RANDPOISSON Function (Gnumeric)

• Introduction: Gnumeric 0.49 NEWS explicitly lists this new function. This release-history evidence is separate from a native test of that old version.
• Source support: Registered in the Gnumeric 1.12.62 random plugin. No retirement is established; the last functioning native application version is unverified.
• Evidence dates: Official Gnumeric source retained October 13–14, 2026 was reviewed October 15. Mathematical controls were independently recalculated October 15. No native application, compiled Gnumeric helper or random generator was executed.

The required lambda must be nonnegative. The worksheet callback and helper reject a negative value with #NUM!. The helper returns 0 immediately for lambda 0; otherwise it passes a finite-resolution unit driver to the Poisson quantile routine.

The ideal distribution has mean lambda and variance lambda. Lambda is not a maximum count, a probability, or necessarily an integer. A valid lambda 0.5 does not instruct the function to round the rate to zero.

Create PoissonControls. Put expected interval count 2 in B2. In D2 enter =RANDPOISSON($B$2). It returns a changing nonnegative integer count, not the fixed number 2.

In F2 enter =B2, in F3 =B2, and in F4 =SQRT(B2). The ideal mean and variance are both 2, and standard deviation is approximately 1.41421356237. These are population summaries rather than exact results of the next few worksheet draws.

Put counts 0 through 5 in A7:A12. In B7 enter =EXP(-$B$2)*$B$2^A7/FACT(A7) and fill down. This is the ideal probability of each fixed count, not another random call:

The probabilities for 0 and 1 sum to approximately 0.40600584971. Thus the probability of at least 2 is approximately 0.59399415029. The six displayed rows do not exhaust the distribution; counts above 5 remain possible.

The source uses a quantile inversion helper rather than evaluating this PMF to produce each draw. The table validates the intended count model independently, without claiming a specific sequence of random outputs.

If an illustrative process has mean 6 events per hour, a 15-minute interval corresponds to lambda =6*15/60, or 1.5. Use =RANDPOISSON(1.5) for that interval, rather than retaining 6 and labeling the result a 15-minute count.

The ideal probability of zero events then is =EXP(-1.5), approximately 0.223130160148. Scaling the interval scales lambda under the model’s constant-rate assumption; it does not independently establish that a real event process is Poisson.

Independent ideal Poisson variables with means 2 and 3 sum to a Poisson variable with mean 5 and variance 5. If a worksheet instead doubles one lambda 2 random result, the mean becomes 4 and variance 8; the resulting counts are all even, unlike a Poisson distribution with mean 4.

This distinction is visible without sampling. A repeated cell reference reuses one draw. Use separate variables and state the independence assumption when applying the Poisson-sum relationship.

For lambda 0.5, the ideal zero-count probability is =EXP(-0.5), approximately 0.606530659713, while the output still counts whole events. A fractional expected value is compatible with integer-valued observations.

=RANDPOISSON(0) returns 0 from an early helper branch before a unit driver is requested. =RANDPOISSON(-1) gives #NUM!. The zero-lambda case is a degenerate distribution, not an instruction to use the previous nonzero parameter.

The ideal positive-lambda distribution has no finite maximum count. A count greater than lambda is ordinary behavior, and no single draw is required to equal the mean. Conversely, a finite generator and floating-point quantile search do not certify arbitrary mathematical tail reach.

The function generates a count; it does not calculate a Poisson probability or perform a goodness-of-fit test. Keep the random D2 cell separate from the fixed probability table when explaining or auditing a workbook.

The plugin marks this function volatile. Recalculation can replace a result, and the worksheet signature has no seed argument. Exact GUI triggers and saved-file replay were not tested. Separate calls do not promise matched internal drivers, while repeated references to one random cell do not constitute independent observations.

The controls here are independent mathematical or source-branch expectations, not generated samples. Ideal probability distributions are distinct from the finite-resolution implementation. Rounding, underflow, overflow and non-finite intermediates can affect extreme inputs; non-finite numeric results become #NUM!. No native application, compiled Gnumeric helper, random generator, empirical distribution-quality test or file-transfer test was executed. Use explicit finite numeric inputs; text and blank coercion are not certified by these controls.

• Function help and worksheet callback
• Pinned transformation helper, finite-resolution unit generator, and volatile registrations
• Finite-number result and error conversion
• Poisson quantile helper and discrete bounded inversion
• Gnumeric official function catalogue

RANDGEOM, RANDHYPERG.