← Home

Documentation

Quantum@Home demonstrates small, home-scale physics experiments. It uses off-the-shelf, low-cost detectors and instruments that provide live data related to quantum events. The hardware is connected to the web server in Bladel, the Netherlands.

Available experiments

Natural radiation

The Natural Radiation page receives detection counts from a GQ GMC-300S Geiger counter, which uses a Geiger-Müller tube for detection. When ionizing radiation interacts with the gas in the tube, it can eject an electron from a gas atom. The resulting ionization triggers an avalanche of further ionizations and produces the electrical pulse recorded as a detection.

Live detections

The Live detections chart shows the most recent 60 one-second samples. A low baseline marker represents a recorded zero. A taller bar represents one or more detections during that second, and values above one are printed above the bar.

Detection history

The first row shows counts this minute, counts this hour, and counts today.

A value marked partial has incomplete live-detector coverage for that local period. The number is the count actually recorded; it is not an estimate of what the missing interval might have contained.

The second row shows average counts per day, week, and month:

• the daily average uses only completed earlier local days;
• the weekly average uses only completed Monday-through-Sunday weeks;
• the monthly average uses only completed calendar months.

A day qualifies as complete when at least 99% of its expected one-second live samples were received. Weeks and months qualify only when every included day is complete. Current and incomplete periods are excluded from these averages.

The long-term chart offers three views:

• the last 30 local days, with one bar per day;
• the last 26 local weeks, with one bar per Monday-through-Sunday week;
• the last 52 local weeks, with one bar per Monday-through-Sunday week.

A partial bar shows the subtotal that was actually recorded. A subdued baseline marker means no numeric measurement is available for that period. Recorded zero and unavailable data are therefore not treated as the same thing. Weekly partial totals include the recorded days that are available, without estimating missing days.

Counts per minute distribution

The distribution chart uses complete one-minute intervals from the retained live-detector history. Each bar shows the percentage of complete minutes containing a particular number of detections. The line shows the Poisson distribution expected from the observed mean count rate.

A Poisson distribution is expected when detections occur independently at an approximately steady average rate. Short data sets fluctuate visibly; the observed bars should generally approach the expected curve as more complete minutes accumulate. Partial minutes, unavailable periods, simulated data, and daily summary records are excluded.

About the detector

When the detector is live, the page shows its location, source status, model, firmware, detector type, and the manufacturer-documented safe range. The displayed range of 0–50 counts per minute corresponds to 72,000 counts per day. This manufacturer figure is contextual information for this detector; it is not a general medical or environmental safety assessment.

If the physical detector is not available, hardware-specific values are shown as unavailable.

Quantum randomness

The Quantum Randomness page contains a timed distribution experiment designed for random bytes produced by a Crypta Labs Cicada 0.5 USB quantum random number generator.

A byte can have any integer value from 0 through 255. If the QRNG output is uniform, every value has the same expected frequency:

1 / 256 = 0.390625%, displayed on the chart as approximately 0.391%.

Running an experiment

Select a duration of 0.1, 1, 10, 30, or 60 seconds and press Start. While the experiment is running, the chart updates with only the bytes processed during that run. Press Stop to finish early. After completion or stopping, the final chart and summary values remain visible until another run begins.

The six summary boxes show:

• the selected duration;
• elapsed time;
• bytes processed;
• the arithmetic mean of the byte values;
• the standard deviation of the 256 observed percentages around the expected percentage;
• experiment status.

For a perfectly uniform byte stream, the expected mean is 127.5. The displayed standard deviation summarizes the spread of the 256 observed bin percentages around 0.390625%. It is shown as an absolute percentage difference; technically, that difference is measured in percentage points.

Distribution chart

The chart contains one bar for each possible byte value from 0 through 255. The dashed reference line marks the expected uniform frequency of approximately 0.391% per value.

Short experiments naturally show larger irregularities because they contain fewer bytes. As more bytes are processed, the distribution should usually become more even, although a genuinely random result is never required to look perfectly uniform.

