# 8.1. Expressions¶

Scientific simulations often keep track of several dozen variables as they
run. However, only a small subset of those variables are usually written
to a simulation database to save disk space. Sometimes variables can be
derived from other variables using an *expression*. VisIt provides
expressions to allow scientists to create derived variables using
variables that are stored in the database. Expressions are extremely powerful
because they allow users to analyze new data without necessarily having to
rerun a simulation. Variables created using expressions behave just like
variables stored in a database; they appear in menus where database variables
appear and can be visualized like any other database variable.

## 8.1.1. Expression Window¶

VisIt provides an **Expression Window**, shown in
Figure 8.1, that allows users to create new
variables that can be used in visualizations. Users can open the
**Expression Window** by clicking on the **Expressions** option in the
**Main Window’s Controls** menu. The **Expression Window** is divided
vertically into two main areas with the **Expression list** on the left
and the **Definition** area on the right.
The **Expression list** contains the list of expressions. The **Definition**
area displays the definition of the expression that is highlighted in
the **Expression list** and provides controls to edit the expression
definition.

Expressions in VisIt are created either manually by the user or automatically by various means including…

Preferences

Mesh quality expressions

Time derivative expressions

Vector magnitude expressions

GUI wizards

Operators

Databases

By default, the **Expression list** will display only those expressions
created manually by the user. A check box near the bottom of the
**Expression list** controls the display of automatically created
expressions. When this box is checked, the **Expression list** will also
include expressions created automatically by **Preferences** and **Databases**
but not expressions created automatically by **GUI wizards** or **Operators**.

### 8.1.1.1. Creating a new expression¶

Users can create a new expression by clicking on the **Expression Window’s New**
button. When the user clicks on the **New** button, VisIt adds a new expression
and shows its new, empty definition in the **Definitions** area. The initial
name for a new expression is *“unnamed”* followed by some integer suffix. After
the user types a new name for the expression into the **Name** text field,
the expression’s name in the **Expression list** will update. If the user types
a name that already exists in the expression list, then Visit will
automatically append a number to the end of the name to avoid duplicate
expression names.

Each expression also has a **Type** that specifies the type of variable
the expression produces. The available types are:

Scalar

Vector

Tensor

Symmetric Tensor

Array

Curve

Users must be sure to select the appropriate type for any expression they create. The selected type determines the menu in which the variable appears and subsequently the plots that can operate on the variable.

To edit an expression’s definition, users can type a new expression
comprised of constants, variable names, and even other VisIt expressions into
the **Definition** text field. The expression definition can span multiple
lines as the VisIt expression parser ignores whitespace. For a complete
list of VisIt’s built-in expressions, refer to the table in section
Built-in expressions. Users can also use the **Insert Function…**
menu, shown in Figure 8.2, to
insert any of VisIt’s built-in expressions directly into the expression
definition. The list of built-in expressions is divided into certain
categories as shown by the structure of the **Insert Function…**
menu.

In the example shown in Figure 8.2,
the **Insert Function…** operation inserted a sort of *template* for the
function giving some indication of the argument(s) to the function and their
meanings. Users can then simply edit those parts of the function template that
need to be specified.

In addition to the **Insert Function…** menu, which lets users insert built-in
functions into the expression definition, VisIt’s **Expression Window**
provides an **Insert Variable…** menu that allows users to insert variables
from the active database into the expression definition. The
**Insert Variable…** menu, shown in
Figure 8.3, is broken up into Scalars,
Vectors, Meshes, etc. and has the available variables under the appropriate
heading so they are easy to find.

Some variables can only be expressed as very complex expressions containing
several intermediate subexpressions that are only used to simplify the
overall expression definition. These types of subexpressions are seldom
visualized on their own. If users want to prevent them from being added to
the **Plot** menu, turn off the **Show variable in plot menus** check box.

### 8.1.1.2. Deleting an expression¶

Users can delete an expression by clicking on it in the **Expression list**
and then clicking on the **Delete** button. Deleting an expression removes
it from the list of defined expressions and will cause unresolved references
for any other expressions that use the deleted expression. If a plot uses
an expression with unresolved references, VisIt will not be able to generate
it until the user resolves the reference.

## 8.1.2. Expression grammar¶

VisIt allows expressions to be written using a host of unary and binary math operators as well as built-in and user-defined functions. VisIt’s expressions follow C-language syntax, although there are a few differences. The following paragraphs detail the syntax of VisIt expressions.

### 8.1.2.1. Math operators¶

These include use of +, -, *, /, ^ as addition, subtraction, multiplication, division, and exponentiation as infix operators, as well as the unary minus, in their normal precedence and associativity. Parentheses may be used as well to force a desired associativity.

*Examples: a+b^-c (a+b)*c*

### 8.1.2.2. Constants¶

Scalar constants include floating point numbers and integers, as well as booleans (true, false, on, off) and strings.

*Examples: 3e4 10 “mauve” true false*

### 8.1.2.3. Vectors¶

Expressions can be grouped into two or three dimensional vector variables using curly braces.

*Examples: {xc, yc} {0,0,1}*

### 8.1.2.4. Lists¶

Lists are used to specify multiple items or ranges, using colons to create ranges of integers, possibly with strides, or using comma-separated lists of integers, integer ranges, floating points numbers, or strings.

*Examples: [1,3,2] [1:2, 10:20:5, 22] [silver, gold] [1.1, 2.5, 3.9] [level1, level2]*

### 8.1.2.5. Identifiers¶

Identifiers include function names, defined variable and function names, and file variable names. They may include alphabetic characters, numeric characters, and underscores in any order. Identifiers should have at least one non-numeric character so that they are not confused with integers, and they should not look identical to floating point numbers such as 1e6.

*Examples: density x y z 3d_mesh*

### 8.1.2.6. Functions¶

These are used for built in functions, but they may also be used for functions/macros defined by the user. They take specific types and numbers of arguments within the parentheses, separated by commas. Some functions may accept named arguments in the form identifier=value.

*Examples: sin(pi / 2) cross(vec1, {0,0,1}) my_xform(mesh1) subselect(materials=[a,b])*

### 8.1.2.7. Database variables¶

These are like identifiers, but may also include periods, plus, and minus characters. A normal identifier will map to a file variable when it is not defined as another expression. To force variables that look like integers or floating point numbers to be interpreted as variable names, or to force variable names which are defined by another expression to map to a variable in a file, they should be enclosed with < and >, the left and right carats/angle brackets. Note that quotation marks will cause them to be interpreted as string constants, not variable names. In addition, variables in files may be in directories within a file, so they may include slashes in a path when in angle brackets.

*Examples: density <pressure> <a.001> <a.002> <domain1/density>*

### 8.1.2.8. Databases¶

A database specification looks similar to a database variable contained in angle brackets, but it is followed by a colon before the closing angle bracket, and it may also contain extra information. A database specification includes a file specification possibly followed a machine name, a time specification by itself, or a file/machine specification followed by a time specification. A file specification is just a file name with a path if needed. A machine specification is an at-sign @ followed by a host name. A time specification looks much like a list in that it contains integer numbers or ranges, or floating point numbers, separated by commas and enclosed in square brackets. However, it may also be followed by a letter c, t, or i to specify if the time specification refers to cycles, times, or indices, respectively. If no letter is specified, then the parser guesses that integers refer to cycles, floating point numbers refer to times. There is also an alternative to force indices which is the pound sign # after the opening square bracket.

*Examples: </dir/file:> <file@host.gov:> <[# 0:10]:> <file[1.234]:> <file[000, 023, 047]:> <file[10]c:>*

### 8.1.2.9. Qualified file variables¶

Just like variables may be in directories within a file, they may also be in other timesteps within the same database, within other databases, and even within databases on other machines. To specify where a variable is located, use the angle brackets again, and prefix the variable name with a database specification, using the colon after the database specification as a delimiter.

*Examples: <file:var> </dir/file:/domain/var> <file@192.168.1.1:/var> <[#0]:zerocyclevar>*

## 8.1.3. Built-in expressions¶

The following table lists built-in expressions that can be used to create more advanced expressions. Unless otherwise noted in the description, each expression takes scalar variables as its arguments.

### 8.1.3.1. Arithmetic Operator Expressions (Math Expressions)¶

In binary arithmetic operator expressions, each operand must evaluate to
the same type field. For example, both must evaluate to a
*scalar* field or both must evaluate to a *vector* field.

In addition, if the two expressions differ in centering (e.g. one is *zone*
or *cell* centered or *piecewise-constant* over mesh cells while the other is
*node* or *point* centered or *pieceiwse-linear* over mesh cells), VisIt will
*recenter* any node-centered fields to *zone* centering to compute the
expression. This may not always be desirable. When it is not, the
recenter() may be used to explicitly
control the centering of specific operands in an expression.

- Sum Operator (
`+`

)`exprL + exprR`

Creates a new expression which is the sum of the

`exprL`

and`exprR`

expressions.

- Difference Operator (
`-`

)`exprL - exprR`

Creates a new expression which is the difference of the

`exprL`

and`exprR`

expressions.

- Product Operator (
`*`

)`exprL * exprR`

Creates a new expression which is the product of the

`exprL`

and`exprR`

expressions.

- Division Operator (
`/`

)`exprL / exprR`

Creates a new expression which is the quotient after dividing the

`exprL`

expression by the`exprR`

expression.- Division Operator
`divide(val_numerator, val_denominator, [div_by_zero_value, tolerance])`

Creates a new expression which is the quotient after dividing the

`val_numerator`

expression by the`val_denominator`

expression. The`div_by_zero_value`

is used wherever the`val_denominator`

is within`tolerance`

of zero.

- Exponent Operator (
`^`

)`exprL ^ exprR`

Creates a new expression which is the product after multiplying the

`exprL`

expression by itself`exprR`

times.

- Logical AND Operator (
`&`

)`exprL & exprR`

Creates a new expression which is the logical

*AND*of the`exprL`

and`exprR`

expressions treating each value as a binary bit field. It is probably most useful for expressions involving integer data but can be applied to expressions involving any type.

