text
stringlengths
1
81
start
float64
0
10.1k
duration
float64
0
24.9
How many steps does it take to sort the right half of elements?
4,516.09
2.95
And how many steps does it take to sort-- rather,
4,519.04
5.12
to merge the sorted halves?
4,524.16
1.56
So we have three questions to answer.
4,525.72
2.96
And let me propose this.
4,528.68
1.385
You can kind of punt.
4,530.065
0.875
And this is the beauty of recursion.
4,530.94
1.18
You can sort of answer the question with the question.
4,532.12
2.249
The running time to sort n elements is really the running time
4,534.369
3.531
to sort n over 2 elements, the left half, plus the running time to sort n
4,537.9
4.37
over 2 elements, the right half, plus the merging step,
4,542.27
3.12
which I claimed per my sort of left/right finger intuition
4,545.39
4.7
is big O of n steps linear.
4,550.09
2.08
And now this holds if n is greater than or equal to 2.
4,552.17
3.68
Now, unfortunately, this is like a recurrence if you will, mathematically.
4,555.85
3.96
This really is a cyclical definition.
4,559.81
2
We really need to be able to generalize this and explain
4,561.81
3.53
it a little more formulaically.
4,565.34
2.4
But indeed, take a guess.
4,567.74
1.45
If you have your old physics textbook or your math textbook,
4,569.19
2.73
what does this recurrence actually equal if you multiply it all out
4,571.92
3.09
and prove it mathematically?
4,575.01
1.33
What's the running time total of something
4,578.74
1.75
where you do some function of n over 2 plus some function of n over 2
4,580.49
4.2
plus some number of linear steps where each of those inner running times
4,584.69
5
is recursively defined?
4,589.69
2.03
And just to be clear, the reason this is like non-obvious
4,591.72
3.61
and should be kind of over most heads, at least if you're
4,595.33
2.75
less familiar with this world, is that I'm saying that the running time, t, is
4,598.08
4.15
a function of the running time, t.
4,602.23
1.5
Like again, that's defining a word with the word itself.
4,603.73
4.63
But it turns out there are ways to prove this inductively, mathematically.
4,608.36
3.24
And indeed, if you actually multiply this all out in your textbook,
4,611.6
2.791
you will get something on the order of n times log n.
4,614.391
4.449
So where has that left us?
4,618.84
1.204
We've introduced a whole number of algorithms here, the last of which,
4,620.044
2.916
merge sort, gives us this ability to actually solve problems
4,622.96
3.66
much faster than any of the others.
4,626.62
2.01
But in each case, are we able to apply some general formulaic analysis,
4,628.63
4.34
asymptotic notation so to speak-- big O, big Omega, and big Theta, that
4,632.97
5.35
describes in general terms how much time these kinds of things take.
4,638.32
5.2
So let's now translate one of today's key ingredients to actual C code
4,643.52
4.36
so that we can see how we can leverage this programmatically.
4,647.88
2.97
So here in CS50 IDE, I've got an example, sigma 0 and sigma 1.
4,650.85
5
Both of which at the end of the day do the same thing,
4,655.85
2.6
but they do it differently.
4,658.45
1.22
And in sigma 0 here, we have the following code.
4,659.67
2.26
A sigma function that takes as input an integer m as its sole argument
4,661.93
4.96
and it has a return value of type int.
4,666.89
1.59
So it's going to return a number.
4,668.48
1.375
Main takes no command line arguments.
4,669.855
1.545
So no arg v, no arg c this time.
4,671.4
1.84
But inside of main is code that's going to pester the user again and again
4,673.24
3.72
until he or she gives us a positive integer as per this condition
4,676.96
4.43
down here.
4,681.39
0.94
And then, notice down here I'm going to call a function sigma,
4,682.33
3.51
passing in whatever the user's input was.
4,685.84
1.85
10, for instance.
4,687.69
1.08
Or 11.
4,688.77
0.76
Or some other value.
4,689.53
1.15
Storing the answer in answer, and then just printing it out.
4,690.68
3.67
And what's nice about the fact that frankly you
4,694.35
2.12
can't see anything else on the screen is that for all intents and purposes,
4,696.47
3.21
sigma, the function at the moment, is a black box.
4,699.68
2.72
It is a function whose purpose in life is to add all of the numbers from 1 up
4,702.4
3.23
until the user's input and return it.
4,705.63
1.96
And we don't know yet or need to care how it's actually implemented.
4,707.59
3.99
If we do take a look inside that black box down lower in the file,
4,711.58
4.76
you'll see this code here.
4,716.34
1.62
Sigma takes an input, m, as an int.
4,717.96
3.08
Returns an int.
4,721.04
0.84
And what does it do inside?
4,721.88
1.22
Well, I've got what I'll call a counter-variable.
4,723.1
2.51
Called it sum.
4,725.61
0.9
Initialized it to 0.
4,726.51
1.38
And then using a for loop, iterating from 1 up to m,
4,727.89
4.01
because I want to sum all the numbers from 1 up through the value
4,731.9
3.63
the human has typed in.
4,735.53
1.48
I just want to keep adding to the sum that value.
4,737.01
2.73
So it's not just sum plus plus.
4,739.74
1.59
If I did sum plus plus as we keep seeing syntactically in examples,
4,741.33
3.684
that would just add 1.
4,745.014
0.916
I want to add in whatever i is.
4,745.93
2.33
Whether it's 1, or 2, or 3, as I'm counting all the way
4,748.26
4.16
up through the user's input.
4,752.42
1.52
And then I return sum.
4,753.94
1.95
But the neat thing about sigma is that it
4,755.89
2.76
affords an opportunity for recursion.
4,758.65
2.13
This isn't necessarily a better opportunity or more correct,
4,760.78
3.66
but we could view the world a little differently.
4,764.44
2.35
Sigma1.c is identical code in main, but the implementation of sigma
4,766.79
6.29
is fundamentally different.
4,773.08
1.58
And here's where it's kind of mind-blowing, at least at first glance.
4,774.66
3.21
You can implement a function, like sigma,
4,777.87
2.9
in such a way that the function calls itself.
4,780.77
3.53
Much like the running time for merge sort is defined in terms of itself.
4,784.3
3.76
Much like searching or sorting a list, it
4,788.06
2.756
can be recursively defined as searching or sorting a smaller list,
4,790.816
4.664
but with the same steps.
4,795.48
1.43
We implement this code in C as follows.
4,796.91
1.97