Pointing to a bar shows its byte value, count, and percentage. Bars that exceed the fixed visible chart range are clipped at the top, but their underlying values and tooltips remain unchanged.

What is quantum here?

Inside the QRNG, a controlled light source emits photons toward a sensor. Even when the average light level is steady, the exact number of photons emitted and detected during each very short sampling interval fluctuates unpredictably. This photon shot noise arises from the quantum nature of light: individual photon-emission and detection events have probabilities, but their precise outcomes cannot be predicted.

The sensor converts those fluctuations into an electrical signal, which is digitized into raw values. The device checks that the measured noise remains dominated by the quantum process rather than by ordinary electronic or environmental noise. The raw data is then conditioned to remove bias and produce a uniform stream of random bits. Conditioning does not create the randomness; the quantum measurement process remains its source.

The resulting bits are sent to the computer through USB and combined into the byte values shown in the experiment.

About the QRNG

The page shows the QRNG location, device status, model, firmware, quantum source, and output rate.

Device status refers exclusively to the physical QRNG hardware. It shows Connected only when that device is actually detected and connected; otherwise it shows Disconnected. It does not change merely because an experiment starts or stops.

When the physical device is connected, the quantum source is photon shot noise and the specified output rate is 500 kb/s. Hardware-specific values that are not available are displayed accordingly.

Schrödinger’s cat

The Schrödinger’s Cat page uses quantum random data from the Crypta Labs Cicada 0.5 USB QRNG to simulate the radioactive decay of a single rubidium-82 nucleus.

What you are seeing

The box contains a virtual cat and a simulated radioactive device that releases poisonous gas if the nucleus decays. Rubidium-82 has a half-life of approximately 75 seconds. While the box remains closed, the display shows the changing probabilities of the cat being alive or dead. After 75 seconds, the box opens automatically and reveals one of those outcomes.

A half-life of 75 seconds means that a rubidium-82 nucleus has a 50% probability of decaying during that interval. It does not mean that every nucleus decays after exactly 75 seconds. The decay time of an individual nucleus is unpredictable, and shorter decay times are more likely than longer ones.

What is quantum here?

Schrödinger’s cat is a thought experiment proposed in 1935 to expose the strange consequences of extending quantum mechanics from microscopic systems to everyday objects. In the idealized setup, a cat is sealed in a box with a radioactive nucleus, a detector, and a mechanism that releases poison if the nucleus decays. Before measurement, quantum theory describes the nucleus as a superposition of decayed and undecayed states. Because the detector, poison mechanism, and cat are linked to that event, the complete system is described as entangled with it until the box is opened and an outcome is observed.

This website does not create a macroscopic quantum superposition. Instead, the QRNG supplies random data generated from photon shot noise, a genuinely quantum physical process. The software converts that data into a simulated decay time following the same exponential probability law as radioactive decay. If the simulated nucleus decays before the box opens, the cat dies; otherwise, it remains alive. The physical quantum randomness is real, while the rubidium-82 nucleus, decay, poison mechanism, and cat are simulated.

About the experiment

The page shows the experiment location and type, the randomness source, the QRNG’s quantum source, the simulated nucleus, its half-life, and how the decay time is obtained. In absence of actual radioactive material, the decay time is simulated from the physical quantum randomness supplied by the QRNG.

Atomic spectrum

The Atomic Spectrum page observes the light from a continuously operating neon discharge lamp with a Thunder Optics Mini USB Spectrometer. Excited neon atoms emit light only at particular wavelengths because their electrons can move between discrete quantum energy levels. The spectrometer separates that light by wavelength, and its camera records the resulting spectrum.

This experiment is live

The neon lamp is burning continuously. Inside it, free electrons are colliding with neon atoms and exciting their electrons to higher energy levels. Those excited electrons are returning to lower levels and emitting photons at this very moment. The spectrometer separates that freshly emitted light by wavelength, its camera records the spectrum, and the resulting data are analyzed and sent directly to the webpage.