- Associative Operator (
`()`

)`( expr0 OP expr1 )`

Parenthesis,

*()*are used to explicitly group partial results of sub expressions and control evaluation order.For example, the expression

`(a + b) / c`

first computes the sum,`a+b`

and then divides by`c`

.

- Absolute Value Function (
`abs()`

)`abs(expr0)`

Creates a new expression which is everywhere the absolute value if its argument.

- Ceiling Function (
`ceil()`

)`ceil(expr0)`

Creates a new expression which is everywhere the

*ceiling*(smallest integer greater than or equal to) of its argument.

- Exponent Function (
`exp()`

)`exp(expr0)`

Creates a new expression which is everywhere

*e*(base of the natural logarithm) raised to the power of its argument.

- Floor Function (
`floor()`

)`floor(expr0)`

Creates a new expression which is everywhere the

*floor*(greatest integer less than or equal to) of its argument.

- Natural Logarithm Function (
`ln()`

)`ln(expr0)`

Creates a new expression which is everywhere the natural logarithm of its argument.

- Base 10 Logarithm Function (
`log10()`

)`log10(expr0)`

Creates a new expression which is everywhere the base 10 logarithm of its argument.

- Max Function (
`max()`

)`max(expr0, exrp1 [, ...])`

Creates a new expression which is everywhere the maximum among all input variables.

- Min Function (
`min()`

)`min(expr0, exrp1 [, ...])`

Creates a new expression which is everywhere the minimum among all input variables.

- Modulo Function (
`mod()`

)`mod(expr0,exrp1)`

Creates a new expression which is everywhere the first argument,

`expr0`

, modulo the second argument,`expr1`

.

- Random Function (
`random()`

)`random(expr0)`

Creates a new expression which is everywhere a random floating point number between 0 and 1, as computed by \((\text{rand()} \% 1024) \div 1024\) where

`rand()`

is the standard C library rand() random number generator. The argument,`expr0`

, must be a mesh variable. The seed used on each block of the mesh is the absolute domain number.

- Round Function (
`round()`

)`round(expr0)`

Creates a new expression which is everywhere the result of rounding its argument.

- Square Function (
`sqr()`

)`sqr(expr0)`

Creates a new expression which is everywhere the result of squaring its argument.

- Square Root Function (
`sqrt()`

)`sqrt(expr0)`

Creates a new expression which is everywhere the square root of its argument.

### 8.1.3.2. Relational, Conditional and Logical Expressions¶

The `if()`

conditional expression is designed to be used in concert with
the **Relational** and **Logical** expressions. Together, these expressions
can be used to build up more complex expressions in which very different
evaluations are performed depending on the outcome of other evaluations.
For example, the `if()`

conditional expression can be used together with
one or more relational expressions to create a new expression which evaluates
to a dot-product on part of a mesh and to the magnitude of a divergence operator
on another part of a mesh. However, the **Relational** and **Logical** expressions
alone (e.g. when not used *within* an `if()`

expression) do not produce a
useful result.

- If Function (
`if()`

)`if(exprCondition, exprTrue, exprFalse)`

Creates a new expression which is equal to

`exprTrue`

wherever the condition,`exprCondition`

is non-zero and which is equal to`exprFalse`

wherever`exprCondition`

is zero.For example, the expression

`if(and(gt(pressure, 2.0), lt(pressure, 4.0)), pressure, 0.0)`

combines the`if`

expression with the`gt`

and`lt`

expressions to create a new expression that is equal to`pressure`

wherever it is between 2.0 and 4.0 and 0 otherwise.

- Equal Function (
`eq()`

)`eq(exprL,exprR)`

Creates a new expression which is everywhere a boolean value (1 or 0) indicating whether its two arguments are equal. A value of 1 is produced everywhere the arguments

*are*equal and 0 otherwise.

- Greater Than Function (
`gt()`

)`gt(exprL,exprR)`

Creates a new expression which is everywhere a boolean value (1 or 0) indicating whether

`exprL`

is greater than`exprR`

. A value of 1 is produced everywhere`exprL`

is greater than`exprR`

and 0 otherwise.

- Greater Than or Equal Function (
`ge()`

)`ge(exprL,exprR)`

Creates a new expression which is everywhere a boolean value (1 or 0) indicating whether

`exprL`

is greater than or equal to`exprR`

. A value of 1 is produced everywhere`exprL`

is greater than or equal to`exprR`

and 0 otherwise.

- Less Than Function (
`lt()`

)`lt(exprL,exprR)`

Creates a new expression which is everywhere a boolean value (1 or 0) indicating whether

`exprL`

is less than`exprR`

. A value of 1 is produced everywhere`exprL`

is less than`exprR`

and 0 otherwise.

- Less Than or Equal Function (
`le()`

)`le(exprL,exprR)`

Creates a new expression which is everywhere a boolean value (1 or 0) indicating whether

`exprL`

is less than or equal to`exprR`

. A value of 1 is produced everywhere`exprL`

is less than or equal to`exprR`

and 0 otherwise.

- Equal Function (
`ne()`

)`ne(exprL,exprR)`

Creates a new expression which is everywhere a boolean value (1 or 0) indicating whether its two arguments are

*not*equal. A value of 1 is produced everywhere the arguments are*not*equal and 0 otherwise.

- Logical And Function (
`and()`

)`and(exprL,exprR)`

Creates a new expression which is everywhere the logical

*and*of its two arguments. Non-zero values are treated as true whereas zero values are treated as false.

- Logical Or Function (
`or()`

)`or(exprL,exprR)`

Creates a new expression which is everywhere the logical

*or*of its two arguments. Non-zero values are treated as true whereas zero values are treated as false.

- Logical Not Function (
`not()`

)`not(expr0)`

Creates a new expression which is everywhere the logical

*not*of its argument. Non-zero values are treated as true whereas zero values are treated as false.

### 8.1.3.3. Trigonometric Expressions¶

- Arc Cosine Function (
`acos()`

)`acos(expr0)`

Creates a new expression which is everywhere the arc cosine of its argument. The returned value is in

*radians*.

- Arc Sine Function (
`asin()`

)`asin(expr0)`

Creates a new expression which is everywhere the arc sine of its argument. The returned value is in

*radians*.

- Arc Tangent Function (
`atan()`

)`atan(expr0)`

Creates a new expression which is everywhere the arc tangent of its argument. The returned value is in

*radians*.

- Arc Tangent 2 Function (
`atan2()`

)`atan2(expr0,expr1)`

Creates a new expression which is everywhere the arc tangent of its arguments. The returned value is in

*radians*.

- Cosine Function (
`cos()`

)`cos(expr0)`

Creates a new expression which is everywhere the cosine of its argument. The argument is treated as in units of

*radians*.

- Hyperbolic Cosine Function (
`cosh()`

)`cosh(expr0)`

Creates a new expression which is everywhere the hyperbolic cosine of its argument. The argument is the

*hyperbolic angle*.

- Sine Function (
`sin()`

)`sin(expr0)`

Creates a new expression which is everywhere the sine of its argument. The argument is treated as in units of

*radians*.

- Hyperbolic Sine Function (
`sinh()`

)`sinh(expr0)`

Creates a new expression which is everywhere the hyperbolic sine of its argument. The argument is the

*hyperbolic angle*.

- Tangent Function (
`tan()`

)`tan(expr0)`

Creates a new expression which is everywhere the tangent of its argument. The argument is treated as in units of

*radians*.

- Hyperbolic Tangent Function (
`tanh()`

)`tanh(expr0)`

Creates a new expression which is everywhere the hyperbolic tangent of its argument. The argument is the

*hyperbolic angle*.

- Degrees To Radians Conversion Function (
`deg2rad()`

)`deg2rad(expr0)`

Creates a new expression which is everywhere the conversion from degrees to radians of its argument. The argument should be a variable defined in units of degrees.

- Radians To Degrees Conversion Function (
`rad2deg()`

)`rad2deg(expr0)`

Creates a new expression which is everywhere the conversion from radians to degrees of its argument. The argument should be a variable defined in units of radians.

### 8.1.3.4. Vector and Color Expressions¶

- Vector Compose Operator (
`{}`

)`{expr0, expr1, ... , exprN-1}`

Curly braces,

`{}`

, are used to create a vector from scalars or a tensor from vectors. A common use is to compose 3 scalar expressions to form a vector expression as in`{a, b, c}`

. For 2D data, the 3rd component must still be provided and must be zero as in`{a,b,0}`

. The component expressions,`expr0`

,`expr1`

, etc. must all be the same type (e.g. scalar, vector) and must all be the same centering. Scalars compose into (row) vectors and (row) vectors compose into tensors, row-by-row.If constant values (e.g.

`1`

or`0`

) are needed in composing a vector expression, then use the expression functions designed to create constant expressions such as`nodal_constant(<mesh>,value)`

(for node-centered constant expressions) or`zonal_constant(<mesh>,value)`

(for zone-centered consntant expressions). Using the constant values themselves (e.g.`0`

or`1`

) directly in the compose operator does not always work as expected depending on VisIt’s ability to infer the intended*mesh*and/or*centering*.

- Vector Component Operator (
`[]`

)`expr[I]`

Square brackets,

*[]*, are used to create a new expression of lower tensor rank by extracting a component from an expression of higher tensor rank. Components are indexed starting from 0. If`expr`

is a tensor of rank 2, the result will be a tensor of rank 1 (e.g. a vector). If`expr`

is a tensor of rank 1, the result will be a tensor of rank 0 (e.g. a scalar). To obtain the`J`

-th component of the`I`

-th row of a tensor of rank 2, the expression would be`expr[I][J]`

- Color Function (
`color()`

)`color(exprR,exprG,exprB)`

Creates a new, RGB

*vector*, expression which defines a*color*vector where`exprR`

defines the*red*component,`exprG`

defines the*green*component and`exprB`

defines the*blue*component of the color vector. The resulting expression is suitable for plotting with the Truecolor Plot. The arguments are used to define color values in the range 0…255. Values outside that range are clamped. No normalization is performed. If the arguments have much smaller or larger range than [0…255], it may be appropriate to select a suitable multiplicative scale factor.

- Color4 Function (
`color4()`

)`color4(exprR,exprG,exprB,exprA)`

See color(). This function is similar to the

`color()`

function but also supports*alpha-transparency*as the fourth argument, again in the range 0…255.

