Calculator overview
Inputs and outputs
This summary comes from the calculator's published input and output contract.
Inputs
- Outcome 1 weight (required)
-
Default 1 Range 0 to 1000
About this input
Discrete: the relative weight of outcome 1, from 0 to 1000; its probability is its weight divided by the total of the six weights.
- Outcome 2 weight (required)
-
Default 1 Range 0 to 1000
About this input
Discrete: the relative weight of outcome 2, from 0 to 1000; its probability is its weight divided by the total of the six weights.
- Outcome 3 weight (required)
-
Default 1 Range 0 to 1000
About this input
Discrete: the relative weight of outcome 3, from 0 to 1000; its probability is its weight divided by the total of the six weights.
- Outcome 6 value (required)
-
Default 6 Range -1000 to 1000
About this input
Discrete: the value of outcome 6, from -1000 to 1000, for example a payout.
- Outcome 3 value (required)
-
Default 3 Range -1000 to 1000
About this input
Discrete: the value of outcome 3, from -1000 to 1000, for example a payout.
- Outcome 4 value (required)
-
Default 4 Range -1000 to 1000
About this input
Discrete: the value of outcome 4, from -1000 to 1000, for example a payout.
- Outcome 5 value (required)
-
Default 5 Range -1000 to 1000
About this input
Discrete: the value of outcome 5, from -1000 to 1000, for example a payout.
- Prediction, low (required)
-
Default 0 Range -1000 to 1000
About this input
Challenge 1: the smallest number of heads in your range for 500 flips.
- Seed (required)
-
Default 2026 Range 1 to 2147483646
About this input
The starting number of the random sequence, a whole number from 1 to 2147483646. The same seed always gives exactly the same draws, so a run can be repeated and checked.
- Number of trials (required)
-
Default 500 Range 10 to 500
About this input
How many trials to run: a multiple of 10 from 10 to 500. More trials narrow the confidence band, roughly by the square root of the count.
- Prediction, high (required)
-
Default 0 Range -1000 to 1000
About this input
Challenge 1: the largest number of heads in your range for 500 flips.
- Outcome 4 weight (required)
-
Default 1 Range 0 to 1000
About this input
Discrete: the relative weight of outcome 4, from 0 to 1000; its probability is its weight divided by the total of the six weights.
- Outcome 5 weight (required)
-
Default 1 Range 0 to 1000
About this input
Discrete: the relative weight of outcome 5, from 0 to 1000; its probability is its weight divided by the total of the six weights.
- Outcome 6 weight (required)
-
Default 1 Range 0 to 1000
About this input
Discrete: the relative weight of outcome 6, from 0 to 1000; its probability is its weight divided by the total of the six weights.
- Distribution (required)
-
Default Uniform Allowed Uniform, Triangular, Exponential, Discrete
About this input
Distribution: Uniform (every value from the low to the high end equally likely), Triangular (most likely at the mode), Exponential (waiting times with the given mean) or Discrete (the six outcomes and weights below).
- Distribution high end (required)
-
Default 1 Range -1000 to 1000
About this input
Uniform and Triangular: the highest possible value, from -1000 to 1000 and above the low end.
- Distribution low end (required)
-
Default 0 Range -1000 to 1000
About this input
Uniform and Triangular: the lowest possible value, from -1000 to 1000 and below the high end.
- Dice a trial (required)
-
Default 2 Range 1 to 6
About this input
Dice: how many fair six-sided dice each trial rolls, from 1 to 6; the outcome is their total.
- Challenge number (required)
-
Default 0 Range 0 to 5
About this input
0 for free exploration, or 1 to 5 to load a challenge. A challenge fixes its own settings and leaves you only the inputs it names.
- Coins a trial (required)
-
Default 1 Range 1 to 6
About this input
Coin flips: how many coins each trial flips, from 1 to 6; the outcome is the number of heads.
- Confidence level (required)
-
Default 0.95 Range 0.8 to 0.99
About this input
How sure the intervals should be, from 0.8 to 0.99: 0.95 gives 95% intervals, which miss the exact value about one run in twenty.
- Probability of heads (required)
-
Default 0.5 Range 0 to 1
About this input
Coin flips: the chance each coin lands heads, from 0 to 1; 0.5 is a fair coin. For a reliability model, read heads as a part that works.
- Outcome 1 value (required)
-
Default 1 Range -1000 to 1000
About this input
Discrete: the value of outcome 1, from -1000 to 1000, for example a payout.
