update Sum function for easy use

source/Main.cpp

@@ -52,9 +52,14 @@ int main( int argc, const char ** argv )
         a.Set2D(1.0F, 0, 0);
         a.Set2D(1.0F, 1, 1);
 
-        b = Sum(a, a);
+        b = Sum(a, Sum(a, a));
+        XTensor c = b;
+        
+        a.Dump(stderr, "a: ");
         b.Dump(stderr, "b: ");
     }
+    
+    return 0;
 
     if(argc > 1 && !strcmp(argv[1], "-test"))
         Test();

source/XLink.cpp

@@ -197,37 +197,16 @@ create a hyperedge with two input tensors and a output tensor
 >> h - head tensor
 >> id - id of the edge type
 */
-void XLink::MakeLink(XTensor * t1, XTensor * t2, XTensor * h, int id)
+void XLink::MakeLink(const XTensor * t1, const XTensor * t2, XTensor * h, int id)
 {
     if(h != NULL)
         return;
+    
+    XList list(2);
+    list.Add(t1);
+    list.Add(t2);
 
-    /* forward */
-    XLink &income = h->income;
-    income.Reset();
-    income.SetHead(h);
-    if(t1 != NULL && t2 != NULL)
-        income.AddTwoTails(t1, t2);
-    else if(t1 != NULL)
-        income.AddTail(t1);
-    else{
-        ShowNTErrors("TODO!");
-    }
-    income.SetType(id);
-
-    /* backward for t1 */
-    if(t1 != NULL){
-        XLink &outgo = t1->outgo;
-        CheckNTErrors(outgo.head != t1, "Wrong head of the hyperedge!");
-        outgo.AddTail(h);
-    }
-
-    /* backward for t2 */
-    if(t2 != NULL){
-        XLink &outgo = t2->outgo;
-        CheckNTErrors(outgo.head != t2, "Wrong head of the hyperedge!");
-        outgo.AddTail(h);
-    }
+    MakeLink(&list, h, id);
 }
 
 /* 
@@ -287,7 +266,7 @@ replace a node with another, i.e., we redirect the links to the new node
 >> oldOne - the node to be replaced
 >> newOne - the new node
 */
-void XLink::Replace(XTensor * oldOne, XTensor * newOne)
+void XLink::Replace(const XTensor * oldOne, XTensor * newOne)
 {
     if(oldOne == NULL || newOne == NULL)
         return;
@@ -329,10 +308,6 @@ void XLink::Replace(XTensor * oldOne, XTensor * newOne)
             }
         }
     }
-
-    XLink &oldOutgo = oldOne->outgo;
-    ClearIncoming(oldOne);
-    oldOne->outgo.tailNum = 0;
 }
     
 } // namespace nts(NiuTrans.Tensor)

source/XLink.h

@@ -116,7 +116,7 @@ struct XLink
 
     /* create a hyper edge with two input tensors and a output tensor */
     static
-    void MakeLink(XTensor * t1, XTensor * t2, XTensor * h, int id);
+    void MakeLink(const XTensor * t1, const XTensor * t2, XTensor * h, int id);
 
     /* create a hyper edge with a list of input tensors and a output tensor */
     static
@@ -132,7 +132,7 @@ struct XLink
 
     /* replace a node with another, i.e., we redirect the links to the new node */
     static 
-    void Replace(XTensor * oldOne, XTensor * newOne);
+    void Replace(const XTensor * oldOne, XTensor * newOne);
 };
     
 } // namespace nts(NiuTrans.Tensor)

source/XList.cpp

@@ -111,7 +111,7 @@ void XList::Create(int myMaxNum, XMem * myMem)
 add an item into the list
 >> item - pointer to the item
 */
-void XList::Add(void * item)
+void XList::Add(const void * item)
 {
     if( count == maxNum ){
         void ** newItems;
@@ -126,7 +126,8 @@ void XList::Add(void * item)
         maxNum = maxNum * 2 + 1;
     }
     
-    items[count++] = item;
+    MTYPE p = (MTYPE)item;
+    items[count++] = (MTYPE*)p;
 
