[Matlab] MxArray 与 MwArray 使用区别

引子

在外部编程语言与matlab的交互中，Array是最单元的交互元素，怎么都绕不过去。

在matlab提供的Array接口有两个，一个是C的MxArray, 另一个是Cpp（C++)的MwArray.

看下两着的分别介绍：

mxArray：Matlab C 函数库的结构体
mwArray：Matlab C++ 函数库中对mxArray的包装类

声明：

mxArray:    mxArray *a;

mwArray：   mwArray a;

销毁

mxArray：   mxDestroyArray a;

mwArray：   mwArray类的析构函数自动销毁对象

变量传递

mxArray：  mxArray *dest_ptr =mxCreateDoubleMatrix(rows,cols, mxREAL);

          memcpy(dest_ptr,source_ptr,MAX_SIZE);

mwArray：   mwArray in1(rows, cols, mxDOUBLE_CLASS, mxREAL); 

          mwArray in2(rows, cols, mxDOUBLE_CLASS, mxREAL); 

          in1.SetData(data, rows*cols); 

          in2.SetData(data, rows*cols);

比较而言，  １。mwArray的声明更简洁，不用考虑指针   ２。mwArray不用手动释放内存

mxArray 介绍

mxArray *mxCreateDoubleMatrix(int m, int n, mxComplexity ComplexFlag); 

参数m和n为矩阵的函数和列数。ComplexFlag为常数，用来区分矩阵中元素是实数还是复数，取值分别为mxREAL和mxCOMPLEX。

类似的创建函数还有：

mxArray *mxCreateString(const char *str);  创建一个字符串类型并初始化为str字符串。

对应的，要删除一个数组mxDestroyArray，该函数声明如下： 

void mxDestroyArray(mxArray *array_ptr);

要获得mxArray数组每一维上元素的个数，可以用mxGetM和mxGetN函数。其中mxGetM用来获得数组第一维的元素个数，对于矩阵来说就是行数。 

int mxGetM(const mxArray *array_ptr); //返回array_ptr对应数组第一维的元素个数（行数）

int mxGetN(const mxArray *array_ptr); //返回array_ptr对应数组其它维的元素个数，对于矩阵来说是列数。对于多维数组来说是从第2维到最后一维的各维元素个数的乘积。 

要获得某一特定维的元素个数，则要用函数：

const int *mxGetDimensions(const mxArray *array_ptr); 

该函数返回array_ptr各维的元素个数保存在一个int数组中返回。对于常用的矩阵来说，用mxGetM和mxGetN两个函数就可以了。

另外还可以通过mxGetNumberOfDimensions来获得数组的总的维数，用mxSetM、mxSetN设置矩阵的行数和列数，函数说明如下：

int mxGetNumberOfDimensions(const mxArray *array_ptr); //返回数组的维数

void mxSetM(mxArray *array_ptr, int m); //设置数组为m行

void mxSetN(mxArray *array_ptr, int n); //设置数组为n列

在对 mxArray 类型的变量进行操作之前，可以验证以下其中的数组的数据类型，比如是否为 double 数组、整数、字符串、逻辑值等，以及是否为某种结构、类、或者是特殊类型，比如是否为空数组，是否为 inf 、 NaN 等。常见的判断函数有：

Use these functions to validate input arguments.

C Functions

mxIsDouble	Determine whether mxArray represents data as double-precision, floating-point numbers
mxIsSingle	Determine whether array represents data as single-precision, floating-point numbers
mxIsComplex	Determine whether data is complex
mxIsNumeric	Determine whether array is numeric
mxIsInt64	Determine whether array represents data as signed 64-bit integers
mxIsUint64	Determine whether array represents data as unsigned 64-bit integers
mxIsInt32	Determine whether array represents data as signed 32-bit integers
mxIsUint32	Determine whether array represents data as unsigned 32-bit integers
mxIsInt16	Determine whether array represents data as signed 16-bit integers
mxIsUint16	Determine whether array represents data as unsigned 16-bit integers
mxIsInt8	Determine whether array represents data as signed 8-bit integers
mxIsUint8	Determine whether array represents data as unsigned 8-bit integers
mxIsScalar	Determine whether array is scalar array

mxIsChar

Determine whether input is mxChar array

mxIsLogical	Determine whether array is of type mxLogical
mxIsLogicalScalar	Determine whether scalar array is of type mxLogical
mxIsLogicalScalarTrue	Determine whether scalar array of type mxLogical is true

mxIsStruct

Determine whether input is structure array

mxIsCell

Determine whether input is cell array

mxIsClass

Determine whether array is member of specified class

mxIsInf	Determine whether input is infinite
mxIsFinite	Determine whether input is finite
mxIsNaN	Determine whether input is NaN (Not-a-Number)
mxIsEmpty	Determine whether array is empty
mxIsSparse	Determine whether input is sparse array

mxIsFromGlobalWS

Determine whether array was copied from MATLAB global workspace

mxAssert	Check assertion value for debugging purposes
mxAssertS	Check assertion value without printing assertion text

对于常用的double类型的数组，可以用mxGetPr和mxGetPi两个函数分别获得其实部和虚部的数据指针，这两个函数的声明如下：

double *mxGetPr(const mxArray *array_ptr); //返回数组array_ptr的实部指针

double *mxGetPi(const mxArray *array_ptr); //返回数组array_ptr的虚部指针

Utilities for manipulating strings and structures.