- Outcome 2 value (required)
-
Default 2 Range -1000 to 1000
About this input
Discrete: the value of outcome 2, from -1000 to 1000, for example a payout.
- Exponential mean (required)
-
Default 1 Range 0.01 to 1000
About this input
Exponential: the mean of the distribution, from 0.01 to 1000.
- Distribution mode (required)
-
Default 0.5 Range -1000 to 1000
About this input
Triangular: the most likely value, from the low end to the high end.
- Event threshold (required)
-
Default 1 Range -1000 to 1000
About this input
The event counted in every experiment except pi: an outcome at or above this number, for example a dice total of at least 10, or at least one working part.
- Experiment (required)
-
Default Estimate pi Allowed Estimate pi, Coin flips, Dice, Distribution
About this input
What each trial does: Estimate pi drops a point in a square; Coin flips flips coins; Dice rolls dice; Distribution draws one value from the distribution you set.
Outputs
- Confidence interval, high
-
No unit declared
About this output
The high end of the confidence interval: the estimate plus z standard errors.
- Histogram of the outcomes against the exact expected counts
-
No unit declared
About this output
The outcomes counted in bins, beside the counts the exact probabilities predict.
- Coins or dice a trial
-
No unit declared
About this output
The coins or dice each trial used; blank for pi and Distribution.
- Confidence interval, low
-
No unit declared
About this output
The low end of the confidence interval: the estimate minus z standard errors.
- Probability of heads used
-
No unit declared
About this output
The heads probability of a coin run; blank otherwise.
- Exact value inside the interval
-
No unit declared
About this output
Yes when the exact value lies inside the confidence interval, No when it does not (which happens by chance in about one run in twenty at 95%).
- Event threshold used
-
No unit declared
About this output
The event threshold of the run; blank for pi, whose event is a point inside the circle.
- Experiment run
-
No unit declared
About this output
The experiment that was run; in a challenge, the challenge's own.
- Exact value
-
No unit declared
About this output
The exact value the estimate aims at: pi, the expected number of heads, the expected dice total, or the distribution's mean.
- Standard error
-
No unit declared
About this output
The estimated standard deviation of the estimate itself: the sample standard deviation divided by the square root of the number of trials.
- Seed used
-
No unit declared
About this output
The seed of the run.
- Trials run
-
No unit declared
About this output
The number of trials run.
- The first 10 trials, coin by coin or die by die
-
No unit declared
About this output
The first 10 trials coin by coin (1 heads, 0 tails) or die by die, for revealing the results one at a time. Blank for pi and Distribution.
- Trials: 500 rows computed from the seeded stream
-
No unit declared
About this output
One row per trial: x and y (the point for pi; the first two coins or dice; the draw and the sampled value for a distribution), the outcome, and the histogram bin the trial falls in. Rows past the trial count are blank.
- Model version
-
No unit declared
About this output
The model version.
- Model Status
-
No unit declared
About this output
OK when the run is computed with nothing to flag; CHECK when it needs attention, for example fewer than 30 trials or an interval that misses the exact value; NOT VALID, with the reason, when an input is out of range.
- Running estimate every 10 trials, with its confidence band
-
No unit declared
About this output
The estimate after every 10 trials with its confidence band, and 1 when the exact value lies inside that band (0 when it does not): the band narrows as trials accumulate.
- Formula module versions
-
No unit declared
About this output
The formula modules the model is built from, with their versions: the seeded generator and the two interval methods.
- Critical value z
-
No unit declared
About this output
The normal critical value z of the confidence level, 1.96 for 95%.
- Confidence level used
-
No unit declared
About this output
The confidence level of both intervals in this run, for example 0.95.
- Distribution used
-
No unit declared
About this output
The distribution sampled in a Distribution run; blank otherwise.
- The discrete distribution used
-
No unit declared
About this output
The discrete distribution used: each outcome's value, weight and probability. Blank unless a Discrete distribution was sampled.
- Challenge result
-
No unit declared
About this output
MET when the selected challenge's goal is reached, NOT MET when it is not, or No challenge selected.
- Challenge goal
-
No unit declared
About this output
What the selected challenge asks you to do.
- Challenge feedback
-
No unit declared
About this output
A short explanation when the challenge is met, or a hint when it is not.
- Challenge score
-
No unit declared
About this output
100 when the selected challenge is met, otherwise 0.
- Challenge measure
-
No unit declared
About this output
The number the selected challenge is scored on, for example the exact probability of your range in challenge 1.
- Event
-
No unit declared
About this output
What the event is in this experiment.