 }
 
@@ -355,4 +356,4 @@ void XList::Shuffle(int nround, int beg, int len)
 }
 
 } 
-/* end of the nts (NiuTrans.Tensor) namespace */
\ No newline at end of file
+/* end of the nts (NiuTrans.Tensor) namespace */

source/XList.h

@@ -69,7 +69,7 @@ public:
 
     /* utilities */
     void Create(int myMaxNum, XMem * myMem);
-    void Add(void * item);
+    void Add(const void * item);
     void Add(void ** inputItems, int inputItemCount);
     void AddList(XList * l);
     void AddInt(int i);
@@ -99,4 +99,4 @@ public:
 } 
 /* end of the nts (NiuTrans.Tensor) namespace */
 
-#endif
\ No newline at end of file
+#endif

source/XTensor.cpp

@@ -81,6 +81,7 @@ constructor
 XTensor::XTensor()
 {
     memset(this, 0, sizeof(XTensor));
+    SetDataPointer();
 
     id = MakeTensorID();
     order = -1;
@@ -104,9 +105,10 @@ XTensor::XTensor()
 }
 
 /* constructor */
-XTensor::XTensor(XTensor * reference)
+XTensor::XTensor(const XTensor * reference)
 {
     memset(this, 0, sizeof(XTensor));
+    SetDataPointer();
 
     id = MakeTensorID();
     dataType = DEFAULT_DTYPE;
@@ -129,6 +131,7 @@ XTensor::XTensor(const int myOrder, int myDevID, XMem * myMem)
 {
     CheckNTErrors((myOrder > 0), "Illegal tensor order1");
 
+    SetDataPointer();
     id = MakeTensorID();
     order = myOrder;
     memset(dimSize, 0, sizeof(int) * MAX_TENSOR_DIM_NUM);
@@ -166,6 +169,7 @@ XTensor::XTensor(const int myOrder, const int * myDimSize, const TENSOR_DATA_TYP
 {
     CheckNTErrors((myOrder > 0), "Illegal tensor order1");
 
+    SetDataPointer();
     id = MakeTensorID();
     order = myOrder;
     memset(dimSize, 0, sizeof(int) * MAX_TENSOR_DIM_NUM);
@@ -188,8 +192,9 @@ XTensor::XTensor(const int myOrder, const int * myDimSize, const TENSOR_DATA_TYP
 }
 
 /* copy constructor */
-XTensor::XTensor(XTensor &reference)
+XTensor::XTensor(const XTensor &reference)
 {
+    SetDataPointer();
     id = MakeTensorID();
     ShallowCopy(reference);
     isInit = false;
@@ -200,7 +205,13 @@ XTensor::XTensor(XTensor &reference)
         devID = reference.devID;
         mem = reference.mem;
         data = reference.data;
-        reference.data = NULL;
+        
+        /* what we really want to do is "reference.data = NULL;"
+           As "reference" is constant, we cannot reset reference.data
+           here. So we save the ADDRESS of reference.data in
+           reference.dataP, and do this work by updating "*reference.dataP".
+           This is VERY trick and might not be the best solution :) */
+        *reference.dataP = NULL;
     }
     else{
         DestroyData();
@@ -263,7 +274,7 @@ void XTensor::ShallowCopy(const XTensor &tensor)
 }
 
 /* overloading of the equal-sign */
-XTensor& XTensor::operator= (XTensor& tensor)
+XTensor& XTensor::operator= (const XTensor& tensor)
 {
     /* hard copy of data array */
     int size = unitNum * unitSize;
@@ -284,7 +295,6 @@ XTensor& XTensor::operator= (XTensor& tensor)
         }
 
         Resize(tensor.order, tensor.dimSize, tensor.dataType, tensor.denseRatio);
-
         CopyValues(&tensor, this);
     }
 
@@ -620,6 +630,12 @@ bool XTensor::CheckData(const void * d, int num, int beg)
 #endif
     return true;
 }
+    
+/* set the pointer to "data" */
+void XTensor::SetDataPointer()
+{
+    dataP = &data;
+}
 
 bool XTensor::CheckData(const void * d, int num, float tolerance, int beg)
 {
@@ -969,7 +985,7 @@ increase the value of a cell in a 2d tensor
 }
 
