APEX Daily Pulse — 2026-11-12
Matching Samples Do Not Identify Frequency
[✅ Verified] Pillar: Sensing and collection
[❓ Unverified] Estimated duration: 3–5 minutes
Reactivation target: Expose a frequency alias, then change the acquisition rate to distinguish two candidate sources.
Prerequisites
- Python 3.12+ with its standard math module.
- A Bash-compatible terminal on owned equipment.
- No privileges, network access, SDR hardware, or RF transmission.
Actions
- 1. Before running the fixture, predict whether 125 Hz and 875 Hz can produce the same real-valued samples at 1,000 samples per second.
- 2. Run the complete comparison. Both signals have unit amplitude, zero phase, and 64 samples:
python3 - <<'PY'
import math
def capture(frequency_hz: float, rate_hz: float) -> list[float]:
"""Generate 64 ideal real-valued cosine samples."""
return [
math.cos(math.tau * frequency_hz * n / rate_hz)
for n in range(64)
]
def maximum_error(first: list[float], second: list[float]) -> float:
"""Return the largest corresponding sample difference."""
return max(
abs(a - b)
for a, b in zip(first, second, strict=True)
)
tolerance = 1e-10
error_1000 = maximum_error(
capture(125.0, 1000.0),
capture(875.0, 1000.0),
)
error_1200 = maximum_error(
capture(125.0, 1200.0),
capture(875.0, 1200.0),
)
indistinguishable_1000 = error_1000 < tolerance
distinguishable_1200 = error_1200 > 1.0
passed = (
indistinguishable_1000
and distinguishable_1200
)
print(f"samples_per_capture=64")
print(f"error_at_1000={error_1000:.3e}")
print(f"error_at_1200={error_1200:.3e}")
print(f"indistinguishable_at_1000={indistinguishable_1000}")
print(f"distinguishable_at_1200={distinguishable_1200}")
print(f"drill_passed={passed}")
raise SystemExit(0 if passed else 1)
PY
- 3. Compare the two error values. At 1,000 samples per second, the maximum difference must be below
1e-10. At 1,200 samples per second, it must exceed1.0.
- 4. State the permitted conclusion: the second rate distinguishes these two candidates. It does not establish a unique source frequency among all possible frequencies.
Observable signal
[❓ Unverified] Expected flags; the fixture was not executed in this run:
samples_per_capture=64
indistinguishable_at_1000=True
distinguishable_at_1200=True
drill_passed=True
[✅ Verified] Mechanism: at integer sample index n, the 875 Hz signal at a 1,000 Hz sample rate is cos(2πn − 2π·125n/1000). This equals the sampled 125 Hz cosine. The sample sequences therefore match, apart from floating-point error. A longer capture at the same rate cannot separate this ideal pair.
Success condition
Both candidate signals match at the first rate. They differ at the second rate. The operator rejects a unique frequency claim based only on the first capture.
Failure signature
[✅ Verified] Reporting “source confirmed at 125 Hz” from the first capture alone is unsupported: the 875 Hz fixture produces the same observations under the declared conditions.
Python unavailable: POSIX awk alternative
Use awk and its built-in cosine function:
awk '
BEGIN {
pi = atan2(0, -1)
rate[1] = 1000
rate[2] = 1200
for (k = 1; k <= 2; k++) {
peak[k] = 0
for (n = 0; n < 64; n++) {
a = cos(2 * pi * 125 * n / rate[k])
b = cos(2 * pi * 875 * n / rate[k])
difference = a - b
if (difference < 0) difference = -difference
if (difference > peak[k]) peak[k] = difference
}
printf "rate=%d maximum_error=%.3e\n", rate[k], peak[k]
}
passed = peak[1] < 1e-10 && peak[2] > 1
print "drill_passed=" (passed ? "true" : "false")
exit(passed ? 0 : 1)
}'
Boundary check
[✅ Verified] Hidden assumption: this fixture models real-valued sampling without an analog input filter. Complex I/Q sampling has different signed-frequency behavior. A correctly selected analog filter can reject the higher-frequency candidate before sampling; a digital filter applied after aliasing cannot recover the lost distinction.
[❌ Speculative] In a hardware drill, a second acquisition rate adds operator work and can change gain, bandwidth, or timing. Record those settings. Prefer an input filter when the permitted source band is known; use rate changes as a diagnostic when it is not.
Adaptation question
Which APEX frequency claims depend on a single sample rate without a verified input-band limit?