C Functions

mxArrayToString	Array to string
mxArrayToUTF8String	Array to string in UTF-8 encoding
mxGetString	mxChar array to C-style string or Fortran character array

另外一种操作mxArray的方法（在mathworks上居然没有搜索到..... 囧）

[cpp]  view plain  copy

1. //为了调用matlab中的函数，必须使用数组数据类型，并其后调用matlab函数将其转化为矩阵格式（matlab的基本数据类型是矩阵）  
2.   
3. static double x1[1]={1.0};  
4. static double x2[1]={2.5};  
5.   
6. double result;  
7. //调用matlab创建3个矩阵  
8. mxArray *A=mclGetUninitializedArray();  
9. mxArray *B=mclGetUninitializedArray();  
10. mxArray *C=mclGetUninitializedArray();  
11.   
12. //将C语言中的变量值赋给matlab中的矩阵  
13. mlfAssign(&A,mlfDoubleMatrix(1,1,x1,NULL));  
14. mlfAssign(&B,mlfDoubleMatrix(1,1,x2,NULL));  
15. mlfAssign(&C,mlfMyfunct(A,B));               //调m函数  
16.   
17. //将matlab中的矩阵的指针传递给C语言中的指向double的指针  
18.   
19. double * md=mxGetPr(C);  
20.   
21. result=md[0];  
22.   
23. //释放这些矩阵  
24. mxDestroyArray(A);  
25. mxDestroyArray(B);  
26. mxDestroyArray(C);  

C++ Utility Classes

• mwArray	Class used to pass input/output arguments to C functions generated by MATLAB Compiler SDK
  mwException	Exception type used by the mwArray API and the C++ interface functions
  mwString	String class used by the mwArray API to pass string data as output from certain methods

mwArray 介绍

构造函数Constructors

mwArray()    Description    创建空的Matlab阵列,类型为mxDOUBLE_CLASS

mwArray(mxClassID mxID)  Description  创建mxID指定类型的Matlab阵列

Arguments

mxClassID mxID

Valid mxClassID specifying the type of array to construct. See the Work with mxArrays for more information on mxClassID.

mwArray(mwSize num_rows, mwSize num_cols, mxClassID mxID, mxComplexity cmplx = mxREAL)

Description  创建行数为num_rows，列数为num_cols，类型为mxID的Matalb阵列，对于数值型阵列,将complx做为最后一个参数,确定待创建阵列是否为复数阵列

Arguments

mwSize num_rows	Number of rows in the array
mwSize num_cols	Number of columns in the array
mxClassID mxID	Valid mxClassID specifying the type of array to construct. See the Work with mxArrays for more information on mxClassID.
mxComplexity cmplx	Complexity of the array to create. Valid values are mxREAL and mxCOMPLEX. The default value is mxREAL.

mwArray(mwSize num_dims, const mwSize* dims, mxClassID mxID, mxComplexity cmplx = mxREAL)

Description

创建任意维数的Matlab阵列，维数由num_dims指定，各维大小由dims指定，mxID指定阵列的类型。对于数值型阵列，将cmplx作为最后的一个参数，确定待创建阵列是否为复型的阵列。

All elements are initialized to zero. For cell arrays, all elements are initialized to empty cells.

Arguments

mwSize num_dims	Number of dimensions in the array
const mwSize* dims	Dimensions of the array
mxClassID mxID	Valid mxClassID specifying the type of array to construct. See the Work with mxArrays for more information on mxClassID.
mxComplexity cmplx	Complexity of the array to create. Valid values are mxREAL and mxCOMPLEX. The default value is mxREAL.

mwArray(const char* str)

Description

Create a 1-by-n array of type mxCHAR_CLASS, with n = strlen(str), and initialize the array's data with the characters in the supplied string.

根据字符串str创建一个新的字符型阵列

Arguments

const char* str

Null-terminated character buffer used to initialize the array

mwArray(mwSize num_strings, const char** str)

Description

创建字符型阵列(mxCHAR_CLASS)，字符串由str指定. The created array has dimensions m-by-max, where max is the length of the longest string in str.

Arguments

mwSize num_strings	Number of strings in the input array
const char** str	Array of null-terminated strings

mwArray(mwSize num_rows, mwSize num_cols, int num_fields, const char** fieldnames)

Description

Create a matrix of type mxSTRUCT_CLASS, with the specified field names. All elements are initialized with empty cells. 

创建行数为num_rows，列数为num_cols结构体阵列(mxSTRUCT_CLASS),  结构体域名为由fieldnames指定,域名个数由num_fields指定

Arguments

mwSize num_rows	Number of rows in the array
mwSize num_cols	Number of columns in the array
int num_fields	Number of fields in the struct matrix.
const char** fieldnames	Array of null-terminated strings representing the field names

mwArray(mwSize num_dims, const mwSize* dims, int num_fields, const char** fieldnames)

Description

Create an n-dimensional array of type mxSTRUCT_CLASS, with the specified field names. All elements are initialized with empty cells.

创建任意维数的结构体阵列，维数由num_dims指定，各维大小由dims指定，结构体域名由fieldnames指定，域名个数由num_fields指定.

Arguments

mwSize num_dims	Number of dimensions in the array
const mwSize* dims	Dimensions of the array
int num_fields	Number of fields in the struct matrix.
const char** fieldnames	Array of null-terminated strings representing the field names

mwArray(const mwArray& arr)

Description

Create a deep copy of an existing array.  根据当前的阵列arr中创建一个新的阵列(复制)

Arguments

mwArray& arr

mwArray to copy

mwArray(<type> re)

Description

Create a real scalar array.  创建一个新的数值阵列，实部为re.

The scalar array is created with the type of the input argument.

Arguments

<type> re

Scalar value to initialize the array. <type> can be any of the following: mxDouble\mxSingle\mxInt8\mxUint8\mxInt16\mxUint16 mxInt32\mxUint32\mxInt64\mxUint64\mxLogical

mwArray(<type> re, <type> im)

Description

Create a complex scalar array.  创建一个新的数值阵列，实部为re,虚部为im

The scalar array is created with the type of the input argument.

Arguments

<type> re	Scalar value to initialize the real part of the array
<type> im	Scalar value to initialize the imaginary part of the array

<type> can be any of the following:       mxDouble\mxSingle\mxInt8\mxUint8\mxInt16\mxUint16\mxInt32\mxUint32\mxInt64\mxUint64\mxLogical

Methods

mwArray Clone() const

Description

Create a new array representing deep copy of array.

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
mwArray b = a.Clone();

mwArray SharedCopy() const

Description

Create a shared copy of an existing array. The new array and the original array both point to the same data.

返回一个新的共享数据型mwArray阵列，此阵列与现有的mwArray阵列指向同一个数据块。

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
mwArray b = a.SharedCopy();

mwArray Serialize() const

Description

Serialize an array into bytes. A 1-by-n numeric matrix of type mxUINT8_CLASS is returned containing the serialized data. The data can be deserialized back into the original representation by calling mwArray::Deserialize().

将mwArray序列化一个新的阵列，新的阵列为mxUINT8_CLASS类型

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
mwArray b = a.Serialize();

mxClassID ClassID() const

Description

Determine the type of the array. See the Work with mxArrays for more information on mxClassID.

