revise the comment

source/tensor/core/arithmetic/Div.cpp

@@ -168,7 +168,7 @@ int GetDivDimIndex(const XTensor &a, const XTensor &b)
 }
 
 /*
-element-wise division of two tensors (return a XTensor structure)
+element-wise division of two tensors (return an XTensor structure)
 make a new tensor c to keep the result and return it
 
 c(i) = a(i)*b(i)

source/tensor/core/arithmetic/Div.h

@@ -42,7 +42,7 @@ where i is the index of the element
 void _DivMe(XTensor * a, const XTensor * b, DTYPE alpha = 0.0, int leadingDim = 0);
 
 /* 
-element-wise division of two tensors (return a XTensor structure)
+element-wise division of two tensors (return an XTensor structure)
 make a new tensor to keep the result and return it
 c(i) = a(i)/b(i)
 where i is the index of the element 

source/tensor/core/arithmetic/DivDim.cpp

@@ -133,7 +133,7 @@ void _DivDim(XTensor * a, const XTensor * b, int n, DTYPE alpha)
 }
     
 /*
-tensor division of two tensors (return a XTensor structure and make tensor connections)
+tensor division of two tensors (return an XTensor structure and make tensor connections)
 make a new tensor to keep the result and return it
 
 c = a/b + \alpha * c

source/tensor/core/arithmetic/MatrixMul.cpp

@@ -203,7 +203,7 @@ void _MatrixMul(const XTensor * a, MATRIX_TRANS_TYPE transposedA,
 }
 
 /*
-matrix multiplication (return a XTensor structure) c = trans(a) * trans(b) * alpha
+matrix multiplication (return an XTensor structure) c = trans(a) * trans(b) * alpha
 make a new tensor to keep the result and return it
 
 For the input tensors a and b, we perform matrix multiplication on the first two dimentsions. 

source/tensor/core/arithmetic/MatrixMul.h

@@ -44,7 +44,7 @@ void _MatrixMul(const XTensor * a, MATRIX_TRANS_TYPE transposedA, const XTensor
                 DTYPE alpha = (DTYPE)1.0, DTYPE beta = 0, XPRunner * parallelRunner = NULL);
 
 /* 
-matrix multiplication (return a XTensor structure) c = trans(a) * trans(b) * alpha
+matrix multiplication (return an XTensor structure) c = trans(a) * trans(b) * alpha
 make a new tensor c to keep the result and return it
 
 For the input tensors a and b, we perform matrix multiplicationon the first two dimentsions. 

source/tensor/core/arithmetic/MatrixMulBatched.h

@@ -62,7 +62,7 @@ void _MatrixMulBatchedCPU(const XList * a, MATRIX_TRANS_TYPE transposedA, const
                           XList * c, DTYPE alpha = (DTYPE)1.0, DTYPE beta = 0);
 
 /*
-matrix multiplication of the two tensors (return a XTensor structure) c = trans(a) * trans(b) * alpha
+matrix multiplication of the two tensors (return an XTensor structure) c = trans(a) * trans(b) * alpha
 make a new tensor to keep the result and return it
 
 for each 2-dimensional data array in a (denoted as ai) and
@@ -74,7 +74,7 @@ XTensor MatrixMulBatched(const XTensor &a, MATRIX_TRANS_TYPE transposedA, const
                          DTYPE alpha = (DTYPE)1.0, XPRunner * parallelRunner = NULL);
 
 /*
-matrix multiplication of the two tensors (return a XTensor structure) c = a * b * alpha
+matrix multiplication of the two tensors (return an XTensor structure) c = a * b * alpha
 make a new tensor to keep the result and return it
 
 for each 2-dimensional data array in a (denoted as ai) and

source/tensor/core/arithmetic/Multiply.cpp

@@ -169,7 +169,7 @@ int GetMultiplyDimIndex(const XTensor &a, const XTensor &b)
 }
 
 /*
-element-wise product of two tensors (return a XTensor structure)
+element-wise product of two tensors (return an XTensor structure)
 make a new tensor c to keep the result and return it
 
 c(i) = a(i)*b(i)

source/tensor/core/arithmetic/Multiply.h

@@ -42,7 +42,7 @@ where i is the index of the element
 void _MultiplyMe(XTensor * a, const XTensor * b, DTYPE alpha = 0.0, int leadingDim = 0);
 
 /* 
-element-wise product of two tensors (return a XTensor structure)
+element-wise product of two tensors (return an XTensor structure)
 make a new tensor to keep the result and return it
 c(i) = a(i)*b(i)
 where i is the index of the element 

source/tensor/core/arithmetic/MultiplyDim.cpp

@@ -132,7 +132,7 @@ void _MultiplyDimMe(XTensor * a, const XTensor * b, int n, DTYPE alpha)
 }
 