- Event interval (Wilson), low
-
No unit declared
About this output
The low end of Wilson's interval for the event probability.
- Event probability, exact
-
No unit declared
About this output
The exact probability of the event.
- Event share, estimated
-
No unit declared
About this output
The share of trials with the event.
- Event interval (Wilson), high
-
No unit declared
About this output
The high end of Wilson's interval for the event probability.
- Estimate minus exact
-
No unit declared
About this output
The estimate minus the exact value.
- Estimate
-
No unit declared
About this output
The Monte Carlo estimate: the mean outcome over all trials.
- Trials with the event
-
No unit declared
About this output
How many trials had the event.
- What is estimated
-
No unit declared
About this output
What the estimate measures in this experiment, for example 4 times the share of points inside the circle.
LogicCommons is in beta. If a result, label, or reference looks wrong, tell us here; we read every message.
Methodology
Purpose and model boundary
This lab runs a seeded Monte Carlo experiment of up to 500 trials and shows how random sampling estimates a probability or an average, and how uncertain that estimate is. Each trial does one of four things: drops a point in a square to estimate π, flips 1 to 6 coins, rolls 1 to 6 fair six-sided dice, or draws one value from a Uniform, Triangular, Exponential or six-outcome Discrete distribution. For the run it returns every trial, the estimate (the mean outcome) with its confidence interval, the share of trials with an event and an interval for it, the exact values both should approach, a histogram against the exact expected counts, and a running estimate every 10 trials. Five scored challenges use the same model.
It is a learning tool. The random numbers come from a small pseudo-random generator, the run is limited to 500 trials, and the intervals at these counts are wide. It is not a tool for gambling, security, or any decision that depends on the outcomes. Sampling without replacement (urns, birthday collisions) and other distributions are not modelled.
The model is the calculation authority. The page sends the values you set to the model and draws only what the model returns; every point, coin, die, bar and band on the canvas is a model value, and nothing on the page draws a random number.
Inputs and units
All inputs are unitless numbers or choices. Every input is required and is checked on its entered value even when the chosen experiment does not use it.
| Input | Meaning and unit |
|---|---|
| Experiment | Estimate pi, Coin flips, Dice or Distribution: what each trial does. |
| Seed | The starting number of the random sequence, a whole number from 1 to 2,147,483,646. The same seed always gives exactly the same trials. |
| Number of trials | T, a multiple of 10 from 10 to 500. |
| Coins a trial | Coin flips: 1 to 6 coins a trial; the outcome is the number of heads. |
| Probability of heads | Coin flips: the chance p that each coin lands heads, 0 to 1 (0.5 is a fair coin). Read heads as "a part works" for a reliability model. |
| Dice a trial | Dice: 1 to 6 fair six-sided dice a trial; the outcome is their total. |
| Event threshold | The event in every experiment except π is an outcome at or above this number, -1,000 to 1,000. For π the event is a point inside the circle. |
| Distribution | Distribution: Uniform, Triangular, Exponential or Discrete. |
| Distribution low end, high end | Uniform and Triangular: the lowest and highest possible values a and b, -1,000 to 1,000, with a below b. |
| Distribution mode | Triangular: the most likely value c, from a to b. |
| Exponential mean | Exponential: the mean μ, 0.01 to 1,000. |
| Outcome 1 to 6 value | Discrete: the value of each of the six outcomes, -1,000 to 1,000 (for example a payout). |
| Outcome 1 to 6 weight | Discrete: the relative weight of each outcome, 0 to 1,000, with the six adding to more than 0. An outcome's probability is its weight over the total weight. |
| Confidence level | The level of both intervals, 0.8 to 0.99; 0.95 gives 95% intervals. |
| Challenge number | 0 for free exploration, or 1 to 5 to load a challenge. |
| Prediction, low and high | Challenge 1 only: the fewest and most heads of your predicted range, -1,000 to 1,000. |
Where each input is set:
- In the input panel: the seed, the number of trials and the challenge number.
- On the canvas only: the experiment (chosen from its tiles), the distribution (its buttons), the settings of the chosen experiment (coins, heads probability, dice, threshold, the distribution's ends, mode or mean, and the six discrete values and weights), the confidence level, and challenge 1's two predictions.
- Also on the canvas: the number of trials can be dragged along the n axis of the running estimate (in tens from 10 to 500), the event threshold can be dragged along the histogram's axis, and the seed can be typed or stepped through a fixed list of seeds with New seed.