The spectral lines shown as live measurements are therefore not a prerecorded illustration. They are derived continuously from light produced by atomic transitions occurring in the lamp at the time the page is being viewed. If live acquisition is unavailable, the page says so explicitly and can show a saved real measurement instead.

What you are seeing

The four aligned displays use the same 450–700 nm wavelength geometry:

• Full visible spectrum is a continuous color reference across the displayed wavelength range.
• Live spectrometer image is the current cropped camera image from the spectrometer, without cosmetic enhancement.
• Detected spectral features shows wavelengths found by the blind detector. The vertical lines are equal-height display markers; their height, width, brightness, and color intensity do not represent measured optical intensity.
• NIST strong Ne I reference lines shows 56 strong neutral-neon reference wavelengths from the National Institute of Standards and Technology for comparison.

Pointing a mouse pointer anywhere across the detected or NIST strips shows the wavelength corresponding to that horizontal position.

The bright NIST reference lines were observed by the spectrometer during validation; the faint lines are additional NIST lines that were not observed. This is because not every reference line is detectable: lamp conditions and the spectrometer’s sensitivity and resolution (2–3 nm) limit which lines can be detected.

The NIST line brightness is therefore only a fixed observed/not-observed classification from validation. It does not represent the reference-line intensity, the lamp intensity, or the measurement intensity. All 56 NIST lines remain present in the display, and changing Low, Medium, or High sensitivity does not change which NIST lines are bright or faint.

You may see lines appear or disappear, more so at higher detection sensitivity. Lines may also shift slightly left or right. Both effects are caused by noise in the raw detector data, which can move a line’s measured signal above or below the detection threshold and can slightly change the position at which its peak is detected. The spectrometer’s finite resolution also limits how precisely a line’s wavelength can be located.

What is quantum here?

Neon is a noble gas with atomic number 10: a neutral neon atom has 10 protons in its nucleus and 10 electrons surrounding it. Inside the lamp, an electric discharge accelerates free electrons, which collide with neon atoms and can transfer enough energy to push one of their electrons into a higher, excited energy level. Those excited states are temporary.

Quantum mechanics allows the electron to return to a lower energy level, and when it does, the atom emits a photon whose energy exactly matches the difference between the two levels. Because photon energy determines wavelength, each transition produces light at a particular wavelength. Neon has many possible transitions, so its light contains many distinct spectral lines rather than a continuous spread of colors.

The spectrometer separates this light by wavelength and projects it onto a camera sensor, where different wavelengths appear at different positions. From that live detector image, the experiment identifies the spectral lines produced by the neon lamp. Their distinctive pattern is a directly observable consequence of the quantized energy structure of neon atoms.

This experiment does not detect individual photons and does not claim that the camera image provides calibrated optical line intensities. The measured camera signal is used internally to locate spectral lines; the public detected-line display intentionally uses equal-height markers so that wavelength, rather than uncalibrated intensity, is the quantity being compared.

Measurement and line detection

The spectrometer camera supplies 1920 × 1080 video frames at approximately 5 frames per second. Processing first applies a horizontal flip, then selects pixels 0 through 1423, which corresponds to a range of approximately 450–700 nm. Vertically, it uses camera rows 450 through 609 inclusive, a fixed 160-pixel-high region.

For each usable frame, the software converts that region into a one-dimensional spectral profile by taking the brightest 5% of its vertical pixels at each horizontal position. This favors the illuminated spectral trace while reducing the influence of darker parts of the camera frame. No smoothing or cosmetic image enhancement is applied.

A production measurement is formed from the profiles accumulated over at least approximately 0.5 seconds, normally two or three camera frames. Their profiles are averaged for line detection. The displayed live camera strip is the latest raw individual frame from the same measurement period, rather than a temporally averaged or enhanced image.

Line detection is deliberately blind to the NIST reference list. The software first estimates the profile background from its median and a robust noise level from the median absolute deviation. Candidate lines must exceed both an intensity threshold and a prominence threshold measured against the surrounding profile within ±5 nm; candidates closer than 2.5 nm are reduced to the strongest one. There is no NIST-assisted peak selection, line fitting, smoothing, or reference-assisted calibration.