 /*
-tensor multiplication (return a XTensor structure and make tensor connections)
+tensor multiplication (return an XTensor structure and make tensor connections)
 make a new tensor to keep the result and return it
 
 c = a * b + \alpha * c

source/tensor/core/arithmetic/Negate.cpp

@@ -60,7 +60,7 @@ void _NegateMe(XTensor * a)
 }
 
 /*
-set every entry to its minus value (return a XTensor structure)
+set every entry to its minus value (return an XTensor structure)
 make a new tensor to keep the result and return it
 >> a - input tensor we are processing
 << return - the minus value of input tensor

source/tensor/core/arithmetic/Negate.h

@@ -36,7 +36,7 @@ keep the result in the input tensor a and return nothing
 void _NegateMe(XTensor * a);
 
 /* 
-set every entry to its minus value (return a XTensor structure)
+set every entry to its minus value (return an XTensor structure)
 make a new tensor to keep the result and return it
 */
 XTensor Negate(const XTensor & a);

source/tensor/core/arithmetic/Sign.cpp

@@ -66,7 +66,7 @@ void _SignMe(XTensor * a)
 }
 
 /*
-set every entry to its sign value (return a XTensor structure)
+set every entry to its sign value (return an XTensor structure)
 make a new tensor to keep the result and return it
 >> a - input tensor we are processing
 << return - the sign value of the input tensor

source/tensor/core/arithmetic/Sign.h

@@ -36,7 +36,7 @@ keep the result in the input tensor a and return nothing
 void _SignMe(XTensor * a);
 
 /* 
-set every entry to its sign value  (return a XTensor structure)
+set every entry to its sign value  (return an XTensor structure)
 make a new tensor to keep the result and return it
 */
 XTensor Sign(const XTensor & a);

source/tensor/core/arithmetic/Sub.cpp

@@ -155,7 +155,7 @@ int GetSubDimIndex(const XTensor &a, const XTensor &b)
 }
 
 /*
-tensor subtraction c = a - b * \beta (return a XTensor structure)
+tensor subtraction c = a - b * \beta (return an XTensor structure)
 make a new tensor c to keep the result and return it
 
 >> a - a tensor

source/tensor/core/arithmetic/SubDim.cpp

@@ -134,7 +134,7 @@ void _SubDim(XTensor * a, const XTensor * b, int n, DTYPE beta)
 }
 
 /*
-tensor subtraction (return a XTensor structure and make tensor connections)
+tensor subtraction (return an XTensor structure and make tensor connections)
 make a new tensor to keep the result and return it
 
 c = a - b * \beta

source/tensor/core/arithmetic/Sum.cpp

@@ -160,7 +160,7 @@ int GetSumDimIndex(const XTensor &a, const XTensor &b)
 }
     
 /*
-tensor summation c = a + b * \beta (return a XTensor structure)
+tensor summation c = a + b * \beta (return an XTensor structure)
 make a new tensor c to keep the result and return it
 
 >> a - a tensor

source/tensor/core/arithmetic/SumDim.cpp

@@ -134,7 +134,7 @@ void _SumDim(XTensor * a, const XTensor * b, int n, DTYPE beta)
 }
     
 /*
-tensor summation (return a XTensor structure and make tensor connections)
+tensor summation (return an XTensor structure and make tensor connections)
 make a new tensor to keep the result and return it
 
 c = a + b * \beta

source/tensor/core/getandset/Select.cpp

@@ -79,7 +79,7 @@ void _SelectRange(const XTensor * a, XTensor * c, int dim, int low, int high)
 }
 
 /* 
-generate a tensor with selected data in range[low,high] along the given dimension (return a XTensor structure)
+generate a tensor with selected data in range[low,high] along the given dimension (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 c = select(a) 

source/tensor/core/getandset/Select.h

@@ -42,7 +42,7 @@ c = select(a)
 void _SelectRange(const XTensor * a, XTensor * c, int dim, int low, int high);
 
 /* 
-generate a tensor with selected data in range[low,high] along the given dimension (return a XTensor structure)
+generate a tensor with selected data in range[low,high] along the given dimension (return an XTensor structure)
 make a new tensor to keep the result and return it
 c = select(a)
 */

source/tensor/core/math/Clip.cpp

@@ -71,7 +71,7 @@ void _ClipMe(XTensor * a, DTYPE lower, DTYPE upper)
 }
 
 /*
-set every entry to its clip value (return a XTensor structure)
+set every entry to its clip value (return an XTensor structure)
 make a new tensor to keep the result and return it
 >> a - input tensor we are processing
 >> lower - the lower border

source/tensor/core/math/Clip.h

@@ -34,7 +34,7 @@ void _Clip(const XTensor * a, XTensor * b, DTYPE lower, DTYPE upper);
    keep the result in the input tensor a and return nothing */
 void _ClipMe(XTensor * a, DTYPE lower, DTYPE upper);
 
-/* set every entry to its clip value  (return a XTensor structure)
+/* set every entry to its clip value  (return an XTensor structure)
    make a new tensor to keep the result and return it */
 XTensor Clip(const XTensor & a, DTYPE lower, DTYPE upper);
 

source/tensor/core/math/Normalize.cpp

@@ -114,7 +114,7 @@ void _NormalizeMe(XTensor * input, int dim, const XTensor * mean, const XTensor
     _Normalize(input, input, dim, mean, var, a, b, epsilon);
 }
 /*
-normalized the data with normal distribution (return a XTensor structure)
+normalized the data with normal distribution (return an XTensor structure)
 make a new tensor to keep the result and return it 
 