- Color lookup Function (
`colorlookup()`

)`colorlookup(expr0,tabname,scalmode,skewfac)`

Creates a new

*vector*expression that is the color that each value in`expr0`

maps to. The`tabname`

argument is the name of the color table. The`expr0`

and`tabname`

arguments are*required*. The`scalmode`

and`skewfac`

arguments are optional. Possible values for`scalmode`

are`0`

(for*linear*scaling mode),`1`

(for*log*scaling mode) and`2`

(for*skew*scaling mode). The`skewfac`

argument is*required*only for a`scalmode`

of`2`

.

- Cross Product Function (
`cross()`

)`cross(exprV0,exprV1)`

Creates a new

*vector*expression which is the vector cross product created by crossing`exprV0`

*into*`exprv1`

. Both arguments must be*vector*expression.

- Dot Product Function (
`dot()`

)`dot(exprV0,exprV1)`

Creates a new

*scalar*expression which is the vector dot product of`exprV0`

with`exprV1`

.

- HSV Color Function (
`hsvcolor()`

)`hsvcolor(exprH,exprS,exprV)`

See color(). This function is similar to the

`color()`

function but takes*Hue*,*Saturation*and*Value*(Lightness) arguments as inputs and produces an RGB*vector*expression.

- Magnitude Function (
`magnitude()`

)`magnitude(exprV0)`

Creates a new

*scalar*expression which is everywhere the magnitude of the`exprV0`

.

- Normalize Function (
`normalize()`

)`normalize(exprV0)`

Creates a new

*vector*expression which is everywhere a normalized vector (e.g. same direction but unit magnitude) of`exprV0`

.

- Curl Function:
`curl()`

`curl(expr0)`

Creates a new

*vector*expression which is everywhere the curl of its input argument, which must be vector valued. In a 3D context, the result is also a vector. However, in a 2D context the result*vector*would always be`[0,0,V]`

so expression instead returns only the*scalar*V.

- Divergence Function:
`divergence()`

`divergence(expr0)`

Creates a new

*scalar*expression which is everywhere the divergence of its input argument, which must be vector valued.

- Gradient Function:
`gradient()`

`gradient(expr0)`

Creates a new

*vector*expression which is everywhere the gradient of its input argument, which must be*scalar*. The method of calculation varies depending on the type of mesh upon which the input is defined. See also ij_gradient() and ijk_gradient().

- IJ_Gradient Function:
`ij_gradient()`

`ij_gradient(expr0)`

No description available.

- IJK_Gradient Function:
`ijk_gradient()`

`ijk_gradient(expr0)`

No description available.

- Laplacian Function:
`laplacian()`

`laplacian(expr0)`

No description available.

- Surface Normal Function:
`surface_normal()`

`surface_normal(expr0)`

This function is an

*alias*for cell_surface_normal()

- Point Surface Normal Function:
`point_surface_normal()`

`point_surface_normal(expr0)`

Like cell_surface_normal() except that after computing face normals, they are averaged to the nodes.

- Cell Surface Normal Function:
`cell_surface_normal()`

`cell_surface_normal(<Mesh>)`

Computes a

*vector*variable which is the normal to a*surface*. The input argument is a*Mesh*variable. In addition, this function cannot be used in isolation. It must be used in combination the external surface,*first*, and the defer expression,*second*, operators.

- Edge Normal Function:
`edge_normal()`

`edge_normal(expr0)`

No description available.

- Point Edge Normal Function:
`point_edge_normal()`

`point_edge_normal(expr0)`

No description available.

- Cell Edge Normal Function:
`cell_edge_normal()`

`cell_edge_normal(expr0)`

No description available.

### 8.1.3.5. Tensor Expressions¶

Tensor expressions can be constructed either by direct *composition* (e.g. using the compose operator, `{}`

) or by using a tensor expression function.

When *composing* tensors with the compose operator, `{}`

, 9 scalar (e.g. `{{xx,xy,xz},{yx,yy,yz},{zx,zy,zz}}`

or 3 vector expressions (e.g. `{r1,r2,r3}`

are typically used.
However, tensors can be composed from combinations of these as well (e.g. `{r1,{a,b,c},r3}`

).
Note that in the preceding example expressions, the compose operator is used in a nested manner twice.
The inner instances compose sets of scalars into row vectors and the outer instance composes the row vectors into the final tensor.
Tensor expressions in 2D still require 9 scalar components but those in the 3rd row and column must be all zeros.
Symmetric tensor expressions also still require 9 scalar components but must also exhibit symmetry.

If constant values (e.g. `1`

or `0`

) are needed in composing a tensor expression, then use the expression functions designed to create constant expressions such as `nodal_constant(<mesh>,value)`

(for node-centered constant expressions) or `zonal_constant(<mesh>,value)`

(for zone-centered consntant expressions).
Using the constant values themselves (e.g. `0`

or `1`

) directly in the compose operator does not always work as expected depending on VisIt’s ability to infer the intended *mesh* and/or *centering*.

Often, using the tensor expression functions described here necessitates a detailed understanding of the actual numerical calculations VisIt uses in evaluating the expressions. Therefore, in many cases here, we provide collapsible sections that can be expanded to show the actual C++ source code VisIt is compiled with to compute a given tensor expression. The code displayed in these sections is derived from links to the actual source code files. So, the reader can be assured it is a faithful representation of the numerical operations VisIt is actually performing.

- Contraction Function:
`contraction()`

`contraction(expr0)`

Creates a

*scalar*expression which is everywhere the*contraction*of`expr0`

which must be a*tensor*valued expression. The contraction is the sum of pairwise dot-products of each of the column vectors of the tensor with itself as shown in the code snip-it below.

**Show/Hide Code for** `contraction()`

```
// Conceptually it is like as doting each column vector with
// itself and adding the column results
//
ctract +=vals[0] * vals[0] + vals[1] * vals[1] + vals[2] * vals[2];
ctract +=vals[3] * vals[3] + vals[4] * vals[4] + vals[5] * vals[5];
ctract +=vals[6] * vals[6] + vals[7] * vals[7] + vals[8] * vals[8];
```

- Determinant Function:
`determinant()`

`determinant(expr0)`

Creates a

*scalar*expression which is everywhere the determinant of`expr0`

which must be*tensor*valued.

- Effective Tensor Function:
`effective_tensor()`

`effective_tensor(expr0)`

Creates a

*scalar*expression which is everywhere the square root of three times the*second principal invariant of the stress deviator tenosr*, \(\sqrt{3*J_2}\), where \(J_2\) is the*second principal invariant of the stress deviator tensor*. This is also known as the*von Mises stress*or the*Huber-Mises stress*or the*Mises effective stress*.

**Show/Hide Code for** `effective_tensor()`

```
double s11 = vals[0], s12 = vals[1], s13 = vals[2];
double s21 = vals[3], s22 = vals[4], s23 = vals[5];
double s31 = vals[6], s32 = vals[7], s33 = vals[8];
// First invariant of the stress tensor
// aka "pressure" of incompressible fluid in motion
// aka "mean effective stress"
double trace = (s11 + s22 + s33) / 3.;
// components of the deviatoric stress
double dev0 = s11 - trace;
double dev1 = s22 - trace;
double dev2 = s33 - trace;
// The second invariant of the stress deviator
// aka "J2"
double out2 = 0.5*(dev0*dev0 + dev1*dev1 + dev2*dev2) +
s12*s12 + s13*s13 + s23*s23;
// stress deviator
out2 = sqrt(3.*out2);
```

- Eigenvalue Function:
`eigenvalue()`

`eigenvalue(expr0)`

The

`expr0`

argument must evaluate to a 3x3*symmetric*tensor. The eigenvalue expression returns the eigenvalues of the 3x3*symmetric*matrix argument as a vector valued expression where each eigenvalue is a component of the vector. Use the vector component operator, [], to access individual eigenvalues. If a non-symmetric tensor is supplied, results are indeterminate.

- Eigenvector Function:
`eigenvector()`

`eigenvector(expr0)`

The

`expr0`

argument must evaluate to a 3x3*symmetric*tensor. The eigenvector expression returns the eigenvectors of the 3x3 matrix argument as a tensor (3x3 matrix) valued expression where each column in the tensor is one of the eigenvectors.In order to use the vector component operator [], to access individual eigenvectors, the result must be

*transposed*with the transpose(), expression function.For example, if

`evecs = transpose(eigenvector(tensor))`

, the expression`evecs[1]`

will return the second eigenvector.

- Inverse Function:
`inverse()`

`inverse(expr0)`

Creates a new tensor expression which is everywhere the inverse of its input argument, which must also be a tensor.

- Principal Deviatoric Tensor Function:
`principal_deviatoric_tensor()`

`principal_deviatoric_tensor(expr0)`

Deviatoric stress is the stress tensor which results after subtracting the hydrostatic stress tensor. Hydrostatic stress is a

*scalar*quantity also often referred to as*average pressure*or just*pressure*. However, it is often characterized in*tensor*form by multiplying it through a 3x3 identity matrix.The

`principal_deviatoric_tensor()`

expression function creates a new*vector*expression which is everywhere the principal components of the deviatoric stress tensor computed from the*symmetric*tensor argument`expr0`

. In other words, the*eigenvalues*of the deviatoric stress tensor.Potentially, it would be more appropriate to create a new

*tensor*field here with all zeros for off-diagonal elements and the eigenvalues on the main diagonal.This expression can also be computed by using a combination of the

`trace()`

and`principal_tensor()`

expression functions. The`trace()`

(divided by 3) would be used to subtract out hydrostatic stress and the result could be used in the`principal_tensor()`

expression to arrive at the same result.

