of the mean signal power after a wrong-alias fit. Three velocity products favor the short branch, but a stable sinusoid leaves excess residuals. The products share spectra. Correlated noise, period searches, changing activity, and visibility are not modeled. No new planet or independent instrument confirmation is reported. Code, frozen inputs, source hashes, numerical checks, tables, and offline steps are included.

astronomy · GJ 3543 · HARPS · daily alias · observation design · radial velocity · conditional simulation