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
mxClassID id = a.ClassID();

int ElementSize() const

Description

Determine the size, in bytes, of an element of array type.

 返回mwArray阵列元素大小

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
int size = a.ElementSize();

size_t ElementSize() const

Description

Determine the size, in bytes, of an element of array type.

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
int size = a.ElementSize();

mwSize NumberOfElements() const

Description

Determine the total size of the array.

返回阵列中元素的个数

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
int n = a.NumberOfElements();

mwSize NumberOfNonZeros() const

Description

Determine the size of the of the array's data. If the underlying array is not sparse, this returns the same value as NumberOfElements().

返回稀疏阵列非零元素的个数

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
int n = a.NumberOfNonZeros();

mwSize MaximumNonZeros() const

Description

Determine the allocated size of the of the array's data. If the underlying array is not sparse, this returns the same value as NumberOfElements().

返回稀疏阵列中最大的元素的个数

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
int n = a.MaximumNonZeros();

mwSize NumberOfDimensions() const

Description

Determine the dimensionality of the array. 返回阵列维数

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
int n = a.NumberOfDimensions();

int NumberOfFields() const

Description

Determine the number of fields in a struct array. If the underlying array is not of type struct, zero is returned.

 返回结构体域个数

Example

const char* fields[] = {"a", "b", "c"};
mwArray a(2, 2, 3, fields);
int n = a.NumberOfFields();

mwString GetFieldName(int index)

Description

Determine the name of a given field in a struct array. If the underlying array is not of type struct, an exception is thrown.

Arguments

int index

Index of the field to name. Indexing starts at zero.

Example

const char* fields[] = {"a", "b", "c"};
mwArray a(2, 2, 3, fields);
mwString tempname = a.GetFieldName(1);
const char* name = (const char*)tempname;

mwArray GetDimensions() const

Description

Determine the size of each dimension in the array. The size of the returned array is 1-by-NumberOfDimensions().

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
mwArray dims = a.GetDimensions();

bool IsEmpty() const

Description

Determine if an array is empty.

Example

mwArray a;
bool b = a.IsEmpty();

bool IsSparse() const

Description

Determine if an array is sparse.

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
bool b = a.IsSparse();

bool IsNumeric() const

Description

Determine if an array is numeric.

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
bool b = a.IsNumeric();

bool IsComplex() const

Description

Determine if an array is complex.

Example

mwArray a(2, 2, mxDOUBLE_CLASS, mxCOMPLEX);
bool b = a.IsComplex();

bool Equals(const mwArray& arr) const

Description

Returns true if the input array is byte-wise equal to this array. This method makes a byte-wise comparison of the underlying arrays. Therefore, arrays of the same type should be compared. Arrays of different types will not in general be equal, even if they are initialized with the same data.

Arguments

mwArray& arr

Array to compare to array.

Example

mwArray a(1, 1, mxDOUBLE_CLASS);
mwArray b(1, 1, mxDOUBLE_CLASS);
a = 1.0;
b = 1.0;
bool c = a.Equals(b);

int CompareTo(const mwArray& arr) const

Description

Compares this array with the specified array for order. This method makes a byte-wise comparison of the underlying arrays. Therefore, arrays of the same type should be compared. Arrays of different types will, in general, not be ordered equivalently, even if they are initialized with the same data.

Arguments

mwArray& arr

Array to compare to array.

Example

mwArray a(1, 1, mxDOUBLE_CLASS);
mwArray b(1, 1, mxDOUBLE_CLASS);
a = 1.0;
b = 1.0;
int n = a.CompareTo(b);

int HashCode() const

Description

Constructs a unique hash value form the underlying bytes in the array. Therefore, arrays of different types will have different hash codes, even if they are initialized with the same data.

Example

mwArray a(1, 1, mxDOUBLE_CLASS);
int n = a.HashCode();

mwString ToString() const

Description

Returns a string representation of the underlying array. The string returned is the same string that is returned by typing a variable's name at the MATLAB command prompt.

Example

mwArray a(1, 1, mxDOUBLE_CLASS, mxCOMPLEX);
a.Real() = 1.0;
a.Imag() = 2.0;
printf("%s\n", (const char*)(a.ToString()));

mwArray RowIndex() const

Description

Returns an array of type mxINT32_CLASS representing the row indices (first dimension) of this array. For sparse arrays, the indices are returned for just the non-zero elements and the size of the array returned is 1-by-NumberOfNonZeros(). For nonsparse arrays, the size of the array returned is 1-by-NumberOfElements(), and the row indices of all of the elements are returned.

 返回阵列元素的行索引;对于稀疏阵列,只返回非零原素的行索引

Example

#include <stdio.h>
mwArray a(1, 1, mxDOUBLE_CLASS);
mwArray rows = a.RowIndex();

mwArray ColumnIndex() const

Description

Returns an array of type mxINT32_CLASS representing the column indices (second dimension) of this array. For sparse arrays, the indices are returned for just the non-zero elements and the size of the array returned is 1-by-NumberOfNonZeros(). For nonsparse arrays, the size of the array returned is 1-by-NumberOfElements(), and the column indices of all of the elements are returned.

返回阵列元素的列索引;对于稀疏阵列,只返回非零元素的列索引。

Example

mwArray a(1, 1, mxDOUBLE_CLASS);
mwArray rows = a.ColumnIndex();

void MakeComplex()

Description

Convert a numeric array that has been previously allocated as real to complex. If the underlying array is of a nonnumeric type, an mwException is thrown.

Example

double rdata[4] = {1.0, 2.0, 3.0, 4.0};
double idata[4] = {10.0, 20.0, 30.0, 40.0};
mwArray a(2, 2, mxDOUBLE_CLASS);
a.SetData(rdata, 4);
a.MakeComplex();
a.Imag().SetData(idata, 4);

mwArray Get(mwSize num_indices, ...)

Description

Fetches a single element at a specified index. The index is passed by first passing the number of indices followed by a comma-separated list of 1-based indices. The valid number of indices that can be passed in is either 1 (single subscript indexing), in which case the element at the specified 1-based offset is returned, accessing data in column-wise order, or NumberOfDimensions() (multiple subscript indexing), in which case, the index list is used to access the specified element. The valid range for indices is 1 <= index <= NumberOfElements(), for single subscript indexing. For multiple subscript indexing, the ith index has the valid range: 1 <= index[i] <= GetDimensions().Get(1, i). An mwException is thrown if an invalid number of indices is passed in or if any index is out of bounds.

