“Chance favours the prepared mind.” [Louis Pasteur]

Abstract

It is fairly easy to create a loaded die, let us say on average the 6 should appear twice as often as all the other numbers 1 thru 5: Enter into A1: =MIN(INT(RAND()*7+1),6)

But what if you want to create 7 rolls of this die and all numbers between 1 and 5 should appear exactly once and 6 exactly twice?

Here is my general solution:

sbExactRandHistogrm

The Excel / VBA Function sbExactRandHistogrm

Name

sbExactRandHistogrm – Create an exact double histogram distribution.

Synopsis

sbExactRandHistogrm(ldraw, dmin, dmax, vWeight)

Description

sbExactRandHistogrm creates an exact histogram distribution for ldraw draws of floating point numbers with double precision within range dmin:dmax. This range is divided into vWeight.count classes. Each class has weight vWeight(i), reflecting the probability of occurrence of a value within the class. If weights can’t be achieved exactly for ldraw draws the largest remainder method will be applied to minimize the absolute error. This function calls (needs) RoundToSum.

Parameters

ldraw – Number of draws

dmin – Minimum = lower boundary of range of numbers to draw

dmax – Maximum = upper boundary of range of numbers to draw

vWeight – Array of weights. Array size determines the number of different classes the range dmin : dmax is divided into. Values in this array specify likelihood of this class' numbers to appear (be drawn).

See Also

RoundToSum

Appendix – sbRandHistogrm Code

Please note: this function refers to (needs) RoundToSum.

Please read my Disclaimer.

Function sbExactRandHistogrm(ldraw As Long, _
            dmin As Double, _
            dmax As Double, _
            vWeight As Variant) As Variant
'Creates an exact histogram distribution for ldraw draws within range dmin:dmax.
'This range is divided into vWeight.count classes. Each class has weight vWeight(i)
'reflecting the probability of occurrence of a value within the class.
'If weights can't be achieved exactly for ldraw draws the largest remainder method will
'be applied to minimize the absolute error. This function calls (needs) sbRoundToSum.
'Source (EN): http://www.sulprobil.de/sbexactrandhistogrm_en/
'Source (DE): http://www.berndplumhoff.de/sbexactrandhistogrm_de/
'(C) (P) by Bernd Plumhoff 01-May-2021 PB V0.9

Dim i As Long, j As Long, n As Long
Dim vW As Variant
Dim dSumWeight As Double, dR As Double

Randomize
With Application.WorksheetFunction
vW = .Transpose(vWeight)
On Error GoTo Errhdl
i = vW(1) 'Throw error in case of horizontal array
On Error GoTo 0

n = UBound(vW)
ReDim dWeight(1 To n) As Double
ReDim dSumWeightI(0 To n) As Double
ReDim vR(1 To ldraw) As Variant

For i = 1 To n
    If vW(i) < 0# Then 'A negative weight is an error
        sbExactRandHistogrm = CVErr(xlErrValue)
        Exit Function
    End If
    'Calculate sum of all weights
    dSumWeight = dSumWeight + vW(i)
Next i

If dSumWeight = 0# Then
    'Sum of weights has to be greater zero
    sbExactRandHistogrm = CVErr(xlErrValue)
    Exit Function
End If

For i = 1 To n
    'Align weights to number of draws
    dWeight(i) = CDbl(ldraw) * vW(i) / dSumWeight
Next i

vW = RoundToSum(dWeight, 0)
On Error GoTo Errhdl
i = vW(1) 'Throw error in case of horizontal array
On Error GoTo 0

For j = 1 To ldraw

    dSumWeight = 0#
    dSumWeightI(0) = 0#
    For i = 1 To n
        'Calculate sum of all weights
        dSumWeight = dSumWeight + vW(i)
        'Calculate sum of weights till i
        dSumWeightI(i) = dSumWeight
    Next i
    
    dR = dSumWeight * Rnd
    
    i = n
    Do While dR < dSumWeightI(i)
        i = i - 1
    Loop
    
    vR(j) = dmin + (dmax - dmin) * (CDbl(i) + _
            (dR - dSumWeightI(i)) / vW(i + 1)) / CDbl(n)
    vW(i + 1) = vW(i + 1) - 1#
    
Next j

sbExactRandHistogrm = vR

Exit Function

Errhdl:
'Transpose variants to be able to address
'them with vW(i), not vW(i,1)
vW = .Transpose(vW)
Resume Next
End With

End Function

Download

Please read my Disclaimer.

sbExactRandHistogrm.xlsm [32 KB Excel file, open and use at your own risk]