 /* get the number of non-zero elements (in a sparse tensor) */
-int XTensor::GetNonzeroSize()
+const int XTensor::GetNonzeroSize()
 {
     if(!isSparse){
         XPRINT(1, stderr, "WARNING! Counting non-zero elements in a dense tensor might be slow!\n");
@@ -1736,7 +1752,7 @@ initialize a tensor with a reference tensor
 >> tensor - the tensor we intend to initialize
 >> reference - the reference tensor
 */
-void InitTensor(XTensor * tensor, XTensor * reference)
+void InitTensor(XTensor * tensor, const XTensor * reference)
 {
     if(reference->order < 0)
         return;

source/XTensor.h

@@ -56,11 +56,11 @@ struct XLink;
 #define FAST_MATRIX
 
 /* 
-We implemente the tensor class here though we have defined the class of XMatrix. It
-is the parent class of XMatrix.
+We implemente the tensor class here.
 */
-struct XTensor
+class XTensor
 {
+public:
     /* id */
     int id;
 
@@ -73,6 +73,10 @@ struct XTensor
     /* copy of data on the host memory. It is only activated 
        when the matrix is operated on GPUs */
     void * dataHost;
+    
+    /* a pointer to data (i.e., a pointer to the address of "data".
+       This is for reset "data" when XTensor is used as a const variable. */
+    void ** dataP;
 
     /* 
     device id 
@@ -158,7 +162,7 @@ struct XTensor
     XTensor();
 
     /* constructor */
-    XTensor(XTensor * reference);
+    XTensor(const XTensor * reference);
 
     /* constructor */
     XTensor(const int myOrder, int myDevID, XMem * myMem);
@@ -168,7 +172,7 @@ struct XTensor
             const float myDenseRatio, XMem * myMem);
 
     /* copy constructor */
-    XTensor(XTensor &reference);
+    XTensor(const XTensor &reference);
 
     /* de-constructor */
     ~XTensor();
@@ -180,7 +184,7 @@ struct XTensor
     void ShallowCopy(const XTensor &tensor);
 
     /* overloading of the equal-sign */
-    XTensor& operator= (XTensor &tensor);
+    XTensor& operator= (const XTensor &tensor);
 
     /* judge whether the two matrices are in the same type and size */
     static
@@ -225,6 +229,9 @@ struct XTensor
 
     /* check whether the data array is the same as the answer */
     bool CheckData(const void * answer, int num, float tolerance, int beg = 0);
+    
+    /* set the pointer to "data" */
+    void SetDataPointer();
 
     /* set the cell to the ascending order along a given dimension */
     void SetAscendingOrder(int dim);
@@ -275,7 +282,7 @@ struct XTensor
     bool Add2D(DTYPE value, int ni, int mi);
 
     /* get the number of non-zero elements (in a sparse tensor) */
-    int GetNonzeroSize();
+    const int GetNonzeroSize();
 
     /* set the tensor as "temporary" */
     void SetTMP(bool myIsTmp = true);
@@ -350,7 +357,7 @@ void InitTensor5D(XTensor * tensor, const int d0, const int d1, const int d2, co
                   const TENSOR_DATA_TYPE myDataType = X_FLOAT, const int myDevID = -1, XMem * myMem = NULL);
 
 /* initialize a tensor with a reference tensor */
-void InitTensor(XTensor * tensor, XTensor * reference);
+void InitTensor(XTensor * tensor, const XTensor * reference);
 
 /* generate a XTensor */
 XTensor * NewTensor(const int myOrder, const int * myDimSize, const TENSOR_DATA_TYPE myDataType = X_FLOAT,