根据索引返回阵列元素，其中num_indices表示索引数目。Get函数中输入的索引均从1起始。

Arguments

mwSize num_indices	Number of indices passed in
...	Comma-separated list of input indices. Number of items must equal num_indices but should not exceed 32.

Example

double data[4] = {1.0, 2.0, 3.0, 4.0};
double x;
mwArray a(2, 2, mxDOUBLE_CLASS);
a.SetData(data, 4);
x = a.Get(1,1);
x = a.Get(2, 1, 2);
x = a.Get(2, 2, 2);

mwArray Get(const char* name, mwSize num_indices, ...)

Description

Fetches a single element at a specified field name and index. This method may only be called on an array that is of type mxSTRUCT_CLASS. An mwException is thrown if the underlying array is not a struct array. The field name passed must be a valid field name in the struct array. The index is passed by first passing the number of indices followed by a comma-separated list of 1-based indices. The valid number of indices that can be passed in is either 1 (single subscript indexing), in which case the element at the specified 1-based offset is returned, accessing data in column-wise order, or NumberOfDimensions() (multiple subscript indexing), in which case, the index list is used to access the specified element. The valid range for indices is 1 <= index <= NumberOfElements(), for single subscript indexing. For multiple subscript indexing, the ith index has the valid range: 1 <= index[i] <= GetDimensions().Get(1, i). AnmwException is thrown if an invalid number of indices is passed in or if any index is out of bounds.

返回结构体域名为name，指定索引的结构体域，其中num_indices表示索引的数目。Get函数中输入的索引均从1起始。

Arguments

char* name	Null-terminated character buffer containing the name of the field
mwSize num_indices	Number of indices passed in
...	Comma-separated list of input indices. Number of items must equalnum_indices but should not exceed 32.

Example

const char* fields[] = {"a", "b", "c"};

mwArray a(1, 1, 3, fields);
mwArray b = a.Get("a", 1, 1);
mwArray b = a.Get("b", 2, 1, 1);

mwArray Get(mwSize num_indices, const mwIndex* index)

Description

Fetches a single element at a specified index. The index is passed by first passing the number of indices, followed by an array of 1-based indices. The valid number of indices that can be passed in is either 1 (single subscript indexing), in which case the element at the specified 1-based offset is returned, accessing data in column-wise order, or NumberOfDimensions() (multiple subscript indexing), in which case, the index list is used to access the specified element. The valid range for indices is 1 <= index <= NumberOfElements(), for single subscript indexing. For multiple subscript indexing, the ith index has the valid range: 1 <= index[i] <= GetDimensions().Get(1, i). An mwException is thrown if an invalid number of indices is passed in or if any index is out of bounds.

Arguments

mwSize num_indices	Size of index array
mwIndex* index	Array of at least size num_indices containing the indices

Example

double data[4] = {1.0, 2.0, 3.0, 4.0};
int index[2] = {1, 1};
double x;
mwArray a(2, 2, mxDOUBLE_CLASS);
a.SetData(data, 4);
x = a.Get(1, index);
x = a.Get(2, index);
index[0] = 2;
index[1] = 2;
x = a.Get(2, index);

mwArray Get(const char* name, mwSize num_indices, const mwIndex* index)

Description

Fetches a single element at a specified field name and index. This method may only be called on an array that is of type mxSTRUCT_CLASS. An mwException is thrown if the underlying array is not a struct array. The field name passed must be a valid field name in the struct array. The index is passed by first passing the number of indices followed by an array of 1-based indices. The valid number of indices that can be passed in is either 1 (single subscript indexing), in which case the element at the specified 1-based offset is returned, accessing data in column-wise order, or NumberOfDimensions() (multiple subscript indexing), in which case, the index list is used to access the specified element. The valid range for indices is 1 <= index <= NumberOfElements(), for single subscript indexing. For multiple subscript indexing, the ith index has the valid range: 1 <= index[i] <= GetDimensions().Get(1, i). An mwException is thrown if an invalid number of indices is passed in or if any index is out of bounds.

Arguments

char* name	Null-terminated character buffer containing the name of the field
mwSize num_indices	Number of indices passed in
mwIndex* index	Array of at least size num_indices containing the indices

Example

const char* fields[] = {"a", "b", "c"};
int index[2] = {1, 1};
mwArray a(1, 1, 3, fields);
mwArray b = a.Get("a", 1, index);
mwArray b = a.Get("b", 2, index);

mwArray Real()

Description

Accesses the real part of a complex array. The returned mwArray is considered real and has the same dimensionality and type as the original.

Complex arrays consist of Complex numbers, which are 1 X 2 vectors (pairs). For example, if the number is 3+5i, then the pair is (3,5i). An array of Complex numbers is therefore two dimensional (N X 2), where N is the number of complex numbers in the array. 2+4i, 7-3i, 8+6i would be represented as (2,4i) (7,3i) (8,6i). Complex numbers have two components, real and imaginary.

The MATLAB function Realcan be applied to an array of Complex numbers. It extracts the corresponding part of the Complex number. For example,REAL(3,5i) == 3.

返回数值阵列的实部

Example

double rdata[4] = {1.0, 2.0, 3.0, 4.0};
double idata[4] = {10.0, 20.0, 30.0, 40.0};
mwArray a(2, 2, mxDOUBLE_CLASS, mxCOMPLEX);
a.Real().SetData(rdata, 4);

mwArray Imag()

Description

Accesses the imaginary part of a complex array. The returned mwArray is considered real and has the same dimensionality and type as the original.

Complex arrays consist of Complex numbers, which are 1 X 2 vectors (pairs). For example, if the number is 3+5i, then the pair is (3,5i). An array of Complex numbers is therefore two dimensional (N X 2), where N is the number of complex numbers in the array. 2+4i, 7-3i, 8+6i would be represented as (2,4i) (7,3i) (8,6i). Complex numbers have two components, real and imaginary.

The MATLAB function Imag can be applied to an array of Complex numbers. It extracts the corresponding part of the Complex number. For example,IMAG(3+5i) == 5. Imag returns 5 in this case and not 5i. Imag returns the magnitude of the imaginary part of the number as a real number.

