WEBVTT 00:00.670 --> 00:02.140 Hello and welcome back to the call. 00:02.140 --> 00:05.840 So now we are coming to the core methods of my old course. 00:05.860 --> 00:11.560 So when it comes to large scale and numerical operations none pay as much faster than the pace and standard 00:11.560 --> 00:12.140 library. 00:12.160 --> 00:18.420 Because they are we can use spectre raised element y's operations and we do not need any follow up size 00:18.430 --> 00:20.150 statements and so on. 00:20.380 --> 00:26.680 And in this sense though we can also save many lines of code with an empire in Python we can measure 00:26.680 --> 00:30.010 the time our machine needs for the execution of code. 00:30.070 --> 00:37.030 So with the magic function time it so it's called magic because it starts as a percentage sign. 00:37.390 --> 00:43.720 So there are many so-called magic functions and time it runs the code multiple times and returns as 00:43.870 --> 00:46.260 the ever it's execution time. 00:46.330 --> 00:53.740 So but now let's first impact an umpire and then reset the size of our lists and our race. 00:53.740 --> 01:00.170 First of all to three so we only have three elements and then we create an umpire array with three elements. 01:01.000 --> 01:02.560 So the length is three. 01:02.560 --> 01:05.800 So we have an array that's 0 1 and 2. 01:05.800 --> 01:13.150 And also we create a list with the range function and three elements as you can see now here we have 01:13.180 --> 01:14.970 our array and our list l. 01:15.040 --> 01:20.530 And now we measure the time our machine needs to add to each element in our array and the integer to 01:27.840 --> 01:31.030 so it's 660 nanoseconds. 01:31.290 --> 01:33.500 And the same we are doing now for the list. 01:33.510 --> 01:35.070 So we have a list comprehension. 01:35.110 --> 01:42.090 So for each element in our list there are only three elements we add to let's also measured the time 01:42.090 --> 01:45.930 here. 01:45.940 --> 01:48.600 Here we have two hundred seventy six nanoseconds. 01:48.640 --> 01:54.880 So you can see in the very small scale operation num pi is slower here than than the python standard 01:54.880 --> 01:59.400 library and we can also try other operations like multiplication. 01:59.410 --> 02:02.590 So we multiply each element of our array by two 02:09.210 --> 02:15.240 so we get six hundred fifty nine nanoseconds and we do the same four for the list elements multiplying 02:15.240 --> 02:15.750 by two 02:18.640 --> 02:24.020 and also here two hundred seventy nanoseconds at least at a small scale with only three elements num 02:24.060 --> 02:36.140 pi is slower let's go onto squaring our elements and let's see the average time for the man pi package. 02:36.170 --> 02:42.890 So it's five hundred fifty seven nanoseconds and we do the same with the list comprehension 02:53.700 --> 03:01.710 so now obviously here with the operations squaring num pi is faster than than the python standard library. 03:01.710 --> 03:08.030 So before here with additional multiplication non PI was two times slower than the python standard library 03:08.040 --> 03:11.120 so in the end it also depends which operation you have under. 03:11.220 --> 03:15.120 Let's also try to get to the square root of the number package 03:22.570 --> 03:26.230 eight hundred seventy six nanoseconds and let's do the same. 03:26.260 --> 03:27.530 This list comprehension 03:32.890 --> 03:35.640 now again here the pace and standard library is faster. 03:35.740 --> 03:39.340 So it depends on the operation of the type of the operation. 03:39.340 --> 03:42.840 All right so but it seems very a very small scale that. 03:42.880 --> 03:43.070 Yeah. 03:43.090 --> 03:45.720 That's not really an advantage of the number I package. 03:45.730 --> 03:55.440 But um let's increase our size of elements from three to let's say 1 million. 03:55.880 --> 03:58.460 So we have from our array 1 million elements. 03:58.580 --> 04:04.190 And also on our list and let's read from the whole thing. 04:04.190 --> 04:09.470 So let's begin with Ed.. 04:09.810 --> 04:12.840 So now we have two point two three milliseconds for the month. 04:12.860 --> 04:13.770 Hi Ed.. 04:13.770 --> 04:16.610 Let's see what we get with the list comprehension. 04:24.610 --> 04:29.800 Now we get your ninety two point eight milliseconds so Nampa here was about the hour forty or forty 04:29.800 --> 04:32.560 five times faster than Python standard library. 04:32.560 --> 04:38.380 And let's also try the multiplication with one million elements. 04:38.380 --> 04:41.790 So now I'm pi needs on average two points with six milliseconds. 04:41.890 --> 04:44.950 And let's try the python standard library 04:52.270 --> 04:59.800 also here ninety two point eight milliseconds so about forty forty five times slower than empire and 04:59.800 --> 05:01.360 let's go onto a squaring 05:05.670 --> 05:09.010 two points three eight milliseconds for the number pi package. 05:09.480 --> 05:11.060 And the list comprehension 05:13.690 --> 05:21.630 362 milliseconds num pi is about 250 times faster than the python standard library. 05:21.720 --> 05:23.790 And last but not least the square root 05:32.820 --> 05:42.750 nine point four one milliseconds and this comprehension 217 milliseconds now pays about twenty times 05:42.750 --> 05:47.940 faster than the python standard library so we have seen that. 05:47.980 --> 05:48.190 Yeah. 05:48.240 --> 05:54.360 I'm pi is when it comes to very large scale operations with uh thousands or millions of elements. 05:54.370 --> 05:56.590 The number is really much faster. 05:56.590 --> 06:03.110 So on average about 50 to a 100 times faster than the python standard library. 06:03.250 --> 06:05.410 And it also depends on the type of operations. 06:05.410 --> 06:10.810 All right so this is it for the time being and in the next session we will have a small case study where 06:10.810 --> 06:15.400 we also have a look at the performance between num pi enterprise and standard library so hope to see 06:15.400 --> 06:16.090 you there by.