Governing relationships
The generator. The model implements Park and Miller's minimal standard generator (MINSTD) with the multiplier 48271. Starting from s₀ = the seed, each state is
sₖ = MOD(48271 · sₖ₋₁, 2³¹ − 1), and each draw is uₖ = sₖ / (2³¹ − 1),
which lies strictly between 0 and 1. The first draw is therefore 48271 · seed reduced modulo 2,147,483,647, not the seed itself. Every product stays below 2⁴⁷, so the integer sequence is exact. Every trial uses six consecutive draws, whatever the experiment: trial t uses draws 6t − 5 to 6t, called u₁ to u₆ below. The stream therefore does not depend on the experiment; π and one coin simply leave some of the six unused.
One trial.
- Estimate pi: x = 2u₁ − 1 and y = 2u₂ − 1 give a point in the square from −1 to 1. The outcome Y is 4 when x² + y² ≤ 1 (inside the circle) and 0 otherwise, so the mean of Y estimates π. The event is "inside".
- Coin flips: coin j is heads (1) when uⱼ < p, otherwise tails (0); Y is the number of heads among the first n coins.
- Dice: die j shows INT(6uⱼ) + 1; Y is the total of the first n dice.
- Distribution, by the inverse-transform method on u₁:
- Uniform: Y = a + (b − a)·u₁.
- Triangular: with F_c = (c − a)/(b − a), Y = a + √(u₁(b − a)(c − a)) when u₁ < F_c, otherwise Y = b − √((1 − u₁)(b − a)(b − c)).
- Exponential: Y = −μ·ln(1 − u₁).
- Discrete: with cumulative weights C₁ … C₆ and total W = C₆, the outcome index is 1 plus the number of C₁ … C₅ that u₁·W reaches or passes; Y is that outcome's value.
- For every experiment except π the event is Y ≥ threshold.
Estimate and its interval. Running sums of Y, Y² and the event count are kept trial by trial. After n trials:
mean = ΣY / n, variance s² = max(0, (ΣY² − n·mean²) / (n − 1)), standard error = √(s² / n),
and the confidence interval is mean ± z · standard error, the normal-approximation interval for a mean, with z = NORM.S.INV(1 − (1 − level)/2) (1.96 at 0.95). The final results use n = T; the running estimate repeats the calculation at n = 10, 20, … , T.
Event share and Wilson's interval. The event share is p̂ = (trials with the event) / T. Its interval is Wilson's score interval: with d = 1 + z²/T, the centre is (p̂ + z²/(2T)) / d and the half-width is z·√(p̂(1 − p̂)/T + z²/(4T²)) / d.
Exact values.
| Experiment | Exact mean of Y | Exact probability of the event |
|---|---|---|
| Estimate pi | π | π/4 |
| Coin flips | n·p | with k = the threshold rounded up: 1 when k ≤ 0, 0 when k > n, else 1 − BINOM.DIST(k − 1, n, p, TRUE) |
| Dice | 3.5·n | the sum of the exact dice-total probabilities at or above the threshold |
| Uniform | (a + b)/2 | 1 at or below a, 0 at or above b, else (b − threshold)/(b − a) |
| Triangular | (a + b + c)/3 | 1 − F(threshold), F the triangular distribution function |
| Exponential | μ | 1 when the threshold ≤ 0, else e^(−threshold/μ) |
| Discrete | Σ wₖvₖ / W | (the sum of the weights whose value is at or above the threshold) / W |
The exact dice-total probabilities are built by convolution: one die gives 1/6 for each face, and the total of m dice at s is the average of the (m − 1)-dice probabilities at s − 1 to s − 6.
Histogram. Each trial is assigned a bin, and the count of a bin is the number of trials in it. The expected count is T times the bin's exact probability.
- Estimate pi: two bins, outcome 0 (outside) and outcome 4 (inside).
- Coin flips: one bin for each number of heads, 0 to n.
- Dice: one bin for each total, n to 6n (up to 31 bins for six dice).
- Uniform and Triangular: 20 equal bins from a to b.
- Exponential: 20 equal bins over 0 to 5 means; the last bin holds everything from its low edge up, so its high end reads blank.
- Discrete: one bin for each of the six outcomes.
Calculation sequence
The model works through the relationships above in this order:
- It checks the entered values in the order of the status table under Errors and warnings.
- It fixes the values used: your inputs, or in a challenge the challenge's own values.
- It generates the 3,000 draws of the seeded stream, six a trial.
- It turns each trial's draws into its outcome and event, and keeps the running sums.