 For an input x, y = a * (x-mean)/sqrt(variance+\epsilon) + b

source/tensor/core/math/Normalize.h

@@ -42,7 +42,7 @@ where a and b are the scalar and bias respectively, and \epsilon is the adjustme
 void _NormalizeMe(XTensor * input, int dim, const XTensor * mean, const XTensor * var, const XTensor * a, const XTensor * b, DTYPE epsilon);
 
 /*
-normalized the data with normal distribution (return a XTensor structure)
+normalized the data with normal distribution (return an XTensor structure)
 make a new tensor to keep the result and return it 
 For an input x, y = a * (x-mean)/sqrt(variance+\epsilon) + b
 where a and b are the scalar and bias respectively, and \epsilon is the adjustment parameter.

source/tensor/core/math/Power.cpp

@@ -81,7 +81,7 @@ void _PowerMe(XTensor * a, DTYPE p)
 }
 
 /*
-get the power(a, p) (return a XTensor structure)
+get the power(a, p) (return an XTensor structure)
 make a new tensor to keep the result and return it
 >> a - input tensor
 >> p - parameter

source/tensor/core/math/Power.h

@@ -36,7 +36,7 @@ keep the result in the input tensor a and return nothing
 void _PowerMe(XTensor * a, DTYPE p);
 
 /* 
-get the power(x, y) (return a XTensor structure)
+get the power(x, y) (return an XTensor structure)
 make a new tensor to keep the result and return it
 */
 XTensor Power(const XTensor & a, DTYPE p);

source/tensor/core/math/ScaleAndShift.cpp

@@ -92,7 +92,7 @@ void _ScaleAndShiftMe(XTensor * a, DTYPE scale, DTYPE shift)
 }
 
 /* 
-scale and shift all tensor entires (return a XTensor structure)
+scale and shift all tensor entires (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 b = a * scale + shift

source/tensor/core/math/Unary.h

@@ -32,7 +32,7 @@ void _Absolute(const XTensor * a, XTensor * b);
 /* set every entry to its absolute value (do it on site)
 keep the result in the input tensor a and return nothing */
 void _AbsoluteMe(XTensor * a);
-/* set every entry to its absolute value (return a XTensor structure)
+/* set every entry to its absolute value (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor Absolute(const XTensor & a);
 
@@ -41,7 +41,7 @@ void _Ceil(const XTensor * a, XTensor * b);
 /* set every entry to its ceil value (do it on site)
 keep the result in the input tensor a and return nothing */
 void _CeilMe(XTensor * a);
-/* set every entry to its ceil value (return a XTensor structure)
+/* set every entry to its ceil value (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor Ceil(const XTensor & a);
 
@@ -50,7 +50,7 @@ void _Exp(const XTensor * a, XTensor * b);
 /* set every entry to its exponent value (do it on site)
 keep the result in the input tensor a and return nothing */
 void _ExpMe(XTensor * a);
-/* set every entry to its exponent value (return a XTensor structure)
+/* set every entry to its exponent value (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor Exp(const XTensor & a);
 
@@ -59,7 +59,7 @@ void _Floor(const XTensor * a, XTensor * b);
 /* set every entry to its floor value (do it on site)
 keep the result in the input tensor a and return nothing */
 void _FloorMe(XTensor * a);
-/* set every entry to its floor value (return a XTensor structure)
+/* set every entry to its floor value (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor Floor(const XTensor & a);
 
@@ -68,7 +68,7 @@ void _IsNonZero(const XTensor *a, XTensor *b);
 /* if source entry is non-zero, set target entry to be one, otherwise zero (do it on site)
 keep the result in the input tensor a and return nothing */
 void _IsNonZeroMe(XTensor *a);
-/* if source entry is non-zero, set target entry to be one, otherwise zero (return a XTensor structure)
+/* if source entry is non-zero, set target entry to be one, otherwise zero (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor IsNonZero(const XTensor &a);
 
@@ -77,7 +77,7 @@ void _IsZero(const XTensor *a, XTensor *b);
 /* if source entry is zero, set target entry to be one, otherwise zero (do it on site)
 keep the result in the input tensor a and return nothing */
 void _IsZeroMe(XTensor *a);
-/* if source entry is zero, set target entry to be one, otherwise zero (return a XTensor structure)
+/* if source entry is zero, set target entry to be one, otherwise zero (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor IsZero(const XTensor &a);
 
@@ -86,7 +86,7 @@ void _Log(const XTensor * a, XTensor * b);
 /* set every entry to its logarithm value (do it on site)
 keep the result in the input tensor a and return nothing */
 void _LogMe(XTensor * a);
-/* set every entry to its logarithm value (return a XTensor structure)
+/* set every entry to its logarithm value (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor Log(const XTensor & a);
 