**Show/Hide Code for** `principal_deviatoric_tensor()`

```
double pressure = -(vals[0] + vals[4] + vals[8]) / 3.;
double dev0 = vals[0] + pressure;
double dev1 = vals[4] + pressure;
double dev2 = vals[8] + pressure;
// double invariant0 = dev0 + dev1 + dev2;
double invariant1 = 0.5*(dev0*dev0 + dev1*dev1 + dev2*dev2);
invariant1 += vals[1]*vals[1] + vals[2]*vals[2] + vals[5]*vals[5];
double invariant2 = -dev0*dev1*dev2;
invariant2 += -2.0 *vals[1]*vals[2]*vals[5];
invariant2 += dev0*vals[5]*vals[5];
invariant2 += dev1*vals[2]*vals[2];
invariant2 += dev2*vals[1]*vals[1];
double princ0 = 0.;
double princ1 = 0.;
double princ2 = 0.;
if (invariant1 >= 1e-100)
{
double alpha = -0.5*sqrt(27./invariant1)
*invariant2/invariant1;
if (alpha < 0.)
alpha = (alpha < -1. ? -1 : alpha);
if (alpha > 0.)
alpha = (alpha > +1. ? +1 : alpha);
double angle = acos((double)alpha) / 3.;
double value = 2.0 * sqrt(invariant1 / 3.);
princ0 = value*cos(angle);
angle = angle - 2.0*vtkMath::Pi()/3.;
princ1 = value*cos(angle);
angle = angle + 4.0*vtkMath::Pi()/3.;
princ2 = value*cos(angle);
}
double out3[3];
out3[0] = princ0;
out3[1] = princ1;
out3[2] = princ2;
```

- Principal Tensor Function:
`principal_tensor()`

`principal_tensor(expr0)`

Creates a new

*vector*expression which is everywhere the principal stress components of the input argument, which must a*symmetric*tensor. The principal stress components are the eigenvalues of the stress tensor. So, the vector expression computed here is the same as eigenvalue().Potentially, it would be more appropriate to create a new

*tensor*field here with all zeros for off-diagonal elements and the eigenvalues on the main diagonal.

- Transpose Function:
`transpose()`

`transpose(expr0)`

Creates a new tensor expression which is everywhere the transpose of its input argument which must also be a tensor. The first row vector in the input becomes the first column vector in the output, etc.

- Tensor Maximum Shear Function:
`tensor_maximum_shear()`

`tensor_maximum_shear(expr0)`

Creates a new

*Scalar*expression which is everywhere the*maximum shear stress*as defined in*J.C. Ugural and S.K. Fenster “Advanced Strength and Applied Elasticity”, Prentice Hall 4th Edition, page 81.*the specific mathematical operations of which are shown in the code snip-it below.

**Show/Hide Code for** `tensor_maximum_shear()`

```
double *vals = in->GetTuple9(i);
double s11 = vals[0], s12 = vals[1], s13 = vals[2];
double s21 = vals[3], s22 = vals[4], s23 = vals[5];
double s31 = vals[6], s32 = vals[7], s33 = vals[8];
// Hydro-static component
double pressure = (s11 + s22 + s33) / 3.;
// Deviatoric stress components
double dev0 = s11 - pressure;
double dev1 = s22 - pressure;
double dev2 = s33 - pressure;
// double invariant0 = dev0 + dev1 + dev2;
// Second invariant of stress deviator
double invariant1 = 0.5*(dev0*dev0 + dev1*dev1 + dev2*dev2);
invariant1 += s12*s12 + s13*s13 + s23*s23;
// Third invariant of stress deviator
double invariant2 = -dev0*dev1*dev2;
invariant2 += -2.0*s12*s13*s23;
invariant2 += dev0*s23*s23;
invariant2 += dev1*s13*s13;
invariant2 += dev2*s12*s12;
// Cubic roots of the characteristic equation
// http://mathworld.wolfram.com/CubicFormula.html
double princ0 = 0.;
double princ2 = 0.;
if (invariant1 >= 1e-100)
{
double alpha = -0.5*sqrt(27./invariant1)
*invariant2/invariant1;
if (alpha < 0.)
alpha = (alpha < -1. ? -1 : alpha);
if (alpha > 0.)
alpha = (alpha > +1. ? +1 : alpha);
double angle = acos((double)alpha) / 3.;
double value = 2.0 * sqrt(invariant1 / 3.);
princ0 = value*cos(angle);
// Displace the angle for princ1 (which we don't calculate)
angle = angle - 2.0*vtkMath::Pi()/3.;
// Now displace for princ2
angle = angle + 4.0*vtkMath::Pi()/3.;
princ2 = value*cos(angle);
}
// set the output value
```

- Trace Function:
`trace()`

`trace(expr0)`

Creates a new scalar expression which is everywhere the trace of

`expr0`

which must be a 3x3 tensor. The trace is the sum of the diagonal elements.

- Viscous Stress Function:
`viscous_stress()`

`viscous_stress(expr0)`

Creates a new tensor expression which is everywhere the viscous stress. The key difference between

*viscous*stress and*elastic*stress (which is the kind of stress many of the other functions here deal with) is that*viscous*stress is related to the*rate of change*of deformation whereas*elastic*stress is related to the*amount*of deformation. These two are related in the same way velocity and distance are related.The argument here,

`expr0`

is a*vector*valued velocity. In addition, the current implementation of this function works only for 2D, structured gridded meshes.

**Show/Hide Code for** `viscous_stress()`

```
dx[0] = .5 * (px[0] + px[1] - px[2] - px[3]);
dx[1] = .5 * (px[1] + px[2] - px[3] - px[0]);
dy[0] = .5 * (py[0] + py[1] - py[2] - py[3]);
dy[1] = .5 * (py[1] + py[2] - py[3] - py[0]);
du[0] = .5 * (vx[0] + vx[1] - vx[2] - vx[3]);
du[1] = .5 * (vx[1] + vx[2] - vx[3] - vx[0]);
dv[0] = .5 * (vy[0] + vy[1] - vy[2] - vy[3]);
dv[1] = .5 * (vy[1] + vy[2] - vy[3] - vy[0]);
div = 1.0 / (dx[0] *dy[1] - dx[1] *dy[0] + tiny);
dvx[0] = div * (du[0]*dy[1] - du[1] * dy[0]);
dvx[1] = div * (du[1]*dx[0] - du[0] * dx[1]);
dvy[0] = div * (dv[0]*dy[1] - dv[1] * dy[0]);
dvy[1] = div * (dv[1]*dx[0] - dv[0] * dx[1]);
// create the tensor
// if rz mesh include extra divergence term
if( rz_mesh )
{
cyl_term = (vy[0] + vy[1] + vy[2] + vy[3]) /
(py[0] + py[1] + py[2] + py[3] + tiny);
}
// diag terms
vstress[0] = 1/3.0 * (2.0 * dvx[0] - dvy[1]- cyl_term);
vstress[4] = 1/3.0 * (2.0 * dvy[1] - dvx[0]- cyl_term);
vstress[8] = 0.0;
// other terms
vstress[1] = 0.5 * (dvy[0] + dvx[1]);
vstress[2] = 0.0;
vstress[5] = 0.0;
// use symm to fill out remaining terms
vstress[3] = vstress[1];
vstress[6] = vstress[2];
vstress[7] = vstress[5];
}
```

### 8.1.3.6. Array Expressions¶

- Array Compose Function:
`array_compose()`

`array_compose(expr1, expr2, ..., exprK)`

Create a new

*array*expression variable which is everywhere the array composition of its arguments, which all must be*scalar*type. An array mesh variable is useful when using the label plot or when doing picks and wanting pick values to always return a certain selected set of mesh variables. But, all an array mesh variable really is is a convenient container to hold a group of individual scalar mesh variables. Each argument to the array_compose expression must evaluate to a scalar expression and all of the input expressions must have the same centering. Array variables are collections of scalar variables that are commonly used with certain plots to display the contents of multiple variables simultaneously. For example, the Label plot can display the values in an array variable.

- Array Compose With Bins Function:
`array_compose_with_bins()`

`array_compose_with_bins(expr1,expr2,...,exprK,[b0,...bK+1])`

This expression combines two related concepts. One is the array concept where a group of individual scalar mesh variables are grouped into an array variable. The other is a set of

*coordinate*values (interpreted as histogram bin boundaries), that will be used by VisIt for certain kinds of operations involving the array variable.**Note**that the bin boundaries are specified as a single additional argument to the function as a*list*of values embedded in square brackets. If there are`K`

variables in the array,`expr1, expr2,..., exprK`

, there are K+1 coordinate values (or bin boundaries),`[b0,b1,...,bK+1]`

. When such a variable is picked using one of VisIt’s pick operations, VisIt can plot a**Histogram**plot. Each bar in the**Histogram**plot has a height determined by the associated member of the array and a width determined by the associated bin-boundaries.For example, suppose a user had an array variable, foo, composed of 5 scalar mesh variables,

`a1`

,`a2`

,`a3`

,`a4`

, and`a5`

like so…`array_compose_with_bins(a1,a2,a3,a4,a5,[0,3.5,10.1,10.7,12,22])`

For any given point on a plot, when the user picked foo, there are 5 values returned, the value of each of the 5 scalar variable members of foo. If the user arranged for a pick to return a bar-graph of the variable using the bin-boundaries, the result might look like…

- Array Decompose Function:
`array_decompose()`

`array_decompose(Arr,Idx)`

Creates a new

*scalar*expression which is everywhere the scalar member of the*array*input argument at index`Idx`

(numbered starting from zero).

- Array Decompose 2D Function:
`array_decompose2d()`

`array_decompose2d(expr0)`

No description available.

- Array Component-wise Division Function:
`array_componentwise_division()`

`array_componentwise_division(<Array>,<Divisor>)`

Return a new

*array*variable which is the old input`<Array>`

variable with each of its components divided by the`<Divisor>`

.

- Array Component-wise Product Function:
`array_componentwise_product()`

`array_componentwise_product(<Array>,<Multiplier>)`

Return a new

*array*variable which is the old input`<Array>`

variable with each of its components multiplied by the`<Multiplier>`

.

- Array Sum Function:
`array_sum()`

`array_sum(<Array>)`

Return a new

*scalar*variable which is the sum of the`<Array>`

components.

### 8.1.3.7. Material Expressions¶

- Dominant Material Function:
`dominant_mat()`

`domimant_mat(<Mesh>)`

Creates a new scalar expression which is for every mesh cell/zone the material having the largest volume fraction.