- It computes the estimate, its interval, the event share and Wilson's interval, and the running estimate every 10 trials.
- It computes the exact values, and the histogram counts against the exact expected counts.
- It sets the status and, in a challenge, scores the challenge.
Outputs and interpretation
- Estimate, exact value and error: the mean outcome over all trials (what it measures is stated by "What is estimated", for example 4 times the share of points inside the circle), the exact value it aims at, and the estimate minus the exact value, each to 4 decimals.
- Confidence interval: its low and high ends, the standard error, the critical value z, and "Exact value inside the interval", Yes or No. At 95%, No happens by chance in about one run in twenty.
- Event: what the event is, how many trials had it, the share of trials with it, its exact probability, and the low and high ends of Wilson's interval for it.
- Values used: the experiment, seed, trial count, coins or dice a trial, heads probability, distribution, event threshold and confidence level of the run. In a challenge these are the challenge's own values. A value that does not apply reads blank (for example the coins or dice a trial for π).
- Trials table: one row per trial with x and y (the point for π, the first two coins or dice, or the draw u₁ and the sampled value for a distribution), the outcome, and the trial's histogram bin. Rows past the trial count are blank. The canvas draws the points, coins, dice and stacks from these rows.
- First 10 trials, coin by coin or die by die: each coin (1 heads, 0 tails) or die face; blank for π and Distribution.
- Running estimate: every 10 trials, the estimate, its band low and high, and an "Exact in band" flag (1 inside, 0 outside). The band narrows roughly as one over the square root of the trial count.
- Histogram: bin low and high, the count and the exact expected count of each bin; rows past the bin count are blank. For single-value bins (π, coins, dice, Discrete) the low and high are the same value.
- Discrete distribution used: each outcome's value, weight and probability; blank unless a Discrete distribution was sampled.
- Model version and formula module versions: the generator and the two interval methods the model embeds.
Challenges
Choosing a challenge number from 1 to 5 loads that challenge. The challenge fixes its own experiment, seed, trial count and settings, and leaves the visitor only the inputs it names; every other input takes the challenge's value or the shipped default. The result reads MET (score 100) or NOT MET (score 0), with the goal, the number the challenge is scored on, and a short explanation or hint.
- Predict a heads range. For 500 flips of a fair coin, enter the fewest and most heads of a range no wider than 50 that holds the number of heads at least 95% of the time. Open: the two predictions. Scored on the exact binomial probability of the range.
- A rare dice total. With two dice, find the smallest total that is reached or beaten less than 10% of the time. Open: the event threshold. Scored on the exact event probability, and the total one lower must not also qualify.
- An interval that misses. With 100 points estimating π at 95%, find a seed whose interval does not contain π. Open: the seed. Scored on whether π lies outside the interval.
- Enough spare parts. Each part works 80% of the time; choose the fewest parallel parts that keep the system working at least 99% of the time. Open: the coins (parts) a trial. Scored on the exact probability that at least one part works, and one part fewer must fall short.
- A fair game. A discrete game loses 1 on five equally likely outcomes and pays the sixth; set that payout so the expected value is 0 (within 0.01). Open: outcome 6's value. Scored on the exact mean.
Challenges 1, 2, 4 and 5 are scored on exact probabilities and means, not on the sampled results, so the seed does not decide them; challenge 3 is about the sampling itself.
Validation and status logic
The status reads the first match in the table under Errors and warnings. The NOT VALID checks apply to the entered values; the CHECK items apply to the values used, so a challenge's own settings decide them. A CHECK does not stop the calculation: every result is still shown.
Assumptions and limitations
- At most 500 trials, a multiple of 10, and six draws a trial (3,000 draws, all computed whatever the trial count). The trials table has 500 rows, the running estimate 50 rows, the histogram 31 rows, the coin-by-coin table the first 10 trials, and the discrete table 6 rows.
- The generator is deterministic and reproducible: the same seed gives the same trials in every experiment. It is a small linear congruential generator; consecutive draws fall on a lattice and nearby seeds can give related streams. Seed 2,147,483,646 is the mirror of seed 1 (every draw is 1 − u), so the two give identical π results. It is fine for teaching and wrong for cryptography or serious simulation.
- The interval for the estimate is a normal approximation, rough below 30 trials; the event interval is Wilson's and never sets the status. With 500 trials a 95% interval for π is about 0.3 wide.
- Coins are independent with the same p; dice are fair; draws are with replacement.
- A comparison that sits exactly on its threshold (a point on the circle, a draw equal to p, an outcome equal to the threshold, a value on a bin edge) can go either way; dice faces never sit on a boundary.
