CS计算机代考程序代写 python algorithm DASK HOMEWORK (HW 1) DUE: SEPTEMBER 27, 11:59PM EST

DASK HOMEWORK (HW 1) DUE: SEPTEMBER 27, 11:59PM EST

1. Instructions

1.1. What to look at. The only references you may use for this lab are:

• The dask website for api reference
• Any material on Canvas

In particular, do not use stackoverflow or chegg for two reasons: (1) it will result in an academic integrity
violation and (2) you are likely to get a hilariously incorrect answer from them.

1.2. How to collaborate.

• Write down (in your code) the names of people you collaborate with.
• You cannot look at code/pseudocode written by other students.
• You cannot copy/email/transmit code. Your code must be typed by you without looking at other

people’s code.
• You can discuss algorithmic strategies (pictures are fine, pseudocode is not)

1.3. Getting the code and submitting the code. Run updatestarter. To submit your code, commit
and push it to github, then submit to the gradescope hw1 assignment (tell gradescope to pull the code from
your github repository).

Make sure to only commit python code, and no other files into your repository.

1.4. The Assignment. The homework consists of 3 files:

• hw1.py: This is where you will write your code (you can add functions if you want)
• driver.py: This file helps test your code (type python driver.py in the command line to run it)
• hwfunctions.py: This file contains functions your code needs to use.

The goal of this assignment is to create parallel versions of the following functions:

1 from hwfunctions import fun_inc , fun_factor

2

3 def serial_inc(start , end):

4 return sum([ fun_inc(i) for i in range(start , end)])

5

6 def serial_factor(start , end):

7 return sum([ fun_factor(i) for i in range(start , end)])

Your parallel versions must be significantly faster than the serial versions (your goal is to use the
ideas taught in class to see how much faster you can make it). After your parallel versions have been
added to hw1.py, you can type python driver.py. This will run your parallel code and compare it to the
serial code (it will compare answers and speed).

Question 1 (20 points). In hw1.py, fill in the function delayed_increment(c, start, end). This function
must use dask delayed to create a parallel version of serial_inc. The first parameter c is the client, so
you do not need to create one yourself. The return value of the function must be a dask delayed object,
not the result of the computation (the code in driver.py will take the computation graph defined by your
delayed object and run it to get the answer). Your function must make use of the fun_inc function defined
in hwfunctions.py.

Question 2 (20 points). In hw1.py, fill in the function delayed_factor(c, start, end). This function
must use dask delayed to create a parallel version of serial_factor. The first parameter c is the client, so
you do not need to create one yourself. The return value of the function must be a dask delayed object,
not the result of the computation (the code in driver.py will take the computation graph defined by your
delayed object and run it to get the answer). Your function must make use of the fun_factor function
defined in hwfunctions.py.

1

2 DASK HOMEWORK (HW 1) DUE: SEPTEMBER 27, 11:59PM EST

Question 3 (20 points). In hw1.py, fill in the function future_increment(c, start, end). This function
must use the futures api to create a parallel version of serial_inc (this is like question 1, but using futures
instead of delayed). The first parameter c is the client, so you do not need to create one yourself. The return
value of the function must be a futures object, not the result of the computation (the code in driver.py
will obtain the answer from the futures object). Do not use delayed api in this function. Your function
must make use of the fun_inc function defined in hwfunctions.py.

Question 4 (20 points). In hw1.py, fill in the function future_factor(c, start, end). This function must
use the futures api to create a parallel version of serial_factor. The first parameter c is the client, so you
do not need to create one yourself. The return value of the function must be a futures object, not the
result of the computation (the code in driver.py will obtain the answer from the futures object). Do not
use delayed api in this function. Your function must make use of the fun_factor function defined in
hwfunctions.py.

1. Instructions
1.1. What to look at.
1.2. How to collaborate
1.3. Getting the code and submitting the code
1.4. The Assignment