@@ -95,7 +95,7 @@ void _Round(const XTensor * a, XTensor * b);
 /* set every entry to its round value (do it on site)
 keep the result in the input tensor a and return nothing */
 void _RoundMe(XTensor * a);
-/* set every entry to its round value (return a XTensor structure)
+/* set every entry to its round value (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor Round(const XTensor & a);
 
@@ -104,7 +104,7 @@ void _Sqrt(const XTensor * a, XTensor * b);
 /* set every entry to its sqrt value (do it on site)
 keep the result in the input tensor a and return nothing */
 void _SqrtMe(XTensor * a);
-/* set every entry to its sqrt value (return a XTensor structure)
+/* set every entry to its sqrt value (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor Sqrt(const XTensor & a);
 
@@ -113,7 +113,7 @@ void _Square(const XTensor * a, XTensor * b);
 /* set every entry to its square value (do it on site)
 keep the result in the input tensor a and return nothing */
 void _SquareMe(XTensor * a);
-/* set every entry to its square value (return a XTensor structure)
+/* set every entry to its square value (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor Square(const XTensor & a);
 
@@ -123,7 +123,7 @@ void _Sin(const XTensor * a, XTensor * b);
 /* set every entry to its sine value (do it on site)
 keep the result in the input tensor a and return nothing */
 void _SinMe(XTensor * a);
-/* set every entry to its sine value (return a XTensor structure)
+/* set every entry to its sine value (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor Sin(const XTensor & a);
 
@@ -132,7 +132,7 @@ void _Cos(const XTensor * a, XTensor * b);
 /* set every entry to its cosine value (do it on site)
 keep the result in the input tensor a and return nothing */
 void _CosMe(XTensor * a);
-/* set every entry to its cosine value (return a XTensor structure)
+/* set every entry to its cosine value (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor Cos(const XTensor & a);
 
@@ -141,7 +141,7 @@ void _Tan(const XTensor * a, XTensor * b);
 /* set every entry to its tangent value (do it on site)
 keep the result in the input tensor a and return nothing */
 void _TanMe(XTensor * a);
-/* set every entry to its tangent value (return a XTensor structure)
+/* set every entry to its tangent value (return an XTensor structure)
 make a new tensor to keep the result and return it */
 XTensor Tan(const XTensor & a);
 

source/tensor/core/movement/CopyIndexed.cpp

@@ -99,7 +99,7 @@ void _CopyIndexed(const XTensor * s, XTensor * t, int dim, int * srcIndex, int i
 }
 
 /*
-copy indexed sub-tensors (return a XTensor structure)
+copy indexed sub-tensors (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 >> s - the source tensor

source/tensor/core/movement/CopyIndexed.h

@@ -33,13 +33,13 @@ void _CopyIndexed(const XTensor * s, XTensor * t, int dim, int * srcIndex, int i
 void _CopyIndexed(const XTensor * s, XTensor * t, int dim, const XTensor * srcIndex, const XTensor * tgtIndex);
 
 /* 
-copy selected sub-tensors (return a XTensor structure)
+copy selected sub-tensors (return an XTensor structure)
 make a new tensor to keep the result and return it (remove this???)
 */
 XTensor CopyIndexed(const XTensor &s, int dim, int * srcIndex, int indexSize, int * tgtIndex, int copyNum);
     
 /*
-copy selected sub-tensors where indeces are kept in tensors (return a XTensor structure)
+copy selected sub-tensors where indeces are kept in tensors (return an XTensor structure)
 make a new tensor to keep the result and return it (remove this???)
 */
 void CopyIndexed(const XTensor * s, XTensor * t, int dim, const XTensor * srcIndex, const XTensor * tgtIndex);

source/tensor/core/movement/CopyValues.cpp

@@ -98,7 +98,7 @@ void _CopyValues(const XTensor * s, const int sBeg, const int sLen, XTensor * t,
 }
 
 /*
-copy s to t (return a XTensor structure)
+copy s to t (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 >> s - source

source/tensor/core/movement/CopyValues.h

@@ -33,7 +33,7 @@ void _CopyValues(const XTensor * s, XTensor * t, XStream * stream = NULL);
 void _CopyValues(const XTensor * s, const int sBeg, const int sLen, XTensor * t, const int tBeg, XStream * stream = NULL);
 
 /* 
-copy s to t (return a XTensor structure)
+copy s to t (return an XTensor structure)
 make a new tensor to keep the result and return it
 */
 XTensor CopyValues(const XTensor &s, XStream * stream = NULL);

source/tensor/core/movement/Gather.cpp

@@ -89,7 +89,7 @@ void _Gather(XTensor * s, XTensor * t, XTensor * srcIndex)
 }
 
 /*
-gather indexed sub-tensors (return a XTensor structure)
+gather indexed sub-tensors (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 >> s - the source tensor(2D)

source/tensor/core/movement/Gather.h

@@ -32,7 +32,7 @@ void _Gather(XTensor * s, XTensor * t, int dim, int * srcIndex, int indexSize);
 /* gather selected sub-tensors */
 void _Gather(XTensor * s, XTensor * t, XTensor * srcIndex);
 