Restrictions and non-computing states
Every input is required and is checked on its entered value even when the chosen experiment does not use it.
Under NOT VALID every result and table is blank, except the model and module versions, the challenge texts, the challenge result (NOT MET when a challenge is selected) and a score of 0.
Errors and warnings
The status messages, first match wins:
| Status | Trigger |
|---|---|
| NOT VALID: choose an experiment from the list | The experiment is not one of the four. |
| NOT VALID: the seed must be a whole number from 1 to 2147483646 | The seed is blank, not whole or out of range. |
| NOT VALID: the number of trials must be a multiple of 10 from 10 to 500 | The trial count is out of range or not a multiple of 10. |
| NOT VALID: coins and dice a trial must be whole numbers from 1 to 6, and the heads probability from 0 to 1 | A coin, dice or heads-probability input is out of range. |
| NOT VALID: choose a distribution from the list | The distribution is not one of the four. |
| NOT VALID: the distribution needs a low end below the high end (both from -1000 to 1000), a mode between them, and a mean from 0.01 to 1000 | The low end is not below the high end, the mode lies outside them, or a value or the mean is out of range. |
| NOT VALID: each outcome value must be from -1000 to 1000 and each weight from 0 to 1000, with weights adding to more than 0 | A discrete value or weight is out of range, or every weight is 0. |
| NOT VALID: the event threshold must be from -1000 to 1000 and the confidence level from 0.8 to 0.99 | The threshold or the confidence level is out of range. |
| NOT VALID: the challenge number must be a whole number from 0 to 5, and each prediction from -1000 to 1000 | The challenge number or a prediction is out of range. |
| CHECK: fewer than 30 trials; the interval rests on a normal approximation that is rough this early | Fewer than 30 trials. |
| CHECK: every trial gave the same outcome, so the interval has no width | The sample variance is 0, for example p = 0 or p = 1. |
| CHECK: the exact value lies outside the confidence interval, as happens by chance in a share of runs equal to one minus the confidence level | The exact mean lies outside the estimate's interval. |
| OK | The run is computed with nothing to flag. |
References
This laboratory is an educational model of Monte Carlo estimation. Its random numbers come from the Park-Miller minimal standard (MINSTD) generator, a multiplicative linear congruential generator with modulus 2^31 - 1 and the multiplier 48271 recommended by Park, Miller and Stockmeyer in 1993, so every run is exact and repeatable from its seed. The experiments estimate pi from points in a square, and model coins, dice and a chosen distribution, the last by the inverse-transform method. The estimate after n trials is the mean outcome, with a confidence interval of the mean plus or minus z times the sample standard deviation over the square root of n, a normal approximation. The share of trials with an event gets Wilson's score interval. Exact values are shown alongside for comparison.
Sources and further reading
- The minimal standard generator: S. K. Park and K. W. Miller, "Random number generators: good ones are hard to find", Communications of the ACM 31(10), 1988, pages 1192-1201. doi:10.1145/63039.63042
- The multiplier 48271, proposed in the authors' reply in the journal's technical correspondence: S. K. Park, K. W. Miller and P. K. Stockmeyer, "Technical correspondence", Communications of the ACM 36(7), 1993. doi:10.1145/159544.376068
- Linear congruential generators and their properties: D. E. Knuth, The Art of Computer Programming, volume 2, Seminumerical Algorithms, 3rd edition, Addison-Wesley, 1997, section 3.2.1, The Linear Congruential Method. Author's page
- The inverse-transform (inversion) method: L. Devroye, Non-Uniform Random Variate Generation, Springer-Verlag, 1986, chapter II, General Principles in Random Variate Generation, section 2, The Inversion Method. Author's free edition
- The confidence interval for a mean: NIST/SEMATECH, e-Handbook of Statistical Methods, section 1.3.5.2, Confidence Limits for the Mean. NIST
- The score interval for a proportion: E. B. Wilson, "Probable inference, the law of succession, and statistical inference", Journal of the American Statistical Association 22(158), 1927, pages 209-212. doi:10.1080/01621459.1927.10502953
No table or text from any of these publications is reproduced, and none of their authors or publishers has reviewed or endorses this laboratory.
This laboratory is for learning. Its results hold only within its stated ranges; the generator is meant for repeatable classroom experiments, not for cryptography or production simulation.
Found a problem, or have an idea?
Tell us if a result looks wrong, a label is unclear, or something is missing. We read every message.