返回数值阵列虚部

Example

double rdata[4] = {1.0, 2.0, 3.0, 4.0};
double idata[4] = {10.0, 20.0, 30.0, 40.0};
mwArray a(2, 2, mxDOUBLE_CLASS, mxCOMPLEX);
a.Imag().SetData(idata, 4);

void Set(const mwArray& arr)

Description

Assign shared copy of input array to currently referenced cell for arrays of type mxCELL_CLASS and mxSTRUCT_CLASS.

Arguments

mwArray& arr

mwArray to assign to currently referenced cell

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
mwArray b(2, 2, mxINT16_CLASS);
mwArray c(1, 2, mxCELL_CLASS);
c.Get(1,1).Set(a);
c.Get(1,2).Set(b);

void GetData(<numeric-type>* buffer, mwSize len) const

Description

Copies the array's data into supplied numeric buffer.

The data is copied in column-major order. If the underlying array is not of the same type as the input buffer, the data is converted to this type as it is copied. If a conversion cannot be made, an mwException is thrown.

Arguments

<numeric-type>* buffer	Buffer to receive copy. Valid types for <numeric-type> are: mxDOUBLE_CLASS  /  mxSINGLE_CLASS /  mxINT8_CLASS mxUINT8_CLASS  /  mxINT16_CLASS  /  mxUINT16_CLASS mxINT32_CLASS  /  mxUINT32_CLASS /  mxINT64_CLASS mxUINT64_CLASS
mwSize len	Maximum length of buffer. A maximum of len elements will be copied.

Example

double rdata[4] = {1.0, 2.0, 3.0, 4.0};
double data_copy[4] ;
mwArray a(2, 2, mxDOUBLE_CLASS);
a.SetData(rdata, 4);
a.GetData(data_copy, 4);

void GetLogicalData(mxLogical* buffer, mwSize len) const

Description

Copies the array's data into supplied mxLogical buffer.

The data is copied in column-major order. If the underlying array is not of type mxLOGICAL_CLASS, the data is converted to this type as it is copied. If a conversion cannot be made, an mwException is thrown.

Arguments

mxLogical* buffer	Buffer to receive copy
mwSize len	Maximum length of buffer. A maximum of len elements will be copied.

Example

mxLogical data[4] = {true, false, true, false};
mxLogical data_copy[4] ;
mwArray a(2, 2, mxLOGICAL_CLASS);
a.SetLogicalData(data, 4);
a.GetLogicalData(data_copy, 4);

void GetCharData(mxChar* buffer, mwSize len) const

Description

Copies the array's data into supplied mxChar buffer.

The data is copied in column-major order. If the underlying array is not of type mxCHAR_CLASS, the data is converted to this type as it is copied. If a conversion cannot be made, an mwException is thrown.

Arguments

mxChar** buffer	Buffer to receive copy
mwSize len	Maximum length of buffer. A maximum of len elements will be copied.

Example

mxChar data[6] = {'H', 'e' , `l' , 'l' , 'o' , '\0'};
mxChar data_copy[6] ;
mwArray a(1, 6, mxCHAR_CLASS);
a.SetCharData(data, 6);
a.GetCharData(data_copy, 6);

void SetData(<numeric-type>* buffer, mwSize len) const

Description

Copies the data from supplied numeric buffer into the array.

The data is copied in column-major order. If the underlying array is not of the same type as the input buffer, the data is converted to this type as it is copied. If a conversion cannot be made, an mwException is thrown.

Arguments

<numeric-type>* buffer	Buffer containing data to copy. Valid types for <numeric-type> are: mxDOUBLE_CLASS / mxSINGLE_CLASS / mxINT8_CLASS mxUINT8_CLASS / mxINT16_CLASS / mxUINT16_CLASS mxINT32_CLASS / mxUINT32_CLASS / mxINT64_CLASS mxUINT64_CLASS
mwSize len	Maximum length of buffer. A maximum of len elements will be copied.

Example

double rdata[4] = {1.0, 2.0, 3.0, 4.0};
double data_copy[4] ;
mwArray a(2, 2, mxDOUBLE_CLASS);
a.SetData(rdata, 4);
a.GetData(data_copy, 4);

void SetLogicalData(mxLogical* buffer, mwSize len) const

Description

Copies the data from the supplied mxLogical buffer into the array.

The data is copied in column-major order. If the underlying array is not of type mxLOGICAL_CLASS, the data is converted to this type as it is copied. If a conversion cannot be made, an mwException is thrown.

Arguments

mxLogical* buffer	Buffer containing data to copy
mwSize len	Maximum length of buffer. A maximum of len elements will be copied.

Example

mxLogical data[4] = {true, false, true, false};
mxLogical data_copy[4] ;
mwArray a(2, 2, mxLOGICAL_CLASS);
a.SetLogicalData(data, 4);
a.GetLogicalData(data_copy, 4);

void SetCharData(mxChar* buffer, mwSize len) const

Description

Copies the data from the supplied mxChar buffer into the array.

The data is copied in column-major order. If the underlying array is not of type mxCHAR_CLASS, the data is converted to this type as it is copied. If a conversion cannot be made, an mwException is thrown.

Arguments

mxChar** buffer	Buffer containing data to copy
mwSize len	Maximum length of buffer. A maximum of len elements will be copied.

Example

mxChar data[6] = {'H', 'e' , `l' , 'l' , 'o' , '\0'};
mxChar data_copy[6] ;
mwArray a(1, 6, mxCHAR_CLASS);
a.SetCharData(data, 6);
a.GetCharData(data_copy, 6);

static mwArray Deserialize(const mwArray& arr)

Description

Deserializes an array that has been serialized with mwArray::Serialize(). The input array must be of type mxUINT8_CLASS and contain the data from a serialized array. If the input data does not represent a serialized mwArray, the behavior of this method is undefined.

Arguments

mwArray& arr

mwArray that has been obtained by calling mwArray::Serialize

Example

double rdata[4] = {1.0, 2.0, 3.0, 4.0};
mwArray a(1,4,mxDOUBLE_CLASS); 
a.SetData(rdata, 4);
mwArray b = a.Serialize();
a = mwArray::Deserialize(b);

static mwArray NewSparse(mwSize rowindex_size, const mwIndex* rowindex, mwSize colindex_size, const mwIndex* colindex, mwSize data_size, const mxDouble* rData, mwSize num_rows, mwSize num_cols, mwSize nzmax)

Description

Creates real sparse matrix of type double with specified number of rows and columns.