-/* gather selected sub-tensors (return a XTensor structure)
+/* gather selected sub-tensors (return an XTensor structure)
    make a new tensor to keep the result and return it */
 XTensor Gather(XTensor &s, XTensor &index);
 

source/tensor/core/movement/Spread.h

@@ -30,7 +30,7 @@ namespace nts { // namespace nts(NiuTrans.Tensor)
 void _Spread(XTensor * source, XTensor * collection, int dim, 
              int * srcIndex, int indexSize, int * collIndex);
 
-/* spread a collection tensor to source tensor (return a XTensor structure)
+/* spread a collection tensor to source tensor (return an XTensor structure)
    make a new tensor to keep the result and return it */
 void Spread(XTensor * source, XTensor * collection, 
             XTensor * srcIndex, XTensor * collIndex,

source/tensor/core/reduce/ReduceMax.cpp

@@ -94,7 +94,7 @@ void _ReduceMax(const XTensor * input, XTensor * output, int dim)
 }
 
 /* 
-get the max value of the items along a dimension of the tensor (return a XTensor structure).
+get the max value of the items along a dimension of the tensor (return an XTensor structure).
 make a new tensor to keep the result and return it
 
 >> input - the input tensor

source/tensor/core/reduce/ReduceMax.h

@@ -30,7 +30,7 @@ namespace nts{ // namespace nts(NiuTrans.Tensor)
 void _ReduceMax(const XTensor * input, XTensor * output, int dim);
 
 /* 
-get the max value of the items along a dimension of the tensor (return a XTensor structure)
+get the max value of the items along a dimension of the tensor (return an XTensor structure)
 make a new tensor to keep the result and return it
 */
 XTensor ReduceMax(const XTensor &input, int dim);

source/tensor/core/reduce/ReduceMean.cpp

@@ -47,7 +47,7 @@ void _ReduceMean(const XTensor * input, XTensor * output, int dim)
 }
 
 /* 
-get the mean value along a dimension of the tensor (return a XTensor structure)
+get the mean value along a dimension of the tensor (return an XTensor structure)
 make a new tenosr to keep the result and return it
 
 For a 1-dimensional data array a, mean = (1/n) * sum_i input_i

source/tensor/core/reduce/ReduceMean.h

@@ -33,7 +33,7 @@ For a 1-dimensional data array a, mean = (1/n) * sum_i input_i
 void _ReduceMean(const XTensor * input, XTensor * output, int dim);
 
 /* 
-get the mean value along a dimension of the tensor (return a XTensor structure)
+get the mean value along a dimension of the tensor (return an XTensor structure)
 make a new tenosr to keep the result and return it
 For a 1-dimensional data array a, mean = (1/n) * sum_i input_i
 */

source/tensor/core/reduce/ReduceSum.cpp

@@ -198,7 +198,7 @@ void _ReduceSum(const XTensor * input, XTensor * output, int dim, const XTensor
 }
 
 /* 
-sum the items along a dimension of the tensor (return a XTensor structure)
+sum the items along a dimension of the tensor (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 For a 1-dimensional data array a,
@@ -245,7 +245,7 @@ XTensor ReduceSum(const XTensor &input, int dim, const XTensor &shift, DTYPE pow
 }
 
 /* 
-sum the items along a dimension of the tensor (return a XTensor structure)
+sum the items along a dimension of the tensor (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 For a 1-dimensional data array a,

source/tensor/core/reduce/ReduceSum.h

@@ -36,7 +36,7 @@ void _ReduceSum(const XTensor * input, XTensor * output, int dim, const XTensor
                 DTYPE power = (DTYPE)1.0F, bool isExp = false);
 
 /* 
-sum the items along a dimension of the tensor (return a XTensor structure)
+sum the items along a dimension of the tensor (return an XTensor structure)
 make a new tensor to keep the result and return it
 For a 1-dimensional data array a,
 sum = \sum_i (a_i - shift) if isExp == false
@@ -45,7 +45,7 @@ sum = \sum_i exp(a_i - shift) if isExp == true
 XTensor ReduceSum(const XTensor &input, int dim, const XTensor &shift, DTYPE power = (DTYPE)1.0F, bool isExp = false);
 
 /* 
-sum the items along a dimension of the tensor (return a XTensor structure)
+sum the items along a dimension of the tensor (return an XTensor structure)
 make a new tensor to keep the result and return it
 For a 1-dimensional data array a,
 sum = \sum_i (a_i) if isExp == false

source/tensor/core/reduce/ReduceSumSquared.cpp

@@ -42,7 +42,7 @@ void _ReduceSumSquared(const XTensor * input, XTensor * output, int dim, const X
 
 
 /* 
-squared sum of the items along a dimension of the tensor (return a XTensor structure)
+squared sum of the items along a dimension of the tensor (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 For a 1-dimensional data array a, sum = \sum_i (a_i - shift)^2

source/tensor/core/reduce/ReduceSumSquared.h

@@ -34,7 +34,7 @@ sum = \sum_i (a_i - shift)^2
 void _ReduceSumSquared(const XTensor * input, XTensor * output, int dim, const XTensor * shift);
 
 /* 
-squared sum of the items along a dimension of the tensor (return a XTensor structure)
+squared sum of the items along a dimension of the tensor (return an XTensor structure)
 make a new tensor to keep the result and return it
 For a 1-dimensional data array a, sum = \sum_i (a_i - shift)^2
 */

source/tensor/core/reduce/ReduceVariance.cpp

@@ -45,7 +45,7 @@ void _ReduceVariance(const XTensor * input, XTensor * output, int dim, const XTe
 }
 
 /* 
-variance of the items along a dimension of the tensor (return a XTensor structure)
+variance of the items along a dimension of the tensor (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 For a 1-dimensional data array a, variance = 1/n * \sum_i (a_i - mean)^2

source/tensor/core/reduce/ReduceVariance.h

@@ -33,7 +33,7 @@ For a 1-dimensional data array a, variance = 1/n * \sum_i (a_i - mean)^2
 void _ReduceVariance(const XTensor * input, XTensor * output, int dim, const XTensor * mean);
 
 /* 
-variance of the items along a dimension of the tensor (return a XTensor structure)
+variance of the items along a dimension of the tensor (return an XTensor structure)
 make a new tensor to keep the result and return it
 For a 1-dimensional data array a, variance = 1/n * \sum_i (a_i - mean)^2
 */

source/tensor/core/shape/Concatenate.cpp

@@ -55,7 +55,7 @@ void _Concatenate(const XList * smalls, XTensor * big, int dim)
 }
 
 /*
-concatenate a list of tensors along a given dimension (return a XTensor structure)
+concatenate a list of tensors along a given dimension (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 Note that this is actually a wrapper that selects "ConcatenateSolely"
@@ -155,7 +155,7 @@ void _Concatenate(const XTensor * smallA, const XTensor * smallB, XTensor * big,
 }
 
 /*
-concatenate two tensors along a given dimension (return a XTensor structure).
+concatenate two tensors along a given dimension (return an XTensor structure).
 make a new tensor to keep the result and return it.
 
 >> smallA - one tensor for concatenation

source/tensor/core/shape/Concatenate.h

@@ -34,7 +34,7 @@ Note that this is actually a wrapper that selects
 void _Concatenate(const XList * smalls, XTensor * big, int dim);
 
 /*
-concatenate a list of tensors along a given dimension (return a XTensor structure)
+concatenate a list of tensors along a given dimension (return an XTensor structure)
 make a new tensor to keep the result and return it
 Note that this is actually a wrapper that selects 
 "ConcatenateSolely" or "Merge" by means of the tensor shapes 
@@ -45,7 +45,7 @@ XTensor Concatenate(const XList &smalls, int dim);
 void _Concatenate(const XTensor * smallA, const XTensor * smallB, XTensor * big, int dim);
 
 /* 
-concatenate two tensors along a given dimension (return a XTensor structure)
+concatenate two tensors along a given dimension (return an XTensor structure)
 make a new tensor to keep the result and return it
 */
 XTensor Concatenate(const XTensor &smallA, const XTensor &smallB, int dim);

source/tensor/core/shape/Merge.cpp

@@ -149,7 +149,7 @@ void _Merge(const XTensor * s, XTensor * t, int whereToMerge, int leadingDim)
 }
 
 /*
-transform a tensor by merging it along with a dimension (return a XTensor structure)
+transform a tensor by merging it along with a dimension (return an XTensor structure)
 make a new tensor to keep the result and  return it
 
 e.g., (N/3, M, 3) -> (N, M)
@@ -315,7 +315,7 @@ void _Merge(const XList * smalls, XTensor * big, int whereToMerge)
 }
 
 /*
-merge small tensors into a big tensor (return a XTensor structure)
+merge small tensors into a big tensor (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 >> smalls - the list of the small tensors
@@ -352,7 +352,7 @@ XTensor Merge(const XList &smalls, int whereToMerge)
 }
 
 /* 
-merge two tensors into a big tensor (return a XTensor structure) 
+merge two tensors into a big tensor (return an XTensor structure) 
 >> smalls - the list of the small tensors
 >> whereToMerge - the merging operation is along with which dimension
 << return - the big tensor merged by small tensors

source/tensor/core/shape/Merge.h

@@ -29,17 +29,17 @@ namespace nts { // namespace nts(NiuTrans.Tensor)
 /* transform a tensor by merging it alone with a dimension, e.g., (M, N/3, 3) -> (M, N) */
 void _Merge(const XTensor * s, XTensor * t, int whereToMerge, int leadingDim = -1);
 
-/* transform a tensor by merging it alone with a dimension (return a XTensor structure)
+/* transform a tensor by merging it alone with a dimension (return an XTensor structure)
    e.g., (M, N/3, 3) -> (M, N) */
 XTensor Merge(const XTensor &s, int whereToMerge, int leadingDim = -1);
 
 /* merge small tensors into a big tensor */
 void _Merge(const XList * smalls, XTensor * big, int whereToMerge);
 
-/* merge small tensors into a big tensor (return a XTensor structure) */
+/* merge small tensors into a big tensor (return an XTensor structure) */
 XTensor Merge(const XList &smalls, int whereToMerge);
 