- Material Error Function:
`materror()`

`materror(<Mat>,[Const,Const...])`

Creates a new scalar expression which is everywhere the difference in volume fractions as stored in the database and as computed by VisIt’s material interface reconstruction (MIR) algorithm. The

`<Mat>`

argument is a*material variable*from a database and the`Const`

argument is one of the material names as an quoted string or a material number as an integer. If multiple materials are to be selected from the*material variable*, enclose them in square brackets as a list.Examples…

materror(materials, 1) materror(materials, [1,3]) materror(materials, "copper") materror(materials, ["copper", "steel"])

- Material Volume Fractions Function:
`matvf()`

`matvf(<Mat>,[Const,Const,...])`

Creates a new scalar expression which is everywhere the sum of the volume fraction of the specified materials within the specified material variable. The

`<Mat>`

argument is a*material variable*from a database and the`Const`

argument(s) identify one or more materials within the*material variable*.Examples…

matvf(materials, 1) matvf(materials, [1,3]) matvf(materials, "copper") matvf(materials, ["copper", "steel"])

- NMats Function:
`nmats()`

`nmats(<Mat>)`

Creates a new scalar expression which for each mesh cell/zone is the number of materials in the cell/zone. The

`<Mat>`

argument is a*material variable*from a database.

- Specmf Function:
`specmf()`

`specmf(<Spec>,<MConst>,[Const,Const,...])`

Performs the analogous operation to

`matvf`

for species mass fractions. The`<Spec>`

argument is a*species variable*from a database. The`<MConst>`

argument is a specific material within the*species variable*. The`<Const>`

argument(s) identify which species within the*species variable*to select.Examples:

specmf(species, 1, 1) specmf(species, "copper", 1) specmf(species, "copper", [1,3])

- Value For Material Function:
`value_for_material()`

`value_for_material(<Var>,<Const>)`

Creates a new scalar expression which is everywhere the material-specific value of the variable specified by

`<Var>`

for the material specified by`<Const>`

. If variable specified by`<Var>`

has no material specific values, the values returned from this function will be just the variable’s values.

### 8.1.3.8. Mesh Expressions¶

- Area Function:
`area()`

`area(<Mesh>)`

See the Verdict Manual

- cylindrical Function:
`cylindrical()`

`cylindrical(<Mesh>)`

Creates a new vector variable on the mesh which is the cylindrical coordinate tuple (R,theta,Z) of each mesh node.

- Cylindrical Radius
`cylindrical_radius(<Mesh>)`

Creates a scalar new variable on the mesh which is the radius component of the cylindrical coordinate (from the Z axis) of each mesh node.

- cylindrical theta Function:
`cylindrical_theta()`

`cylindrical_theta(<Mesh>)`

Creates a new scalar variable on the mesh which is the angle component of the cylindrical coordinate (around the Z axis from the +X axis) of each mesh node.

- polar radius Function:
`polar_radius()`

`polar_radius(<Mesh>)`

Creates a new scalar variable on the mesh which is the radius component of the polar coordinate of each mesh node.

- polar theta Function:
`polar_theta()`

`polar_theta(<Mesh>)`

Creates a new scalar variable on the mesh which is the theta component of the polar coordinate of each mesh node.

- polar phi Function:
`polar_phi()`

`polar_phi(<Mesh>)`

Creates a new scalar variable on the mesh which is the phi component of the polar coordinate of each mesh node.

- min coord Function:
`min_coord()`

`min_coord(expr0)`

No description available.

- max coord Function:
`max_coord()`

`max_coord(expr0)`

No description available.

- external node Function:
`external_node()`

`external_node(expr0)`

No description available.

- external cell Function:
`external_cell()`

`external_cell(expr0)`

No description available.

- Zoneid Function:
`zoneid()`

`zoneid(<Mesh>)`

Return a zone-centered

*scalar*variable where the value for each zone/cell is local index of a zone, staring from zero, within its domain.

- Global Zoneid Function:
`global_zoneid()`

`global_zoneid(<Mesh>)`

If global zone ids are specified by the input database, return a zone-centered

*scalar*variable where the value for each zone/cell is the*global*index of a zone, as specified by the data producer.

- Nodeid Function:
`nodeid()`

`nodeid(expr0)`

Return a node-centered

*scalar*variable where the value for each node/vertex/point is local index of a node, staring from zero, within its domain.

- Global Nodeid Function:
`global_nodeid()`

`global_nodeid(expr0)`

If global node ids are specified by the input database, return a node-centered

*scalar*variable where the value for each node/vertex/point is the*global*index of a node, as specified by the data producer.

- Ghost Zoneid Function:
`ghost_zoneid()`

`ghost_zoneid(<Mesh>)`

Returns the ghost zone id of each zone in the mesh. The ghost zone id could be any combination of the following:

DUPLICATED_ZONE_INTERNAL_TO_PROBLEM = 0, ENHANCED_CONNECTIVITY_ZONE = 1, REDUCED_CONNECTIVITY_ZONE = 2, REFINED_ZONE_IN_AMR_GRID = 3, ZONE_EXTERIOR_TO_PROBLEM = 4, ZONE_NOT_APPLICABLE_TO_PROBLEM = 5

where each flag represents a bit shift by the specified number of bits. So if a zone is not a ghost zone, the value returned would be 0, while if it was a

`DUPLICATED_ZONE_INTERNAL_TO_PROBLEM`

and a`REFINED_ZONE_IN_AMR_GRID`

, the value returned would be 1001 in binary, or 9 in decimal.

- Volume Function:
`volume()`

`volume(<Mesh>)`

No description available.

- Volume2 Function:
`volume2()`

`volume2(<Mesh>)`

No description available.

- Revolved Volume Function:
`revolved_volume()`

`revolved_volume(<Mesh>)`

No description available.

- Revolved Surface Area Function:
`revolved_surface_area()`

`revolved_surface_area(<Mesh>)`

No description available.

- Zone Type Function:
`zonetype()`

`zonetype(<Mesh>)`

Return a

*zone*centered, character valued variable which indicates the*shape type*of each zone suitable for being used within the*label*plot. Upper case characters generally denote 3D shapes (e.g.`T`

for`tet`

) while lower case characters denote 2D shapes (e.g.`t`

for`triangle`

).

- Zone Type Rank Function:
`zonetype_rank()`

`zonetype_rank(<Mesh>)`

Return a

*zone*centered, integer valued variable which indicates the*VTK shape type*of each zone. This expression is often useful with the threshold operator to select only certain shapes within the mesh to be displayed.

### 8.1.3.9. Mesh Quality Expressions¶

VisIt employs the *Verdict Mesh Quality Library* to support a number of
expressions related to computing cell-by-cell mesh quality metrics. The
specific definitions of the various mesh quality metrics defined by the
*Verdict Mesh Quality Library* are amply explained in the
`Verdict Manual`

. Below, we
simply list all the mesh quality metrics and describe in detail only
those that are not part of the *Verdict Mesh Quality Library*

In all cases in the **Mesh Quality Expressions**, the input argument is
a *mesh variable* from a database and the output is a *scalar* expression.

- Neighbor Function:
`neighbor()`

`neighbor(<Mesh>)`

See the Verdict Manual

- Node Degree Function:
`node_degree()`

`node_degree(<Mesh>)`

Return a

*node*centered, integer valued variable which indicates the*number*of mesh zones/cells that share each node.

- Degree Function:
`degree()`

`degree(expr0)`

Return a

*node*centered, integer valued variable which indicates the*number*of mesh edges incident to each node.

- Aspect Function:
`aspect()`

`aspect(<Mesh>)`

See the Verdict Manual

- Skew Function:
`skew()`

`skew(<Mesh>)`

See the Verdict Manual

- Taper Function:
`taper()`

`taper(<Mesh>)`

See the Verdict Manual

- Minimum Corner Angle Function:
`min_corner_angle()`

`min_corner_angle(<Mesh>)`

See the Verdict Manual

- Maximum Corner Angle Function:
`max_corner_angle()`

`max_corner_angle(<Mesh>)`

See the Verdict Manual

- Minimum Edge Length Function:
`min_edge_length()`

`min_edge_length(<Mesh>)`

See the Verdict Manual

- Maximum Edge Length Function:
`max_edge_length()`

`max_edge_length(<Mesh>)`

See the Verdict Manual

- Minimum Side Volume Function:
`min_side_volume()`

`min_side_volume(<Mesh>)`

See the Verdict Manual

- Maximum Side Volume Function:
`max_side_volume()`

`max_side_volume(<Mesh>)`

See the Verdict Manual

- Stretch Function:
`stretch()`

`stretch(<Mesh>)`

See the Verdict Manual

- Diagonal Ratio Function:
`diagonal_ratio()`

`diagonal_ratio(<Mesh>)`

See the Verdict Manual

- Maximum Diagonal Function:
`max_diagonal()`

`max_diagonal(<Mesh>)`

See the Verdict Manual

- Minimum Diagonal Function:
`min_diagonal()`

`min_diagonal(<Mesh>)`

See the Verdict Manual

- Dimension Function:
`dimension()`

`dimension(<Mesh>)`

See the Verdict Manual

- Oddy Function:
`oddy()`

`oddy(<Mesh>)`

See the Verdict Manual

- Condition Function:
`condition()`

`condition(<Mesh>)`

See the Verdict Manual

- Jacobian Function:
`jacobian()`

`jacobian(<Mesh>)`

See the Verdict Manual

- Scaled Jacobian Function:
`scaled_jacobian()`

`scaled_jacobian(<Mesh>)`

See the Verdict Manual

- Shear Function:
`shear()`

`shear(<Mesh>)`

See the Verdict Manual

- Shape Function:
`shape()`

`shape(<Mesh>)`

See the Verdict Manual

- Relative Size Function:
`relative_size()`

`relative_size(<Mesh>)`

See the Verdict Manual

- Shape and Size Function:
`shape_and_size()`

`shape_and_size(<Mesh>)`

See the Verdict Manual

- Aspect Gamma Function:
`aspect_gamma()`

`aspect_gamma(<Mesh>)`

See the Verdict Manual

- Warpage Function:
`warpage()`

`warpage(<Mesh>)`

See the Verdict Manual

- Maximum Angle Function:
`maximum_angle()`

`maximum_angle(<Mesh>)`

See the Verdict Manual

- Minimum Angle Function:
`minimum_angle()`

`minimum_angle(<Mesh>)`

See the Verdict Manual

- Minimum Corner Area Function:
`min_corner_area()`

`min_corner_area(<Mesh>)`

See the Verdict Manual

- Minimum Sin Corner Function:
`min_sin_corner()`

`min_sin_corner(<Mesh>)`

See the Verdict Manual

- Minimum Sin Corner CW Function:
`min_sin_corner_cw()`

`min_sin_corner_cw(<Mesh>)`

See the Verdict Manual

- Face Planarity Function:
`face_planarity()`

`face_planarity(<Mesh>)`

Creates a new expression which is everywhere a measure of how close to

*planar*all the points comprising a face are. This is computed for each face of a cell and the maximum over all faces is selected for each cell. Planarity is measured as the maximum distance from an arbitrary plane defined by the first 3 points of a face of the remaining points of the face. Values closer to zero are*better*. A triangle face will always have a planarity measure of zero. This mesh quality expression is not part of the Verdict library.