The lengths of input row, column index, and data arrays must all be the same or equal to 1. In the case where any of these arrays are equal to 1, the value is repeated throughout the construction of the matrix.

If any element of the rowindex or colindex array is greater than the specified values in num_rows or num_cols respectively, an exception is thrown.

Arguments

mwSize rowindex_size	Size of rowindex array
mwIndex* rowindex	Array of row indices of non-zero elements
mwSize colindex_size	Size of colindex array
mwIndex* colindex	Array of column indices of non-zero elements
mwSize data_size	Size of data array
mxDouble* rData	Data associated with non-zero row and column indices
mwSize num_rows	Number of rows in matrix
mwSize num_cols	Number of columns in matrix
mwSize nzmax	Reserved storage for sparse matrix. If nzmax is zero, storage will be set tomax{rowindex_size, colindex_size, data_size}.

Example

This example constructs a sparse 4 X 4 tridiagonal matrix:

2 -1  0  0
-1  2 -1  0
0 -1  2 -1
0  0 -1  2

The following code, when run:

double rdata[]        = 
           {2.0, -1.0, -1.0, 2.0, -1.0, 
            -1.0, 2.0, -1.0, -1.0, 2.0};
mwIndex row_tridiag[] = 
           {1,    2,    1,   2,    3,
            2,   3,    4,    3,   4  };
mwIndex col_tridiag[] = 
           {1,    1,    2,   2,    2, 
              3,   3,    3,    4,   4  };

mwArray mysparse = 
           mwArray::NewSparse(10, row_tridiag, 
                              10, col_tridiag, 
                              10, rdata, 4, 4, 10);
std::cout << mysparse << std::endl;

will display the following output to the screen:

 (1,1)        2
 (2,1)       -1
 (1,2)       -1
 (2,2)        2
 (3,2)       -1
 (2,3)       -1
 (3,3)        2
 (4,3)       -1
 (3,4)       -1
 (4,4)        2

static mwArray NewSparse(mwSize rowindex_size, const mwIndex* rowindex, mwSize colindex_size, const mwIndex* colindex, mwSize data_size, const mxDouble* rdata, mwSize nzmax)

Description

Creates real sparse matrix of type double with number of rows and columns inferred from input data.

The lengths of input row and column index and data arrays must all be the same or equal to 1. In the case where any of these arrays are equal to 1, the value is repeated through out the construction of the matrix.

The number of rows and columns in the created matrix are calculated form the input rowindex and colindex arrays as num_rows = max{rowindex}, num_cols = max{colindex}.

Arguments

mwSize rowindex_size	Size of rowindex array
mwIndex* rowindex	Array of row indices of non-zero elements
mwSize colindex_size	Size of colindex array
mwIndex* colindex	Array of column indices of non-zero elements
mwSize data_size	Size of data array
mxDouble* rData	Data associated with non-zero row and column indices
mwSize nzmax	Reserved storage for sparse matrix. If nzmax is zero, storage will be set tomax{rowindex_size, colindex_size, data_size}.

Example

In this example, we construct a sparse 4 X 4 identity matrix. The value of 1.0 is copied to each non-zero element defined by row and column index arrays:

double one = 1.0;
mwIndex row_diag[] = {1, 2, 3, 4};
mwIndex col_diag[] = {1, 2, 3, 4};

mwArray mysparse = 
  mwArray::NewSparse(4, row_diag, 
                     4, col_diag, 
                     1, &one, 
                     0);
std::cout << mysparse << std::endl;

(1,1)        1
(2,2)        1
(3,3)        1
(4,4)        1

static mwArray NewSparse(mwSize rowindex_size, const mwIndex* rowindex, mwSize colindex_size, const mwIndex* colindex, mwSize data_size, const mxDouble* rdata, const mxDouble* idata, mwSize num_rows, mwSize num_cols, mwSize nzmax)

Description

Creates complex sparse matrix of type double with specified number of rows and columns.

The lengths of input row and column index and data arrays must all be the same or equal to 1. In the case where any of these arrays are equal to 1, the value is repeated through out the construction of the matrix.

If any element of the rowindex or colindex array is greater than the specified values in num_rows, num_cols, respectively, then an exception is thrown.

Arguments

mwSize rowindex_size	Size of rowindex array
mwIndex* rowindex	Array of row indices of non-zero elements
mwSize colindex_size	Size of colindex array
mwIndex* colindex	Array of column indices of non-zero elements
mwSize data_size	Size of data array
mxDouble* rData	Real part of data associated with non-zero row and column indices
mxDouble* iData	Imaginary part of data associated with non-zero row and column indices
mwSize num_rows	Number of rows in matrix
mwSize num_cols	Number of columns in matrix
mwSize nzmax	Reserved storage for sparse matrix. If nzmax is zero, storage will be set tomax{rowindex_size, colindex_size, data_size}.

Example

This example constructs a complex tridiagonal matrix:

double rdata[] =   {2.0,  -1.0,   -1.0,  2.0,  -1.0,  -1.0,  2.0,  -1.0,   -1.0,  2.0};
double idata[] =   {20.0, -10.0, -10.0, 20.0, -10.0, -10.0, 20.0, -10.0,  -10.0, 20.0};
mwIndex row_tridiag[] =   {1,    2,    1,   2,    3,    2,   3,    4,    3,   4};
mwIndex col_tridiag[] =   {1,    1,    2,   2,    2,    3,   3,    3,    4,   4};

mwArray mysparse = mwArray::NewSparse(10, row_tridiag, 
                                      10, col_tridiag, 
                                      10, rdata, 
                                      idata, 4, 4, 10);
std::cout << mysparse << std::endl;

It displays the following output to the screen:

(1,1)      2.0000 +20.0000i
(2,1)     -1.0000 -10.0000i
(1,2)     -1.0000 -10.0000i
(2,2)      2.0000 +20.0000i
(3,2)     -1.0000 -10.0000i
(2,3)     -1.0000 -10.0000i
(3,3)      2.0000 +20.0000i
(4,3)     -1.0000 -10.0000i
(3,4)     -1.0000 -10.0000i
(4,4)      2.0000 +20.0000i

static mwArray NewSparse(mwSize rowindex_size, const mwIndex* rowindex, mwSize colindex_size, const mwIndex* colindex, mwSize data_size, const mxDouble* rdata, const mxDouble* idata, mwSize nzmax)

Description

Creates complex sparse matrix of type double with number of rows and columns inferred from input data.

The lengths of input row and column index and data arrays must all be the same or equal to 1. In the case where any of these arrays are equal to 1, the value is repeated through out the construction of the matrix.

The number of rows and columns in the created matrix are calculated form the input rowindex and colindex arrays as num_rows = max{rowindex}, num_cols = max{colindex}.

Arguments

mwSize rowindex_size	Size of rowindex array
mwIndex* rowindex	Array of row indices of non-zero elements
mwSize colindex_size	Size of colindex array
mwIndex* colindex	Array of column indices of non-zero elements
mwSize data_size	Size of data array
mxDouble* rData	Real part of data associated with non-zero row and column indices
mxDouble* iData	Imaginary part of data associated with non-zero row and column indices
mwSize nzmax	Reserved storage for sparse matrix. If nzmax is zero, storage will be set tomax{rowindex_size, colindex_size, data_size}.

Example

This example constructs a complex matrix by inferring dimensions and storage allocation from the input data.

mwArray mysparse = 
   mwArray::NewSparse(10, row_tridiag, 
                      10, col_tridiag,
                      10, rdata, idata, 
                      0);
std::cout << mysparse << std::endl;

(1,1)      2.0000 +20.0000i
(2,1)     -1.0000 -10.0000i
(1,2)     -1.0000 -10.0000i
(2,2)      2.0000 +20.0000i
(3,2)     -1.0000 -10.0000i
(2,3)     -1.0000 -10.0000i
(3,3)      2.0000 +20.0000i
(4,3)     -1.0000 -10.0000i
(3,4)     -1.0000 -10.0000i
(4,4)      2.0000 +20.0000i

static mwArray NewSparse(mwSize rowindex_size, const mwIndex* rowindex, mwSize colindex_size, const mwIndex* colindex, mwSize data_size, const mxLogical* rdata, mwSize num_rows, mwSize num_cols, mwSize nzmax)

Description

Creates logical sparse matrix with specified number of rows and columns.

The lengths of input row and column index and data arrays must all be the same or equal to 1. In the case where any of these arrays are equal to 1, the value is repeated throughout the construction of the matrix.

If any element of the rowindex or colindex array is greater than the specified values in num_rows, num_cols, respectively, then an exception is thrown.

Arguments

mwSize rowindex_size	Size of rowindex array
mwIndex* rowindex	Array of row indices of non-zero elements
mwSize colindex_size	Size of colindex array
mwIndex* colindex	Array of column indices of non-zero elements
mwSize data_size	Size of data array
mxLogical* rData	Data associated with non-zero row and column indices
mwSize num_rows	Number of rows in matrix
mwSize num_cols	Number of columns in matrix
mwSize nzmax	Reserved storage for sparse matrix. If nzmax is zero, storage will be set tomax{rowindex_size, colindex_size, data_size}.

Example

This example creates a sparse logical 4 X 4 tridiagonal matrix, assigning true to each non-zero value:

mxLogical one = true;
mwIndex row_tridiag[] = {1,    2,    1,   2,    3,    2,   3,    4,    3,   4};
mwIndex col_tridiag[] = {1,    1,    2,   2,    2,    3,   3,    3,    4,   4};

mwArray mysparse = 
      mwArray::NewSparse(10, row_tridiag,
                         10, col_tridiag, 
                          1, &one, 
                          4, 4, 10);
std::cout << mysparse << std::endl;

(1,1)        1
(2,1)        1
(1,2)        1
(2,2)        1
(3,2)        1
(2,3)        1
(3,3)        1
(4,3)        1
(3,4)        1
(4,4)        1

static mwArray NewSparse(mwSize rowindex_size, const mwIndex* rowindex, mwSize colindex_size, const mwIndex* colindex, mwSize data_size, const mxLogical* rdata, mwSize nzmax)

Description

Creates logical sparse matrix with number of rows and columns inferred from input data.

The lengths of input row and column index and data arrays must all be the same or equal to 1. In the case where any of these arrays are equal to 1, the value is repeated through out the construction of the matrix.

The number of rows and columns in the created matrix are calculated form the input rowindex and colindex arrays as num_rows = max {rowindex}, num_cols = max {colindex}.

Arguments

mwSize rowindex_size	Size of rowindex array
mwIndex* rowindex	Array of row indices of non-zero elements
mwSize colindex_size	Size of colindex array
mwIndex* colindex	Array of column indices of non-zero elements
mwSize data_size	Size of data array
mxLogical* rData	Data associated with non-zero row and column indices
mwSize nzmax	Reserved storage for sparse matrix. If nzmax is zero, storage will be set tomax{rowindex_size, colindex_size, data_size}.

Example

This example uses the data from the first example, but allows the number of rows, number of columns, and allocated storage to be calculated from the input data:

mwArray mysparse = 
    mwArray::NewSparse(10, row_tridiag, 
                       10, col_tridiag, 
                       1, &one, 
                       0);
std::cout << mysparse << std::endl;

(1,1)        1
(2,1)        1
(1,2)        1
(2,2)        1
(3,2)        1
(2,3)        1
(3,3)        1
(4,3)        1
(3,4)        1
(4,4)        1

static mwArray NewSparse (mwSize num_rows, mwSize num_cols, mwSize nzmax, mxClassID mxID, mxComplexity cmplx = mxREAL)

Description

Creates an empty sparse matrix. All elements in an empty sparse matrix are initially zero, and the amount of allocated storage for non-zero elements is specified by nzmax.

