Merge pull request #4 from AllanLRH/master

Added cumsum and cumprod functions

MiniPy.py

@@ -9,6 +9,25 @@
 import sublime_plugin
 
 
+def cumsum(lst):
+ return [sum(lst[:i]) for i in range(1, len(lst)+1)]
+
+
+def cumprod(lst, use_logsum=False):
+ def do_cumprod(lst):
+ tmp = 1
+ toReturn = list()
+ for el in lst:
+ tmp *= el
+ toReturn.append(tmp)
+ return toReturn
+
+ if use_logsum:
+ return [exp(el_log) for el_log in cumsum([log(el) for el in lst])]
+ else:
+ return do_cumprod(lst)
+
+
 def dnow():
  return datetime.datetime.strftime(datetime.datetime.now(), '%d/%m/%Y')
 
@@ -50,7 +69,7 @@ def set_symdiff(itr0, itr1):
  return s0.symmetric_difference(s1)
 
 
-def formatnum(num, digits, scientificNotation=None):
+def formatnum(num, digits=8, scientificNotation=None):
 
  def normalFormatting(num, digits):
  return ('{:.%df}' % digits).format(num)

README.md

@@ -76,6 +76,26 @@ The functions takes two iterable arguments, which are turned into sets, and the
  set_symdiff('foo bar', 'foo baz') -> {'z', 'r'}
 
 
+### Computing cumulative sums and products
+
+Compute the cumulative sum of an iterable:
+
+ cumsum([1,2,3,4,5]) -> [1, 3, 6, 10, 15]
+ cumsum([0.02809, 0.05619, 0.08646, 0.11919, 0.15192, 0.18465, 1.31694]) -> [0.02809, 0.08428, 0.17074, 0.28993, 0.44185, 0.6265000000000001, 1.94344]
+
+And a cumulative product
+
+ cumprod([1, 2, 3, 4, 5]) -> [1, 2, 6, 24, 120]
+ cumprod([0.02809, 0.05619, 0.08646, 0.11919, 0.15192, 0.18465, 1.31694]) -> [0.02809, 0.0015783771, 0.000136466484066, 1.626544023582654e-05, 2.471045680626768e-06, 4.5627858492773275e-07, 6.008915196347284e-07]
+
+But doing products of a lot of small numbers are prone to errors, so we can use a math trick:<br>
+<a href="https://www.codecogs.com/eqnedit.php?latex=\exp&space;\left[&space;\sum&space;\log&space;\left[&space;A&space;\right]&space;\right]&space;=&space;\prod&space;A" target="_blank"><img src="https://latex.codecogs.com/gif.latex?\exp&space;\left[&space;\sum&space;\log&space;\left[&space;A&space;\right]&space;\right]&space;=&space;\prod&space;A" title="\exp \left[ \sum \log \left[ A \right] \right] = \prod A" /></a>, where _A_ is the iterable. This will increase the numerical stability, as seen in this example:
+
+ cumprod([1e-8, 1e-9, 1e-10, 1e-11, 1e-12, 1e-13, 1e-14]) -> [1e-08, 1e-17, 1e-27, 1e-38, 9.999999999999999e-51, 9.999999999999999e-64, 1e-77]
+ cumprod([1e-8, 1e-9, 1e-10, 1e-11, 1e-12, 1e-13, 1e-14], use_logsum=True) -> [9.999999999999982e-09, 9.99999999999999e-18, 1.0000000000000022e-27, 9.999999999999936e-39, 9.999999999999944e-51, 1.0000000000000049e-63, 9.999999999999967e-78]
+ cumprod([1e-8, 1e-9, 1e-10, 1e-11, 1e-12, 1e-13, 1e-14], use_logsum=True) -> # same result as above.
+
+
 ### Formatting numbers
 
 The fnuction `formatnum` formats numbers, and takes two mandatory and an optional argument: