DHeap added

[prshnt19]

References to other Issues or PRs or Relevant literature

Closes #102

Brief description of what is fixed or changed

Other comments

[codecov]

Codecov Report

Merging #118 into master will increase coverage by 0.233%.
The diff coverage is 100%.

@@             Coverage Diff              @@
##           master      #118       +/-   ##
============================================
+ Coverage   97.94%   98.173%   +0.233%     
============================================
  Files          20        21        +1     
  Lines        1311      1588      +277     
============================================
+ Hits         1284      1559      +275     
- Misses         27        29        +2

Impacted Files	Coverage Δ
pydatastructs/trees/__init__.py	100% <100%> (ø)	⬆️
pydatastructs/trees/heaps.py	98.314% <100%> (+0.738%)	⬆️
pydatastructs/trees/binary_trees.py	96.894% <0%> (-0.271%)	⬇️
pydatastructs/graphs/graph.py	100% <0%> (ø)	⬆️
pydatastructs/graphs/__init__.py	100% <0%> (ø)	⬆️
pydatastructs/linear_data_structures/__init__.py	100% <0%> (ø)	⬆️
pydatastructs/__init__.py	100% <0%> (ø)	⬆️
...datastructs/linear_data_structures/linked_lists.py	100% <0%> (ø)	⬆️
pydatastructs/graphs/algorithms.py	100% <0%> (ø)
... and 1 more

[prshnt19]

The _build(self) in BinaryHeap doesn't seem to be doing anything. Its called once when the heap is to be built. At that time heap is empty so nothing actually happens.

[czgdp1807]

Well by default, elements can be None so _build does nothing. However, if you pass some list in elements then _build will work.

[czgdp1807]

Is it possible to remove the class BinaryHeap and add a constructor function with a same protoype as BinaryHeap.__new__?

def BinaryHeap(arguments accepted by BinaryHeap.__new__)

We can also add a special constructor for TernaryHeap.

[prshnt19]

I think a better alternative will be to inherit DHeap. BinaryHeap and TernaryHeap will be a child class with d=2 and d=3 respectively.

class BinaryHeap(DHeap):
    def __new__(cls, elements=None, heap_property="min"):
         obj = DHeap.__new__(cls, elements=None, heap_property="min", d=2)
         return obj

class TernaryHeap(DHeap):
    def __new__(cls, elements=None, heap_property="min"):
         obj = DHeap.__new__(cls, elements=None, heap_property="min", d=3)
         return obj

[prshnt19]

ping @czgdp1807

[czgdp1807]

Can we use DynamicOneDimensionalArray here instead of list?

[prshnt19]

Can we use DynamicOneDimensionalArray here instead of list?

The DynamicOneDimensionalArray itself uses list. I think using DynamicOneDimensionalArray would be a good idea only if we use cyptes for DynamicOneDimensionalArray. This will give a performance boost. I can create a new issue for that.

[czgdp1807]

Though DynamicOneDimensionalArray uses list but the deletion time for the same is amortized O(1) with a nice trade off with space complexity. What happens if we use it?

[prshnt19]

Used DynamicOneDimensionalArray instead of List in DHeap.

DynamicOneDimensionalArray could be made faster using ctypes array. ctypes implementations are 10 to 100 times faster than python and takes less memory than python.

[czgdp1807]

DynamicOneDimensionalArray could be made faster using ctypes array. ctypes implementations are 10 to 100 times faster than python and takes less memory than python.

Can you please make an experimental PR for the same?