Arguments

mwSize num_rows	Number of rows in matrix
mwSize num_cols	Number of columns in matrix
mwSize nzmax	Reserved storage for sparse matrix
mxClassID mxID	Type of data to store in matrix. Currently, sparse matrices of type doubleprecision and logical are supported. Pass mxDOUBLE_CLASS to create a doubleprecision sparse matrix. Pass mxLOGICAL_CLASS to create a logical sparse matrix.
mxComplexity cmplx	Complexity of matrix. Pass mxCOMPLEX to create a complex sparse matrix andmxREAL to create a real sparse matrix. This argument may be omitted, in which case the default complexity is real

Example

This example constructs a real 3 X 3 empty sparse matrix of type double with reserved storage for 4 non-zero elements:

mwArray mysparse = mwArray::NewSparse(3, 3, 4, mxDOUBLE_CLASS);
std::cout << mysparse << std::endl;

All zero sparse: 3-by-3

static double GetNaN()

Description

Get value of NaN (Not-a-Number).

Call mwArray::GetNaN to return the value of NaN for your system. NaN is the IEEE arithmetic representation for Not-a-Number. Certain mathematical operations return NaN as a result, for example:

• 0.0/0.0

• Inf-Inf

The value of NaN is built in to the system; you cannot modify it.

Example

double x = mwArray::GetNaN();

static double GetEps()

Description

Returns the value of the MATLAB eps variable. This variable is the distance from 1.0 to the next largest floating-point number. Consequently, it is a measure of floating-point accuracy. The MATLAB pinv and rank functions use eps as a default tolerance.

Example

double x = mwArray::GetEps();

static double GetInf()

Description

Returns the value of the MATLAB internal Inf variable. Inf is a permanent variable representing IEEE arithmetic positive infinity. The value of Inf is built into the system; you cannot modify it.

Operations that return Inf include

• Division by 0. For example, 5/0 returns Inf.

• Operations resulting in overflow. For example, exp(10000) returns Inf because the result is too large to be represented on your machine.

Example

double x = mwArray::GetInf();

static bool IsFinite(double x)

Description

Determine whether or not a value is finite. A number is finite if it is greater than -Inf and less than Inf.

Arguments

doulbe x

Value to test for finiteness

Example

bool x = mwArray::IsFinite(1.0);

static bool IsInf(double x)

Description

Determines whether or not a value is equal to infinity or minus infinity. MATLAB stores the value of infinity in a permanent variable named Inf, which represents IEEE arithmetic positive infinity. The value of the variable, Inf, is built into the system; you cannot modify it.

Operations that return infinity include

• Division by 0. For example, 5/0 returns infinity.

• Operations resulting in overflow. For example, exp(10000) returns infinity because the result is too large to be represented on your machine. If the value equals NaN (Not-a-Number), then mxIsInf returns false. In other words, NaN is not equal to infinity.

Arguments

doulbe x

Value to test for infiniteness

Example

bool x = mwArray::IsInf(1.0);

static bool IsNaN(double x)

Description

Determines whether or not the value is NaN. NaN is the IEEE arithmetic representation for Not-a-Number. NaN is obtained as a result of mathematically undefined operations such as

• 0.0/0.0

• Inf-Inf

The system understands a family of bit patterns as representing NaN. In other words, NaN is not a single value, rather it is a family of numbers that the MATLAB software (and other IEEE-compliant applications) use to represent an error condition or missing data.

Arguments

doulbe x

Value to test for NaN

Example

bool x = mwArray::IsNaN(1.0);

Operators

mwArray operator()(mwIndex i1, mwIndex i2, mwIndex i3, ..., )

Description

Fetches a single element at a specified index. The index is passed as a comma-separated list of 1-based indices. This operator is overloaded to support 1 through 32 indices. The valid number of indices that can be passed in is either 1 (single subscript indexing), in which case the element at the specified 1-based offset is returned, accessing data in column-wise order, or NumberOfDimensions() (multiple subscript indexing), in which case, the index list is used to access the specified element. The valid range for indices is 1 <= index <= NumberOfElements(), for single subscript indexing. For multiple subscript indexing, the ith index has the valid range: 1 <= index[i] <= GetDimensions().Get(1, i). An mwException is thrown if an invalid number of indices is passed in or if any index is out of bounds.

Arguments

mwIndex i1, mwIndex i2, mwIndex i3, ...,

Comma-separated list of input indices

Example

double data[4] = {1.0, 2.0, 3.0, 4.0};
double x;
mwArray a(2, 2, mxDOUBLE_CLASS);
a.SetData(data, 4);
x = a(1,1);
x = a(1,2);
x = a(2,2);

mwArray operator()(const char* name, mwIndex i1, mwIndex i2, mwIndex i3, ..., )

Description

Fetches a single element at a specified field name and index. This method may only be called on an array that is of type mxSTRUCT_CLASS. An mwException is thrown if the underlying array is not a struct array. The field name passed must be a valid field name in the struct array. The index is passed by first passing the number of indices, followed by an array of 1-based indices. This operator is overloaded to support 1 through 32 indices. The valid number of indices that can be passed in is either 1 (single subscript indexing), in which case the element at the specified 1-based offset is returned, accessing data in column-wise order, or NumberOfDimensions() (multiple subscript indexing), in which case, the index list is used to access the specified element. The valid range for indices is 1 <= index <= NumberOfElements(), for single subscript indexing. For multiple subscript indexing, the ith index has the valid range:1 <= index[i] <= GetDimensions().Get(1, i). An mwException is thrown if an invalid number of indices is passed in or if any index is out of bounds.

Arguments

char* name	Null terminated string containing the field name to get
mwIndex i1, mwIndex i2, mwIndex i3, ...,	Comma-separated list of input indices

Example

const char* fields[] = {"a", "b", "c"};
int index[2] = {1, 1};
mwArray a(1, 1, 3, fields);
mwArray b = a("a", 1, 1);
mwArray b = a("b", 1, 1);

mwArray& operator=(const <type>& x)

Description

Sets a single scalar value. This operator is overloaded for all numeric and logical types.

Arguments

const <type>& x

Value to assign

Example

mwArray a(2, 2, mxDOUBLE_CLASS);
a(1,1) = 1.0;
a(1,2) = 2.0;
a(2,1) = 3.0;
a(2,2) = 4.0;

operator <type>() const

Description

Fetches a single scalar value. This operator is overloaded for all numeric and logical types.

Example

double data[4] = {1.0, 2.0, 3.0, 4.0};
double x;
mwArray a(2, 2, mxDOUBLE_CLASS);
a.SetData(data, 4);
x = (double)a(1,1);
x = (double)a(1,2);
x = (double)a(2,1);
x = (double)a(2,2);