Term
| What are the advantages of type theory , explain briefly? |
|
Definition
There are many advantages of type decleration such as -type mismatch error -memory optimization - compile time operator disambiguation -accuracy control -compile time coercion. Type mismatch error is caused if two diffrent variables do not have same declared or implicit types. Memory optimiization is possibe since diffrent types of data objects need diffrent amount of memory. By declearing type at memory appropriate memory alllocation can be done at compile time. Compile time operator disambiguation is possible since we know the type of data objects at compile time. Compile time coercion is possible as variable of the data type with less information can be transformed to variable of data type with more information. |
|
|
Term
|
Definition
Polymorphism is when the same function or operator can be used for more than type of data objects.
|
|
|
Term
| What are two types of polymorphism? |
|
Definition
-Universal,Adhoc -Universal polymorphism - data types are indefinite -adhoc polymorphism - data types are finte |
|
|
Term
| What are two universal polymorphism types? |
|
Definition
parametric , inclusion. Paremtric polymorphism - types are passed as parameters to specialize a function for a a specific type of data object Inclusion polymorphism - a type is declared as subtype of the original type to inherit all the operations of the original type. |
|
|
Term
| What are the two types of adhoc polymorphism? |
|
Definition
overloading,coercion. overloading - an operator that can have multiple meanings is disambiguated by using the information of operands. coercion - a variable with type with less information is transformed to a varaible with more information when an operation involves both types of variables |
|
|
Term
| What types of errors can not be picked by type declarations and why? Name and explain some instance of these types? instances of these types of errors. |
|
Definition
| Type declearation can only pick up type mismatch errors, and can not pick up runtime errors since type information and boundry information is lost after compilation. Some examples are range check errors in arrays, substring arrays in selecting substring, variant record arrays when switch is mixed up , type violations when pointers are treated as distinct types. |
|
|
Term
| What is Name equivalence? |
|
Definition
| Name equivalence says two variable of types having diffrent names cannot be equated even when two types have the same structure. |
|
|
Term
| How can recursive data structures be modeled using set operations.? |
|
Definition
| Recursive types can be modeled as combination of disjoint union and Cartesian product: disjoint union is used to model the union of base case and recursive case in type definition, and cartesian product is used to connect multile catenations in recursive part of defintions. |
|
|
Term
| What are guarded commands and what are their rules of execution for them? |
|
Definition
| Guarded commands is non deterministic style of programming where slection part can be modeled in any order using guards. A block is a set of garded commands where each guraded command has two parts, a gard part and a command pat , gards are booleaon conditions that must be true to execute the command part. Guards in a block can be in any order. Only if the guards scueed command are eexecuted. Guards are necessary conditions thus that sucess of a guard does not ensure a solution. |
|
|
Term
| What do you understand by data datadependecny graph? |
|
Definition
| Data dependency graph is the connectivity of multiple statements in the program to show dependency of execution on of states on one another. It shows the order in which various statements can be executed without destroying the final outcome of the answertil producers have written the value |
|
|
Term
| What are the three golden rules of data depedency graph? |
|
Definition
The three rules of data depedency graphs include 1. consumer of variables can not be executed until producers have written the value of variables. 2. new values can not be written for variables unless the statements consuming the old values of a variable are executed. 3. The statement using aliased variables should follow the order. |
|
|
Term
| What is fine grain conccurancy |
|
Definition
| Fine grain parallelism is when maximum concurrency is exploited at the statement level using data dependency graph. The problem with data dependency graph is that after the processor allocation, many data elements have to be transfered from one process to another. The overhead of packing unpacking data for data transfere is quite high that takes finge grin parallesim unattractive. |
|
|
Term
| what is coarse grain parallelism ? |
|
Definition
| Coarse grain parallelism groups multiple sequential statments on one machine and exploits concurrency between these blocks of multiple sequential statements. The blocks are sepreated such that there is minmum data transfer between these blocks, and these data transfers are grouped to reduce packing unpacking cost. Coarse grain parallelism is more suited because it reduces cost of data transfer between processors that can make concurrent computation inefficient. |
|
|
Term
| what do you understand by program slicing? |
|
Definition
| Program slicing is the means spliting the program into multiple blocks such that the data transfer between the various blocks is minimal. Each of these blocks are executed sequeintailly on diffrent processors concurrently after data dependency graph is created. |
|
|
Term
| What are the two rules used in slicing a program? |
|
Definition
Two standord techniques are used for program slicing: 1. if the number of statemens on one processor are smaller than the cost of transfering the data to another procesor then it will be more efficient to group those statements on the other processor. 2. keep multiple copies of producers that are consumed by statements in multiple processors. |
|
|