-/* merge two tensors into a big tensor (return a XTensor structure) */
+/* merge two tensors into a big tensor (return an XTensor structure) */
 XTensor Merge(const XTensor &smallA, const XTensor &smallB, int whereToMerge);
 
 } // namespace nts(NiuTrans.Tensor)

source/tensor/core/shape/Permute.h

@@ -42,7 +42,7 @@ a = permuted(a)
 void _PermuteMe(XTensor * a, int * dimPermute);
 
 /* 
-make a tensor with permuted dimensions (return a XTensor structure).
+make a tensor with permuted dimensions (return an XTensor structure).
 make a new tensor to keep the result and return it.
 b = permuted(a)
 */

source/tensor/core/shape/Split.cpp

@@ -157,7 +157,7 @@ void _Split(const XTensor * s, XTensor * t, int whereToSplit, int splitNum)
 }
 
 /*
-transform a tensor by splitting it, e.g., (N, M) -> (N/3, M, 3) (return a XTensor structure)
+transform a tensor by splitting it, e.g., (N, M) -> (N/3, M, 3) (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 >> s - the source tensor

source/tensor/core/shape/Split.h

@@ -35,7 +35,7 @@ e.g., (M, N) -> (M, N/3, 3)
 void _Split(const XTensor * s, XTensor * t, int whereToSplit, int splitNum);
 
 /* 
-transform a tensor by splitting it (return a XTensor structure)
+transform a tensor by splitting it (return an XTensor structure)
 make a new tensor to keep the result and return it
 e.g., (M, N) -> (M, N/3, 3) 
 */

source/tensor/core/shape/Squeeze.cpp

@@ -89,7 +89,7 @@ void _SqueezeMe(XTensor * source, int leadingDim)
 }
 
 /*
-squeeze the tensor along the specified dimension (return a XTensor structure)
+squeeze the tensor along the specified dimension (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 >> source - the input tensor

source/tensor/core/shape/Squeeze.h

@@ -33,7 +33,7 @@ void _Squeeze(XTensor * source, XTensor * target, int leadingDim = -1);
    keep the result in the input tensor a and return nothing */
 void _SqueezeMe(XTensor * source, int leadingDim = -1);
 
-/* squeeze the tensor along the specified dimension  (return a XTensor structure)
+/* squeeze the tensor along the specified dimension  (return an XTensor structure)
    make a new tensor to keep the result and return it */
 XTensor Squeeze(XTensor & source, int leadingDim = -1);
 

source/tensor/core/shape/Transpose.cpp

@@ -107,7 +107,7 @@ void _Transpose(const XTensor * a, XTensor * b, const int i, const int j)
 }
 
 /*
-tensor transposition of dimensions i and j (return a XTensor structure).
+tensor transposition of dimensions i and j (return an XTensor structure).
 make a new tensor to keep the result and return it.
 b = transposed(a)
 

source/tensor/core/shape/Transpose.h

@@ -34,7 +34,7 @@ b = transposed(a)
 void _Transpose(const XTensor * a, XTensor * b, const int i, const int j);
 
 /* 
-tensor transposition of dimensions i and j (return a XTensor structure).
+tensor transposition of dimensions i and j (return an XTensor structure).
 make a new tensor to keep the result and return it.
 b = transposed(a)
 */

source/tensor/core/shape/Unsqueeze.h

@@ -31,7 +31,7 @@ namespace nts { // namespace nts(NiuTrans.Tensor)
 void _Unsqueeze(const XTensor * a, XTensor * b, int dim, int dSize);
 
 /* insert a dimension by copying the blocks for x times 
-  (where x is the size of the inerted dimension) (return a XTensor structure)
+  (where x is the size of the inerted dimension) (return an XTensor structure)
    make a new tensor to keep the result and return it */
 XTensor Unsqueeze(const XTensor &a, int dim, int dSize);
 

source/tensor/core/sort/Sort.cpp

@@ -95,7 +95,7 @@ void _SortMe(XTensor * a, XTensor * index, int dim)
 }
 
 /*
-sort the tensor along a given dimension (return a XTensor structure)
+sort the tensor along a given dimension (return an XTensor structure)
 make a new tensor to keep the result and return it
 
 >> a - input tensor

source/tensor/core/sort/Sort.h

@@ -36,7 +36,7 @@ keep the result in the input tensor a and return nothing
 void _SortMe(XTensor * a, XTensor * index, int dim);
 
 /* 
-sort the data along a given dimension (return a XTensor structure)
+sort the data along a given dimension (return an XTensor structure)
 make a new tensor to keep the result and return it
 */
 void Sort(XTensor & a, XTensor & b, XTensor & index, int dim);

source/tensor/function/HardTanH.cpp

@@ -61,7 +61,7 @@ void _HardTanH(const XTensor * x, XTensor * y)
 }
 
 /* 
-hard tanh function (return a XTensor structure) 
+hard tanh function (return an XTensor structure) 
 make a new tensor to keep the result and return it
 
 y =  1    if x > 1

source/tensor/function/HardTanH.h

@@ -37,7 +37,7 @@ y =  1    if x > 1
 */
 void _HardTanH(const XTensor * x, XTensor * y);
 