The three sensitivity settings change only the detector thresholds:

• Low: intensity above background + 4σ and prominence of at least 6σ;
• Medium: intensity above background + 3σ and prominence of at least 4.5σ;
• High: intensity above background + 2σ and prominence of at least 3σ.

Here, σ is a robust estimate of detector noise, calculated as 1.4826 times the median absolute deviation from the profile median. Higher sensitivity can reveal measured lines closer to the noise level, but it can also admit more transient noise-related detections. Acquisition and profile generation occur only once for each measurement; Low, Medium, and High are alternative analyses of that same profile.

Wavelength calibration and validation

The wavelength scale uses a fixed two-point linear calibration based on two neon lines near 585.25 nm and 640.21 nm. The calibration maps the full camera width to wavelength; the page then displays the 450–700 nm portion.

A separate calibration diagnostic compared multiple measured neon lines with reference wavelengths. A fixed-correspondence affine fit and a robust fit both indicated only negligible additional correction and no meaningful wavelength-dependent trend, so the established two-point calibration was retained unchanged rather than tuned for a prettier match to the reference spectrum.

The instrument’s wavelength resolution is approximately 2–3 nm. Measured line positions therefore need not coincide exactly with NIST wavelengths, and nearby neon transitions can blend into a single detected line. The comparison should be interpreted at that resolution rather than as sub-nanometer spectroscopy.

NIST reference comparison

The reference strip uses 56 strong Ne I air wavelengths between 450 and 700 nm from the NIST reference table:

https://www.physics.nist.gov/PhysRefData/Handbook/Tables/neontable2_a.htm

The NIST list is not used to detect live lines. The bright/faint display classification was created separately from a dedicated validation capture of the real setup lasting about two minutes. High-sensitivity detections were clustered without reference information; only clusters appearing in at least 10% of completed measurements were retained, and NIST matching was performed afterward by assigning each persistent cluster to the nearest NIST strong line within 2.5 nm. Eleven NIST lines were classified as observed by this setup and are shown bright; the remaining 45 remain visible but faint.

This is a fixed classification defined in the source configuration. It does not change with the live spectrum or with the user’s sensitivity selection, and it never feeds back into calibration or line detection.

About the setup

The page reports the spectrometer connection state and identifies the physical setup as a Thunder Optics Mini USB Spectrometer observing a neon discharge lamp over the 450–700 nm displayed range. The spectrometer’s nominal wavelength resolution is approximately 2–3 nm and its output is live USB video.

Time and persistence

Calendar totals and chart periods use Europe/Amsterdam local time. Internally, minute timestamps may remain in UTC so that measurements stay unambiguous during daylight-saving-time transitions.

Live natural-radiation history is stored across application restarts. A sudden interruption can lose a small amount of the newest unsaved data, and any resulting gap is reflected by a partial period rather than by an estimated replacement.

Quantum randomness experiment results are not stored as long-term history. A completed chart remains on the page for the current browser session until another run begins or the page is reloaded.

Schrödinger’s cat outcomes are simulation results rather than detector history and are not included in the Natural Radiation records.

Atomic Spectrum does not maintain a growing historical spectrum. The first valid completed live measurement is saved as one representative fallback data set consisting of the displayed image, the measured one-dimensional profile, and measurement metadata. It is not periodically replaced. If live spectrometer acquisition is unavailable, the page can show this saved real measurement and labels it as saved data rather than live data. Detected lines for the fallback are recalculated from the saved measured profile using the same Low, Medium, or High sensitivity selected by the visitor.

Language support

The site supports English, Dutch, German, and Spanish. The language switch in the header cycles through those languages and keeps the visitor on the same page whenever possible.

Help pages

Documentation, Glossary, About, Privacy Policy, and Terms of Use are served through the shared Help navigation. The Markdown content is loaded per language, while the surrounding layout remains shared.