- Relative Face Planarity Function:
`relative_face_planarity()`

`relative_face_planarity(<Mesh>)`

Performs the same computation as the face_planarity(), except where each face’s value is normalized by the average edge length of the face.

### 8.1.3.10. Comparison Expressions¶

Comparing variables defined on the *same* mesh is often as simple as taking
their difference. What about comparing variables when they are defined on
different meshes? A common example is taking the difference between results
from two runs of the same simulation application. Even if the two runs operate
on computationally *identical* meshes, the fact that each run involves its own
*instance* of that mesh means that as far as VisIt is concerned, they are
different meshes.

In order to compose an expression involving variables on different meshes, the
*first* step is to *map* the variables onto a *common* mesh. The position-based
CMFE function and its friend, the connectivity-based CMFE function,
conn_cmfe() are the work-horse methods
needed when working with variables from *different* meshes in the *same*
expression. *CMFE* is an abbreviation for *cross-mesh field evaluation*.

The syntax for specifying CMFE expressions can be complicated.
Therefore, the GUI supports a *wizard* to help create them.
See the Data-Level Comparisons Wizard for more information.
It sometimes makes sense to use the wizard to create an *initial* CMFE expression and then modify it manually, often to adjust the state indexing.
Here, we describe the details of creating CMFE expressions manually.

All of the comparison expressions involve the concepts of a *donor variable*
and a *target mesh*. The donor variable (e.g. *pressure*) is the variable to
be mapped. The target mesh is the mesh onto which the donor variable is to be
mapped. In addition, the term *donor mesh* refers to the mesh upon which the
donor variable is defined.

- Position-Based CMFE Function:
`pos_cmfe()`

`pos_cmfe(<Donor Variable>,<Target Mesh>,<Fill>)`

The

`pos_cmfe()`

function performs the mapping assuming the two meshes, that is the`<Target Mesh>`

and the mesh upon which the`<Donor Variable>`

(e.g. the*donor mesh*) is defined, share*only*a common spatial (positional) extent. Its friend, the conn_cmfe() function is*optimized*to perform the mapping when the two meshes are also*topologically identical*. In other words, their*coordinate***and***connectivity*arrays are 1:1. In this case, the mapping can be performed with more efficiency and numerical accuracy. Therefore, when it is possible and makes sense to do so, it is always best to use`conn_cmfe()`

.We’ll describe the arguments to

`pos_cmfe()`

working backwards from the last.The last,

`<Fill>`

argument is a numerical constant that VisIt will use to determine the value of the result in places on the target mesh that do not spatially overlap with the mesh of the donor variable. Note that if a value is chosen within the range of the donor variable, it may by difficult to distinguish regions VisIt deemed were non-overlapping. On the other hand, if a value outside the range is chosen, it will effect the range of the mapped variable. A common practice is to choose a value that is an extremum of the donor variable’s range. Another practice is to choose a value that is easily distinguishable and then apply a threshold operator to remove those portions of the result. If the`Fill`

argument is not specified, zero is assumed.Working backwards, the next argument, is the

`<Target Mesh>`

. The`<Target Mesh>`

argument in`pos_cmfe()`

is always interpreted as a mesh*within*the currently*active*database. The CMFE expressions are always mapping data from*other*meshes, possibly in*other*databases onto the`<Target Mesh>`

which is understood to be in the currently*active*database. When mapping data between meshes*in different databases*, the additional information necessary to specify the other database is encoded with a special syntax prepending the`Donor Variable`

argument.

The

`Donor Variable`

argument is a string argument of the form:<PATH-TO-DATABASE-FROM-CWD[SSS]MM:VARNAME>consisting of the donor variable’s name and up to three pre-pending sub-strings which may be optionally needed to specify…

…the

Database(`PATH-TO-DATABASE-FROM-CWD`

) in which the donor variable resides,…the

State Id(`[SSS]`

) from which to take the donor variable,…the

Modality(`MM`

) by which states are identified in theState Idsub-string.Depending on circumstances, specifying the

`Donor-Variable`

argument to the CMFE functions can get cumbersome. For this reason, CMFE expressions are typically created using the Data-Level Comparisons Wizard under theControlsmenu. Nonetheless, here we describe the syntax and provide examples for a number of cases of increasing complexity in specifying where the`Donor Variable`

resides.When the donor variable is in the same database and state as the target mesh, then only the variable’s name is needed. The optional substrings are not. See case A in the examples below.

When the donor variable is in a different database

andthe databases do not have multiple time states, then only sub-string 1, above, is needed to specify the path to the database in the file system. The path to the database can be specified using eitherabsoluteorrelativepaths.Relativepaths are interpreted relative to the current working directory in which the VisIt session was started. See cases B and C in the examples below.When the donor variable is in a different database

andthe databases have multiple states, then all 3 sub-strings, above, are required. The`State Id`

substring is a square-bracket enclosed number used to identifywhich statefrom which to take the donor variable. The`Modality`

substring is a one- or two-character moniker. The first character indicates whether the number in the the`State Id`

substring is a cycle (`c`

), a time (`t`

), or an index (`i`

). The second character, if present, is a`d`

character to indicate the cycle, time or index isrelative(e.g. adelta) to the current state. For example, the substring`[200]c`

means to treat the`200`

as acyclenumber in the donor database whereas the the substring`[-10]id`

means to treat the`-10`

as an (`i`

) index (`d`

) delta. So,`[200]c`

would map thedonorat cycle 200 to thecurrentcycle of thetargetand`[-10]id`

would map thedonorat the currentindex minus 10to thecurrentindex of thetarget. In particular, the string`[0]id`

is needed to create a CMFE that keepsdonorandtargetin lock step. Note that in cases where the donor database does not have an exact match for the specified cycle or time, VisIt will chose the state with the cycle or time which is closest in absolute distance. For theindexmodality, if there is no exact match for the specified index, an error results. See cases D-I in the examples below.Note that the

relativeform of specifying theState Idis needed even when working with different stateswithin the same database. In particular, to create an expression representing atime derivativeof a variable in a database, the key insight is to realize it involves mapping a donor variable from one state in the database onto a mesh at another state. In addition, the value in using therelativeform of specifying the`State Id`

of the donor variable is that as the current time is changed, the expression properly identifies the different states of the donor variable instead of always mapping a fixed state.Examples…

# Case A: Donor variable, "pressure" in same database as mesh, "ucdmesh" # Note that due to a limitation in Expression parsing, the '[0]id:' is # currently required in the donor variable name as a substitute for # specifying a file system path to a database file. The syntax '[0]id:' # means a state index delta of zero within the active database. pos_cmfe(<[0]id:pressure>,<ucdmesh>,1e+15) # Case B: Donor variable in a different database using absolute path pos_cmfe(</var/tmp/foo.silo:pressure>,<ucdmesh>,1e+15) # Case C: Donor variable in a different database using relative path pos_cmfe(<foo/bar.silo:pressure>,<ucdmesh>,1e+15) # Case D: Map "p" from wave.visit at state index=7 onto "mesh" pos_cmfe(<./wave.visit[7]i:p>, mesh, 1e+15) # Case E: Map "p" from wave.visit at state index current-1 onto "mesh" pos_cmfe(<./wave.visit[-1]id:p>, mesh, 1e+15) # Case F: Map "p" from wave.visit at state with cycle~200 onto "mesh" pos_cmfe(<./wave.visit[200]c:p>, mesh, 1e+15) # Case G: Map "p" from wave.visit at state with cycle~cycle(current)-200 onto "mesh" pos_cmfe(<./wave.visit[-200]id:p>, mesh, 1e+15) # Case H: Map "p" from wave.visit at state with time~1.4 onto "mesh" pos_cmfe(<./wave.visit[1.4]t:p>, mesh, 1e+15) # Case I: Map "p" from wave.visit at state with time~time(current)-0.8 onto "mesh" pos_cmfe(<./wave.visit[-0.8]td:p>, mesh, 1e+15)

- Connectivity-Based CMFE Function:
`conn_cmfe()`

`conn_cmfe(<Donor Variable>,<Target Mesh>)`

The connectivity-based CMFE is an

*optimized*version of pos_cmfe() for cases where the`Target Mesh`

and the mesh of the`Donor Variable`

are*topologically and geometrically identical*. In such cases, there is no opportunity for the two meshes to fail to overlap perfectly. Thus, there is no need for the third,`<Fill>`

argument. In all other respects,`conn_cmfe()`

performs the same function as pos_cmfe() except that`conn_cmfe()`

*assumes*that any differences in the coordinates of the two meshes are numerically insignificant to the resulting mapped variable. In other words, differences in the coordinate fields, if they exist, are not incorporated into the resulting mapping.

- Curve CMFE Function:
`curve_cmfe()`

`curve_cmfe(<Donor Curve>,<Target Curve>)`

The curve-based CMFE performs the same function as pos_cmfe() except for curves. The arguments specify the

`Target Curve`

and`Donor Curve`

and the same syntax rules apply for specifying the`Donor Curve`

as for the`Donor Variable`

in pos_cmfe(). However, if curves represent different spatial extents or different numbers of samples or sample spacing, no attempt is made to unify them.