-/* hard tanh function (return a XTensor structure) */
+/* hard tanh function (return an XTensor structure) */
 XTensor HardTanH(const XTensor &x);
 
 /* de/dx */

source/tensor/function/Identity.cpp

@@ -38,7 +38,7 @@ void _Identity(const XTensor * x, XTensor * y)
 }
 
 /* 
-identity function y = x (return a XTensor structure) 
+identity function y = x (return an XTensor structure) 
 make a new tensor to keep the result and return it
 
 >> x - input tensor

source/tensor/function/Identity.h

@@ -30,7 +30,7 @@ namespace nts{ // namespace nts(NiuTrans.Tensor)
 /* identity function y = x */
 void _Identity(const XTensor * x, XTensor * y);
 
-/* identity function y = x (return a XTensor structure) */
+/* identity function y = x (return an XTensor structure) */
 XTensor Identity(const XTensor &x);
 
 /* de/dx */

source/tensor/function/LogSoftmax.cpp

@@ -168,7 +168,7 @@ void _LogSoftmax(const XTensor * x, XTensor * y, int leadDim)
 }
 
 /*
-log scale softmax y = log(e^x / \sum_{i} e^{x_i}) (return a XTensor structure) 
+log scale softmax y = log(e^x / \sum_{i} e^{x_i}) (return an XTensor structure) 
 make a new tensor to keep the result and return it
 
 >> x - input vector

source/tensor/function/LogSoftmax.h

@@ -30,7 +30,7 @@ namespace nts{ // namespace nts(NiuTrans.Tensor)
 /* log scale softmax y = log(e^x / \sum_{i} e^{x_i}) */
 void _LogSoftmax(const XTensor * x, XTensor * y, int leadDim);
 
-/* log scale softmax y = log(e^x / \sum_{i} e^{x_i}) (return a XTensor structure) */
+/* log scale softmax y = log(e^x / \sum_{i} e^{x_i}) (return an XTensor structure) */
 XTensor LogSoftmax(const XTensor &x, int leadDim);
 
 /* log scale softmax y = log(e^x / \sum_{i} e^{x_i}) (with both argument of x and y) */

source/tensor/function/Rectify.cpp

@@ -57,7 +57,7 @@ void _Rectify(const XTensor * x, XTensor * y)
 }
 
 /*
-rectify function y = max(0, x) (return a XTensor structure) 
+rectify function y = max(0, x) (return an XTensor structure) 
 make a new tensor to keep the result and return it
 
 >> input - input tensor

source/tensor/function/Rectify.h

@@ -30,7 +30,7 @@ namespace nts{ // namespace nts(NiuTrans.Tensor)
 /* rectify function y = max(0, x) */
 void _Rectify(const XTensor * x, XTensor * y);
 
-/* rectify function y = max(0, x) (return a XTensor structure) */
+/* rectify function y = max(0, x) (return an XTensor structure) */
 XTensor Rectify(const XTensor &x);
 
 /* de/dx */

source/tensor/function/Sigmoid.cpp

@@ -55,7 +55,7 @@ void _Sigmoid(const XTensor * x, XTensor * y)
 }
 
 /*
-sigmoid function y = 1/(1+exp(-x)) (return a XTensor structure) 
+sigmoid function y = 1/(1+exp(-x)) (return an XTensor structure) 
 make a new tensor to keep the result and return it
 
 >> x - input tensor

source/tensor/function/Sigmoid.h

@@ -30,7 +30,7 @@ namespace nts{ // namespace nts(NiuTrans.Tensor)
 /* sigmoid function y = 1/(1+exp(-x)) */
 void _Sigmoid(const XTensor * x, XTensor * y);
 
-/* sigmoid function y = 1/(1+exp(-x)) (return a XTensor structure) */
+/* sigmoid function y = 1/(1+exp(-x)) (return an XTensor structure) */
 XTensor Sigmoid(const XTensor &x);
 
 /* de/dx */

source/tensor/function/Softmax.cpp

@@ -122,7 +122,7 @@ void _Softmax(const XTensor * x, XTensor * y, int leadDim)
 }
 
 /*
-softmax y = e^x / \sum_{i} e^{x_i} (return a XTensor structure) 
+softmax y = e^x / \sum_{i} e^{x_i} (return an XTensor structure) 
 make a new tensor to keep the result and return it
 
 >> x - input vector

source/tensor/function/Softmax.h

@@ -30,7 +30,7 @@ namespace nts{ // namespace nts(NiuTrans.Tensor)
 /* softmax y = e^x / \sum_{i} e^{x_i} */
 void _Softmax(const XTensor * x, XTensor * y, int leadDim);
 
-/* softmax y = e^x / \sum_{i} e^{x_i} (return a XTensor structure) */
+/* softmax y = e^x / \sum_{i} e^{x_i} (return an XTensor structure) */
 XTensor Softmax(const XTensor &x, int leadDim);
 
 /* de/dx */