Home / Monte Carlo / Model Building / Mark Input / Options

Mark Input Options

Every control in the Mark Input dialog. The controls sit on two tabs, with the Inputs range band above them naming the cells being marked, and the density preview beside them.

How every value box works

Every value box in the dialog takes either a number you type or a cell you point at.

Type a number and the value is fixed in the marking.

Point at a cell and the marking stores a link, so the model reads that cell on every run. There are two ways to point at one, and they do the same thing:

• Click the cell in Excel. The click fills whichever box is currently armed. A box arms itself when the keyboard is in it, so clicking into a box and then clicking a cell links that box.
• Press the box's own ▼ button, tooltip Collapse the dialog and pick a cell in Excel. That arms the box and shrinks the dialog to a one-line strip so the worksheet is not covered. The strip shows the field's name and its current value, and it stays above the other windows while it is showing. Pick a cell and the dialog comes back to full size with the keyboard back in the box you picked for. The ▲ button on the strip brings the dialog back without picking anything.

A chain icon appears to the left of a box that holds a link, and the box text states the link as well, so the icon is never the only sign of it. Typing over a linked box drops the link.

You can also type a cell reference or a defined name instead of clicking, and it becomes a link when you leave the box.

A cell in another workbook cannot be linked. A marking's link is written inside the workbook that holds the marking, so it has no way to name a cell somewhere else. Picking one leaves every field as it was and says so:

That cell is in Other.xlsx, so nothing was linked. A marking can only use cells in Model.xlsx.

A right-click in Excel picks nothing.

Reference mode

When a link is made while more than one cell is being marked, a box appears at the right of the row to say how that link should shift from one marked cell to the next. It offers all four of Excel's own reference forms:

Mode	Written	What it does across the marked cells
Relative	A1	shifts with each cell
Absolute	\(A\)1	never shifts
Row fixed	A$1	pins the row, the column shifts
Column fixed	$A1	pins the column, the row shifts

The box is shown only when the field holds a link and only when more than one cell is being marked, because a single cell has no offset for the mode to describe.

The choice is used to work out which cell each marked cell links to, and then it is finished with: what is stored is the resolved cell for each one, not the rule that produced it.

Messages and the status line

Two places report back to you, and they do different jobs.

A message under a box belongs to that box. It appears when you leave the box or when you press OK, and then stays live while you correct it. An error blocks OK. A caution does not: it tells you something is worth a look and lets the marking be written anyway.

The status line above the buttons reports what just happened rather than what is wrong: which box is armed for a pick, and confirmation when a link is made. It is one line, the full width of the dialog, with room for two lines reserved so the buttons below it do not move when a long message appears.

Inputs range

A read-only band at the top of the dialog naming the cells being marked. It sits above the tabs and is visible on both of them. There is nothing to type into it.

The Distribution tab

Name and distribution

Input name. What this input is called in reports. Type a name, or point at a cell holding one.

Leave the name empty to let the report use the cell's own name or address.

An empty name is not stored as the address it happens to have today: the report reads the cell's current address every time it runs, so the name does not go stale when rows are inserted above the cell or the cell is dragged.

The name box takes a link only from a sheet-qualified reference such as Data!B7, not from a bare B7. An input called T1, R2 or SO2 is a name someone meant to type, and a bare address is treated as one.

A typed name is held to two rules:

The input name cannot contain line breaks or control characters.

Keep the input name to 40 characters or fewer.

Distribution. The list of distributions this input can be drawn from.

The Mark Input dialog, the Distribution Gallery and the design sheet's in-cell dropdown are all shown the same list, so none of them can offer a distribution the others do not.

The list is built in three parts, in this order: the built-in distributions, then this workbook's own custom distributions, then its Empirical variables.

Fourteen distributions are built in. The parameters are listed in the order the dialog shows them.

Distribution	Parameters, in the order shown
Normal	Mean, Standard deviation
Lognormal	Mean (log), Standard deviation (log)
Exponential	Offset, Lambda
Uniform	Lower, Upper
Triangular	Minimum, Mode, Maximum
Weibull	Offset, Beta (shape), Alpha (scale)
Gamma	Offset, Shape, Scale
Logistic	Location, Scale
Log-Logistic	Location, Scale
Uniform-Discrete	Lower, Upper
Poisson	Offset, Mean
Binomial	Offset, Number of trials, Probability of success
Binary (Bernoulli)	Offset, Probability of 1
Constant	Value

Normal opens pre-filled with a mean of 0 and a standard deviation of 1, so a cell that has not been marked before shows a live preview straight away. It is the only one that starts with values in its boxes.

Two kinds of distribution are defined in the workbook rather than chosen from the list, and both are added to the end of it.

A custom distribution is one you build yourself. It appears under the name you gave it.

An Empirical variable is a column of your own sample data. It appears as its name followed by (Empirical).

Neither has parameter boxes, because neither keeps its data in the marking: the data stays in the workbook and the model reads it from there. Selecting one and marking writes only its name.

Distribution cell. Shown only when Linked to a cell is chosen in the list.