- Symmetric Difference By Point Function:
`symm_point()`

`symm_point(<Scalar>,<Fill>,[Px,Py,Pz])`

Return a new

*scalar*variable which is the symmetric difference of`<Scalar>`

reflected about the point`[Px, Py, Pz]`

. In 2D,`Pz`

is still required but ignored. The`<Fill>`

argument is a numerical constant that VisIt will use to determine the value of the result in places symmetry about the point doesn’t overlap with the donor mesh. This operation involves**both**the reflection about the point**and**taking the difference. If the input`<Scalar>`

is indeed symmetric about the point, the result will be a constant valued variable of zero.

- Symmetric Difference By Plane Function:
`symm_plane()`

`symm_plane(<Scalar>,<Fill>,[Nx,Ny,Nz,Px,Py,Pz])`

Return a new

*scalar*variable which is the symmetric difference of`<Scalar>`

reflected about the plane defined by the point`[Px, Py, Pz]`

and normal`[Nx, Ny, Nz]`

. In 2D, the`Nz`

and`Pz`

arguments are still required but ignored. The`<Fill>`

argument is a numerical constant that VisIt will use to determine the value of the result in places symmetry about the plane doesn’t overlap with the donor mesh. This operation involves**both**the reflection about the plane**and**taking the difference. If the input`<Scalar>`

is indeed symmetric about the plane, the result will be a constant valued variable of zero.

- Symmetric Difference By Transform Function:
`symm_transform()`

`symm_transform(<Scalar>,<Fill>,[T00,T01,T02,...,T22])`

Return a new

*scalar*variable which is the symmetric difference of`<Scalar>`

reflected through the 3x3 transformation where each point,`[Px,Py,Pz]`

, in the mesh supporting`<Scalar>`

is transformed by the transform coefficients,`[T00, T01,...,T22]`

as shown below. In 2D, all 9 transform coefficients are still required but the last row and column are ignored. The`<Fill>`

argument is a numerical constant that VisIt will use to determine the value of the result in places symmetry through the transform doesn’t overlap with the donor mesh. This operation involves**both**the transform**and**taking the difference. If the input`<Scalar>`

is indeed symmetric through the transform, the result will be a constant valued variable of zero.

- Evaluate Point Function:
`eval_point()`

`eval_point(<Scalar>,<Fill>,[Px,Py,Pz])`

Performs only the reflection half of the symm_point() operation. That is, it computes a new

*scalar*variable which is the input`<Scalar>`

reflected through the symmetric point. It does not then take the*difference*between with the input`<Scalar>`

as symm_point() does.

- Evaluate Plane Function:
`eval_plane()`

`eval_plane(<Scalar>,<Fill>,[Nx,Ny,Nz,Px,Py,Pz])`

Performs only the reflection half of the symm_plane() operation. That is, it computes a new

*scalar*variable which is the input`<Scalar>`

reflected through the symmetric plane. It does not then take the*difference*between with the input`<Scalar>`

as symm_plane() does.

- Evaluate Transform Function:
`eval_transform()`

`eval_transform(expr0,<Fill>,[T00,T01,T02...T22])`

Performs only the transform half of the symm_transform() operation. That is, it computes a new

*scalar*variable which is the input`<Scalar>`

mapped through the transform. It does not then take the*difference*between with the input`<Scalar>`

as symm_transform() does.

### 8.1.3.11. Image Processing Expressions¶

The image processing expressions defined here are not suitable for multi-block
data. They do not handle domain boundaries properly even if the input database
properly defines suitable layers of *ghost* zones. They do, however, operate
on 2 and 3D data.

- conservative smoothing Function:
`conservative_smoothing()`

`conservative_smoothing(expr0)`

No description available.

- Mean Filter Function:
`mean_filter()`

`mean_filter(<Scalar>,<Int>)`

Return a filtered version of the input

*scalar*variable using the mean filter of width specified by`<Int>`

argument. By default, the filter width is 3 (3x3). The input scalar must be defined on a structured mesh.

- Median Filter Function:
`median_filter()`

`median_filter(expr0)`

Return a filtered version of the input

*scalar*variable using a 3x3 median filter. The input scalar must be defined on a structured mesh.

- Abel Inversion Function:
`abel_inversion()`

`abel_inversion(expr0)`

No description available.

### 8.1.3.12. Miscellaneous Expressions¶

- Zonal Constant Function:
`zonal_constant()`

`zonal_constant(expr0)`

Return a

*scalar*, zone-centered field that is everywhere on`<Mesh>`

the constant value`<Const>`

.

- Zone Constant Function:
`zone_constant()`

`zone_constant(<Mesh>,<Const>)`

An alias for zonal_constant()

- Cell Constant Function:
`cell_constant()`

`cell_constant(expr0)`

An alias for zonal_constant()

- Nodal Constant Function:
`nodal_constant()`

`nodal_constant(<Mesh>,<Const>)`

Return a

*scalar*, node-centered field that is everywhere on`<Mesh>`

the constant value`<Const>`

.

- Node Constant Function:
`node_constant()`

`node_constant(expr0)`

An alias for nodal_constant()

- Point Constant Function:
`point_constant()`

`point_constant(expr0)`

An alias for nodal_constant()

- Time Function:
`time()`

`time(expr0)`

Return a

*constant scalar*variable which is everywhere the time of the associated input argument within its time-series.

- Cycle Function:
`cycle()`

`cycle(expr0)`

Return an integer

*constant scalar*variable which is everywhere the cycle of the associated input argument within its time-series.

- Timestep Function:
`timestep()`

`timestep(expr0)`

Return an integer

*constant scalar*variable which is everywhere the index of the associated input argument within its time-series.

- curve domain Function:
`curve_domain()`

`curve_domain(expr0)`

No description available.

- curve integrate Function:
`curve_integrate()`

`curve_integrate(expr0)`

No description available.

- curve swapxy Function:
`curve_swapxy()`

`curve_swapxy(expr0)`

No description available.

- curve Function:
`curve()`

`curve(expr0)`

No description available.

- Enumerate Function:
`enumerate()`

`enumerate(<Int-Scalar>,<[Int-List]>)`

Map an integer valued

*scalar*variable to a new set of integer values. If*K*is the maximum value in the`Int-Scalar`

input argument, the`[Int-List]`

argument must be a square bracketed list of*K+1*integer values. Value*i*in the`Int-Scalar`

input argument is used to index the*ith*entry in the`[Int-List]`

to produce mapped value.

- Map Function:
`map()`

`map(<Scalar>,<[Input-Value-List]>,<[Output-Value-List]>, fill_value)`

A more general form of enumerate() which supports non-integer input

*scalar*variables and input and output maps which are not required to include all values in the input*scalar*variable. The`[Input-Value-List]`

and`[Output-Value-List]`

must have the same number of entries. A value in the input*scalar*variable that matches the*ith*entry in the`[Input-Value-List]`

is mapped to the new value at the*ith*entry in the`[Output-Value-List]`

. Values that do not match any entry in the`[Input-Value-List]`

are mapped to`fill_value`

, which is`-1`

by default.

- Resample Function:
`resample()`

`resample(<Var>,Nx,Ny,Nz)`

Resample

`<Var>`

onto a regular grid defined by taking the X, Y and for 3D, Z spatial extents of the mesh`<Var>`

is defined on and taking`Nx`

samples over the spatial extents in X,`Ny`

samples over the spatial extents in Y, and, for 3D,`Nz`

samples over the spatial extents in Z. Any samples that*miss*the mesh`<Var>`

is defined on are assigned the value`-FLT_MAX`

. For 2D, the`Nz`

argument is still required but ignored.

- Recenter Expression Function
`recenter(expr, ["nodal", "zonal", "toggle"])`

This function can be used to recenter

`expr`

. The second argument is optional and defaults to*“toggle”*if it is not specified. A value of*“toggle”*for the second argument means that if`expr`

is*node*centered, it is recentered to*zone*centering and if`expr`

is*zone*centered, it is recentered to*node*centering. Note that the quotes are required for the second argument. This function is typically used to force a specific centering among the operands of some other expression.

- Process Id Function:
`procid()`

`procid(<Var>)`

Return an integer

*scalar*variable which is everywhere the MPI rank associated with each of the blocks of the possibly parallel decomposed mesh upon which`<Var>`

is defined. For serial execution or for parallel execution of a single-block mesh, this will produce a constant zero variable. Otherwise, the values will vary block by block.

- Thread Id Function:
`threadid()`

`threadid(expr0)`

Return an integer

*scalar*variable which is everywhere the local thread id associated with each of the blocks of the possibly parallel decomposed mesh upon which`<Var>`

is defined. For non-threaded execution, this will produce a constant zero variable. Otherwise, the values will vary block by block.

- isnan Function:
`isnan()`

`isnan(expr0)`

No description available.

- q criterion Function:
`q_criterion()`

`q_criterion(<gradient(velocity[0])>, <gradient(velocity[1])>, <gradient(velocity[2])>)`

Generates the Q-criterion value developed by Hunt et. al.. It is based on the observation that, in regions where the Q-criterion is greater than zero, rotation exceeds strain and, in conjunction with a pressure min, indicates the presence of a vortex. The three arguments to the function are gradient vectors of the x-, y-, and z-velocity. The gradient function (see gradient()) can be used to create the gradient vectors.

- lambda2 Function:
`lambda2()`

`lambda2(<gradient(velocity[0])>, <gradient(velocity[1])>, <gradient(velocity[2])>)`

Generates the Lambda-2 criterion. It is based on the observation that, in regions where Lambda-2 is less than zero, rotation exceeds strain and, in conjunction with a pressure min, indicates the presence of a vortex. The three arguments to the function are gradient vectors of the x-, y-, and z-velocity. The gradient function (see gradient()) can be used to create the gradient vectors.

- mean curvature Function:
`mean_curvature()`

`mean_curvature(expr0)`

No description available.

- Gauss Curvature Function:
`gauss_curvature()`

`gauss_curvature(expr0)`

No description available.

- agrad Function:
`agrad()`

`agrad(expr0)`

No description available.

- key aggregate Function:
`key_aggregate()`

`key_aggregate(expr0)`

No description available.