source/core/arithmetic/Sum.cpp

@@ -35,11 +35,8 @@ return a pointer
 >> c - where we put a+b*\beta. we save it in a if c is NULL
 >> beta - the scaling factor
 */
-void _Sum(XTensor * a, XTensor * b, XTensor * c, DTYPE beta)
+void _Sum(const XTensor * a, const XTensor * b, XTensor * c, DTYPE beta)
 {
-    if (c == NULL)
-        c = a;
-
     CheckNTErrors(a && b && c, "Empty tensors in addition!");
     CheckNTErrors(a->unitNum == b->unitNum && a->unitNum == c->unitNum,
                   "Unmatched tensors in addition!");
@@ -121,7 +118,7 @@ do it on site
 >> b - another tensor
 >> beta - the scaling factor
 */
-void _SumMe(XTensor * a, XTensor * b, DTYPE beta)
+void _SumMe(XTensor * a, const XTensor * b, DTYPE beta)
 {
     _Sum(a, b, a, beta);
 }
@@ -133,7 +130,7 @@ return a XTensor structure
 >> b - another tensor
 >> beta - the scaling factor
 */
-XTensor Sum(XTensor &a, XTensor &b, DTYPE beta)
+XTensor Sum(const XTensor &a, const XTensor &b, DTYPE beta)
 {
     XTensor c(&a);
     c.SetTMP();
@@ -141,6 +138,8 @@ XTensor Sum(XTensor &a, XTensor &b, DTYPE beta)
     /* computation */
     _Sum(&a, &b, &c, beta);
     
+    c.Dump(stderr, "c: ");
+    
     /* tensor connections */
     XLink::MakeLink(&a, &b, &c, MATH_SUM);
     XLink::AddParamToHead(&c, beta);

source/core/arithmetic/Sum.h

@@ -27,13 +27,13 @@
 namespace nts { // namespace nts(NiuTrans.Tensor)
 
 /* tensor summation c = a + b * \beta */
-void _Sum(XTensor * a, XTensor * b, XTensor * c = NULL, DTYPE beta = (DTYPE)1.0);
+void _Sum(const XTensor * a, const XTensor * b, XTensor * c, DTYPE beta = (DTYPE)1.0);
 
 /* tensor summation a = a + b * \beta (return a pointer) */
-void _SumMe(XTensor * a, XTensor * b, DTYPE beta = (DTYPE)1.0);
+void _SumMe(XTensor * a, const XTensor * b, DTYPE beta = (DTYPE)1.0);
     
 /* tensor summation c = a + b * \beta (return a structure) */
-XTensor Sum(XTensor &a, XTensor &b, DTYPE beta = (DTYPE)1.0);
+XTensor Sum(const XTensor &a, const XTensor &b, DTYPE beta = (DTYPE)1.0);
 
 } // namespace nts(NiuTrans.Tensor)
 

source/core/movement/CopyValues.cpp

@@ -32,7 +32,7 @@ copy s to t
 >> stream - the stream for creating the job pipeline
 << return - succeeded or not
 */
-bool CopyValues(XTensor * s, XTensor * t, XStream * stream)
+bool CopyValues(const XTensor * s, XTensor * t, XStream * stream)
 {
     if (s == NULL || t == NULL)
         return false;
@@ -60,10 +60,10 @@ bool CopyValues(XTensor * s, XTensor * t, XStream * stream)
         memcpy((char*)t->data, (char*)s->data, s->unitSize * s->unitNum);
     }
     else if (s->isSparse && t->isSparse) {
-        int d = s->GetNonzeroSize();
+        int d = s->unitNumNonZero;
         t->Resize(s);
         t->unitNumNonZero = d;
-        memcpy((char*)t->data, (char*)s->data, sizeof(int) + d *(sizeof(int) + sizeof(DTYPE)));
+        memcpy((char*)t->data, (char*)s->data, sizeof(int) + d *(sizeof(int) + t->unitSize));
     }
     else {
         ShowNTErrors("TODO!");

source/core/movement/CopyValues.h

@@ -28,8 +28,8 @@ namespace nts { // namespace nts(NiuTrans.Tensor)
 
 /* copy s to t */
 extern "C"
-bool CopyValues(XTensor * s, XTensor * t, XStream * stream = NULL);
+bool CopyValues(const XTensor * s, XTensor * t, XStream * stream = NULL);
 
 } // namespace nts(NiuTrans.Tensor)
 
-#endif // __COPYVALUES_H__
\ No newline at end of file
+#endif // __COPYVALUES_H__

source/test/TSum.cpp

@@ -55,7 +55,7 @@ bool TestSum1()
     b->SetData(bData, unitNum);
 
     /* call sum function */
-    _Sum(a, b);
+    _Sum(a, b, a);
 
     /* check results */
     cpuTest = a->CheckData(answer, unitNum);