# log2

Base 2 logarithm and floating-point number dissection

## Syntax

``Y = log2(X)``
``[F,E] = log2(X)``

## Description

example

````Y = log2(X)` computes the base 2 logarithm of the elements of `X` such that ${2}^{Y}=X$.```

example

````[F,E] = log2(X)` returns arrays `F` and `E` such that $X=F\cdot {2}^{E}$. The values in `F` are typically in the range `0.5 <= abs(F) < 1`.```

## Examples

collapse all

```X = [0 1 2 10 Inf NaN]; Y = log2(X)```
```Y = 1×6 -Inf 0 1.0000 3.3219 Inf NaN ```

Dissect several numbers into the exponent and mantissa. These operations all follow standard IEEE® arithmetic.

Create a vector X that contains several test values. Calculate the exponent and mantissa for each number.

```X = [1 pi -3 eps realmax realmin]; format rat [F,E] = log2(X)```
```F = Columns 1 through 5 1/2 355/452 -3/4 1/2 1 Column 6 1/2 ```
```E = Columns 1 through 5 1 2 2 -51 1024 Column 6 -1021 ```

Collect the results in a table. Convert the numbers into character vectors for display purposes.

```x = {'1','pi','-3','eps','realmax','realmin'}'; f = strtrim(cellstr(rats(F'))); T = table(x,f,E','VariableNames',{'Value','Mantissa','Exponent'})```
```T=6×3 table Value Mantissa Exponent ___________ ___________ ________ {'1' } {'1/2' } 1 {'pi' } {'355/452'} 2 {'-3' } {'-3/4' } 2 {'eps' } {'1/2' } -51 {'realmax'} {'1' } 1024 {'realmin'} {'1/2' } -1021 ```

The results indicate that, for the first row, $1=\frac{1}{2}\left({2}^{1}\right)$. Similarly, for the fourth row, $\mathrm{eps}=\frac{1}{2}\left({2}^{-51}\right)$.

## Input Arguments

collapse all

Input matrix, specified as a scalar, vector, matrix, or multidimensional array.

For floating-point number dissection `[F,E] = log2(X)`, any zeros in `X` produce `F = 0` and `E = 0`. Input values of `Inf`, `-Inf`, or `NaN` are returned unchanged in `F` with a corresponding exponent of `E = 0`.

Data Types: `single` | `double`
Complex Number Support: Yes

## Output Arguments

collapse all

Base 2 logarithm values, returned as a scalar, vector, matrix, or multidimensional array of the same size as `X`.

Mantissa values, returned as a scalar, vector, matrix, or multidimensional array of the same size as `X`. The values in `F` and `E` satisfy ```X = F.*2.^E```.

Exponent values, returned as a scalar, vector, matrix, or multidimensional array of the same size as `X`. The values in `F` and `E` satisfy ```X = F.*2.^E```.

## Tips

• This function corresponds to the ANSI® C function `frexp()` and the IEEE® floating-point standard function `logb()`. Any zeros in `X` produce `F = 0` and `E = 0`.

## Extended Capabilities

### GPU Code GenerationGenerate CUDA® code for NVIDIA® GPUs using GPU Coder™.

Introduced before R2006a