- rectilinear Laplacian Function:
`rectilinear_laplacian()`

`rectilinear_laplacian(expr0)`

No description available.

- conn components Function:
`conn_components()`

`conn_components(expr0)`

No description available.

- resrad Function:
`resrad()`

`resrad(expr0)`

No description available.

crack width Function: `crack_width()`

: `crack_width(crack_num, <crack1_dir>, <crack2_dir>, <crack3_dir>, <strain_tensor>, volume2(<mesh_name>))`

Calculates crack width using the following formula:

crackwidth = L * (1 - (exp(-delta)) where: L = ZoneVol / (Area perpendicular to crack_dir) find Area by slicing the cell by plane with origin == cell center and Normal == crack_dir. Take area of that slice. delta = T11 for crack dir1 = component 0 of strain_tensor T22 for crack dir2 = component 4 of strain_tensor T33 for crack dir3 = component 8 of strain_tensor

### 8.1.3.13. Time Iteration Expressions¶

- Average Over Time Function:
`average_over_time()`

`average_over_time(<Scalar>,<Start>,<Stop>,<Stride>)`

Return a new

*scalar*variable in which each zonal or nodal value is the average over the times indicated by`Start`

,`Stop`

and`Stride`

.

- Min Over Time Function:
`min_over_time()`

`min_over_time(<Scalar>,<Start>,<Stop>,<Stride>)`

Return a new

*scalar*variable in which each zonal or nodal value is the minimum value the variable,`<Scalar>`

, attained over the times indicated by`Start`

,`Stop`

and`Stride`

.

- Max Over Time Function:
`max_over_time()`

`max_over_time(<Scalar>,<Start>,<Stop>,<Stride>)`

Return a new

*scalar*variable in which each zonal or nodal value is the maximum value the variable,`<Scalar>`

, attains over the times indicated by`Start`

,`Stop`

and`Stride`

.

- Sum Over Time Function:
`sum_over_time()`

`sum_over_time(<Scalar>,<Start>,<Stop>,<Stride>)`

Return a new

*scalar*variable in which each zonal or nodal value is the sum of the values the variable,`<Scalar>`

attains over the times indicated by`Start`

,`Stop`

and`Stride`

.

- First Time When Condition Is True Function:
`first_time_when_condition_is_true()`

`first_time_when_condition_is_true(<Cond>,<Fill>,<Start>,<Stop>,<Stride>)`

Return a new

*scalar*variable in which each zonal or nodal value is the*first*time (not cycle and not time-index, but floating point time) at which the true/false condition,`<Cond>`

is true. The`<Fill>`

value is used if there is no*first*time the condition is true.

- Last Time When Condition Is True Function:
`last_time_when_condition_is_true()`

`last_time_when_condition_is_true(<Cond>,<Fill>,<Start>,<Stop>,<Stride>)`

Return a new

*scalar*variable in which each zonal or nodal value is the*last*time (not cycle and not time-index, but floating point time) at which the true/false condition,`<Cond>`

is true. The`<Fill>`

value is used if there is no*last*time the condition is true.

- First Cycle When Condition Is True Function:
`first_cycle_when_condition_is_true()`

`first_cycle_when_condition_is_true(<Cond>,<Fill>,<Start>,<Stop>,<Stride>)`

Return a new integer valued

*scalar*variable in which each zonal or nodal value is the*first*cycle (not time and not time-index, but integer cycle) at which the true/false condition,`<Cond>`

is true. The`<Fill>`

value is used if there is no*first*cycle the condition is true.

- Last Cycle When Condition Is True Function:
`last_cycle_when_condition_is_true()`

`last_cycle_when_condition_is_true(<Cond>,<Fill>,<Start>,<Stop>,<Stride>)`

Return a new integer valued

*scalar*variable in which each zonal or nodal value is the*last*cycle (not time and not time-index, but integer cycle) at which the true/false condition,`<Cond>`

is true. The`<Fill>`

value is used if there is no*last*cycle the condition is true.

- First Time Index When Condition Is True Function:
`first_time_index_when_condition_is_true()`

`first_time_index_when_condition_is_true(<Cond>,<Fill>,<Start>,<Stop>,<Stride>)`

Return a new integer valued

*scalar*variable in which each zonal or nodal value is the*first*time index (not cycle and not time, but integer time-index) at which the true/false condition,`<Cond>`

is true. The`<Fill>`

value is used if there is no*first*time-index the condition is true.

- Last Time Index When Condition Is True Function:
`last_time_index_when_condition_is_true()`

`last_time_index_when_condition_is_true(<Cond>,<Fill>,<Start>,<Stop>,<Stride>)`

Return a new integer valued

*scalar*variable in which each zonal or nodal value is the*last*time index (not cycle and not time, but integer time-index) at which the true/false condition,`<Cond>`

is true. The`<Fill>`

value is used if there is no*last*time-index the condition is true.

- var when condition is first true Function:
`var_when_condition_is_first_true()`

`var_when_condition_is_first_true(expr0)`

No description available.

- var when condition is last true Function:
`var_when_condition_is_last_true()`

`var_when_condition_is_last_true(expr0)`

No description available.

- time at minimum Function:
`time_at_minimum()`

`time_at_minimum(expr0)`

No description available.

- cycle at minimum Function:
`cycle_at_minimum()`

`cycle_at_minimum(expr0)`

No description available.

- time index at minimum Function:
`time_index_at_minimum()`

`time_index_at_minimum(expr0)`

No description available.

- value at minimum Function:
`value_at_minimum()`

`value_at_minimum(expr0)`

No description available.

- time at maximum Function:
`time_at_maximum()`

`time_at_maximum(expr0)`

No description available.

- cycle at maximum Function:
`cycle_at_maximum()`

`cycle_at_maximum(expr0)`

No description available.

- time index at maximum Function:
`time_index_at_maximum()`

`time_index_at_maximum(expr0)`

No description available.

- value at maximum Function:
`value_at_maximum()`

`value_at_maximum(expr0)`

No description available.

- localized compactness Function:
`localized_compactness()`

`localized_compactness(expr0)`

No description available.

- merge tree Function:
`merge_tree()`

`merge_tree(expr0)`

No description available.

- split tree Function:
`split_tree()`

`split_tree(expr0)`

No description available.

- local threshold Function:
`local_threshold()`

`local_threshold(expr0)`

No description available.

- python Function:
`python()`

`python(expr0)`

No description available.

- relative difference Function:
`relative_difference()`

`relative_difference(expr0)`

No description available.

- var skew Function:
`var_skew()`

`var_skew(expr0)`

No description available.

- apply data binning Function:
`apply_data_binning()`

`apply_data_binning(expr0)`

No description available.

- distance to best fit line Function:
`distance_to_best_fit_line()`

`distance_to_best_fit_line(expr0)`

No description available.

- distance to best fit line2 Function:
`distance_to_best_fit_line2()`

`distance_to_best_fit_line2(expr0)`

No description available.

- geodesic vector quantize Function:
`geodesic_vector_quantize()`

`geodesic_vector_quantize(expr0)`

No description available.

- bin Function:
`bin()`

`bin(expr0)`

No description available.

- biggest neighbor Function:
`biggest_neighbor()`

`biggest_neighbor(expr0)`

No description available.

- smallest neighbor Function:
`smallest_neighbor()`

`smallest_neighbor(expr0)`

No description available.

- neighbor average Function:
`neighbor_average()`

`neighbor_average(expr0)`

No description available.

## 8.1.4. Expression Compatibility Gotchas¶

VisIt will allow you to define expressions that it winds up determining to be
invalid later when it attempts to execute those expressions. Some common
issues are the mixing of incompatible mesh variables in the same expression
*without* the necessary additional functions to make them compatible.

### 8.1.4.1. Tensor Rank Compatibility¶

For example, what happens if you mix scalar and vector mesh variables in the same expression? VisIt will allow users to define such an expression. But, when it is plotted, the plot will fail.

As an aside, as the user goes back and forth between the Expressions window creating and/or adjusting expression definitions, VisIt makes no attempt to keep track of all the changes made in expressions and automatically update plots as expressions change. Users have to manually clear or delete plots to force VisIt to re-draw plots in which the expressions changed.

If what is really intended was a scalar mesh variable, then users must use one of the expression functions that converts a vector to a scalar such as the magnitude() built-in expression or the array de-reference operator.

### 8.1.4.2. Centering Compatibility¶

Some variables are zone centered and some are node centered. What happens if a user combines these in an expression? VisIt will default to zone centering for the result. If this is not the desired result, the recenter() expression function should be used, where appropriate, to adjust centering of some of the terms in the expression. Note that ordering of operations will probably be important. For example

```
node_var + recenter(zone_var)
recenter(zone_var + node_var)
```

both achieve a node-centered result. But, each expression is subtly (and numerically) different. The first recenter’s zone_var to the nodes and then performs the summation operator at each node. In the second, there is an implied recentering of node_var to the zones first. Then, the summation operator is applied at each zone center and finally the results are recentered back to the nodes. In all likelihood this creates in a numerically lower quality result. The moral is that in a complex series of expressions be sure to take care where you want recentering to occur.

### 8.1.4.3. Mesh Compatibility¶

In many cases, especially in Silo databases, all the available variables in a database are not always defined on the same mesh. This can complicate matters involving expressions in variables from different meshes.

Just as in the previous two examples of incompatible variables where the
solution was to apply some function to make the variables compatible, we have
to do the same thing when variables from different meshes are combined in an
expression. The key expression functions which enable this are called
**Cross Mesh Field Evaluation** or **CMFE** expression functions. We will only
briefly touch on these here. **CMFEs** will be discussed in much greater detail
elsewhere.

Just as for centering, we have two options when dealing with variables from
two different meshes. Each of which involves *mapping* one of the variables
onto the other variable’s mesh using one of the CMFE expression functions.

## 8.1.5. Deferring expression evaluation¶

Expressions are generally evaluated before any operators are applied to the data.
In cases where an expression involves the *output* of an operator, or the operator behaves in such a way as to change the outcome of an expression, then it is necessary to *defer* evaluation of such an expression until *after* operators have been applied.
The DeferExpression operator is designed for this purpose.
It will cause expressions in its list to be evaluated at the time of it’s own execution in the pipeline.