The last entry in the Mark Input dialog's list, Linked to a cell, means the distribution is whatever a worksheet cell says it is. Choosing it reveals one more box naming the cell that holds the distribution's name.

The model reads the distribution from this cell every time it runs, so changing the cell changes the distribution with no need to mark the input again.

A cell that does not name a distribution is a caution rather than an error. The marking is still written, and the model is what refuses.

This entry is offered by the Mark Input dialog alone. The design sheet's in-cell dropdown and the Distribution Gallery do not show it.

Parameters

One box for each parameter the selected distribution takes, labelled and ordered as the table above lists them. Each box takes a typed number or a linked cell.

Parameter values stay in their boxes when you change the distribution, slot by slot. Switching from Normal to Lognormal keeps the two numbers that were there, and switching back keeps any edit you made in between. Only opening the dialog on an unmarked cell seeds the standard normal.

Each parameter is checked as you enter it, against the same limits the simulation engine applies, so the dialog refuses exactly what a run would refuse.

Distribution	What is required
Normal	Standard deviation is 0 or greater
Lognormal	Standard deviation (log) is greater than 0
Exponential	Lambda is greater than 0
Uniform	Upper is greater than or equal to Lower
Triangular	Mode is greater than or equal to Minimum, and Maximum is greater than or equal to Mode
Weibull	Beta (shape) and Alpha (scale) are both greater than 0
Gamma	Shape and Scale are both greater than 0
Logistic	Scale is greater than 0
Log-Logistic	Scale is greater than 0
Uniform-Discrete	Lower and Upper are whole numbers, and Upper is greater than or equal to Lower
Poisson	Offset is a whole number, and Mean is greater than 0
Binomial	Offset is a whole number; Number of trials is a whole number greater than 0; Probability of success is between 0 and 1 inclusive
Binary (Bernoulli)	Offset is a whole number, and Probability of 1 is greater than 0 and less than 1
Constant	no restriction on Value

A parameter that has to be a whole number is also held to Excel's own whole-number range, and the message says which of the two rules was broken:

Number of trials must be a whole number.

Number of trials must be between -2147483648 and 2147483647.

A parameter given as a link is checked when the model runs rather than as you type, because its value is not known until then.

While the distribution is itself linked to a cell, the boxes are labelled with the slot number after the name, as Mean (Param 1), because which parameter is which depends on what the cell currently says.

Truncation

Truncate this distribution. A checkbox. Clearing it hides the controls below and keeps whatever you had entered, so turning it back on brings your bounds back.

Truncation removes part of the distribution and keeps the rest. The preview redraws to show what is kept.

Exclude below. The left-hand cut: values below this are excluded. An empty box means no lower limit, and shows a gray -∞ to say so. There is no separate clear button, because deleting the text clears the bound.

Exclude above. The right-hand cut, the same way, with a gray +∞ while empty.

The two are a pair, and crossing them is refused:

Exclude below must be less than exclude above for "Inlet pressure".

Interior exclusion bands (From / To). Zero or more ranges cut out of the middle, each a pair of boxes. + Add band adds one and puts the keyboard in its From box. Overlapping bands are combined, and a band whose From is greater than its To is read the right way round.

Kept support. A line under the bands reporting what is left, as a list of intervals:

Kept support: [2, 5] and [8, 12]

It reads Kept support: none when the bounds you gave leave nothing at all.

A Constant cannot be truncated. Its checkbox is turned off and the reason is shown beside it:

A Constant has a single value, so there is no range to keep. Truncation is not available for it.

Optimization

Whether the Optimizer is allowed to move this input. One choice of three:

Choice	What it means
Yes, this input can be optimized	the Optimizer may change it
No	it may not
Read it from a cell	a worksheet cell decides, and is read on every run

Choosing the third puts the keyboard in the cell box below it and arms it for a pick. The box is enabled only while that answer is chosen, so there is never an empty box on screen for an answer you did not pick.

What the cell has to hold is stated in the dialog:

FALSE, 0, or an empty cell means the input cannot be optimized. Anything else means it can.

The Advanced tab

Both sections on this tab are recorded with the marking and neither changes what is sampled.

Valid range

Where the model is valid. Does not affect sampling.

Valid minimum (low) and Valid maximum (high). Both optional, both take a typed number or a linked cell. The low must be below the high, and OK is blocked while it is not.

Default range for plotting

Optional axis bounds for plots. May extend beyond the valid range.

Plot low and Plot high. Optional axis bounds. The low must be below the high.

Nominal (plot hold value).

Nominal (plot hold value): the value used for this factor when it is not on a plot axis. It may be left blank.

The buttons

Button	What it does
Browse Distribution Gallery	opens the gallery and returns your choice to the distribution list
Fit Distribution from Data	advertises a future version. Pressing it says so in the status line and changes nothing
Help	opens the reference page for the distribution selected at that moment, at its own section
<< Previous Input	saves this input and shows the previous marked input
Next Input >>	saves this input and shows the next marked input
Cancel	closes without writing anything
OK	writes the marking

The two browsing buttons stay enabled even when there is nowhere to go, so they can always take keyboard focus. Pressing one with nowhere to go says why in the status line.