Problem 42793. Fast 1-D Convolution (full shape)
This is the first problem in the fast 1-D convolution series. This problem asks you to find a fast algorithm to compute the 1-D convolution in its full output shape.
There exists a fast 1-D convolution algorithm way more efficient than MATLAB's built-in conv function invoked in the form conv(u,v,'full'), and the performance improvement is more pronounced when length(u) and/or length(v) are large. Do you know how? Try it out.
- Next problem: Fast 1-D Convolution (same shape).
Solution Stats
Problem Comments
-
3 Comments
The same codes, the score is from 5 to 49, too wide.
score.p is invalid now, the score is size, not speed.
The function conv is not slow. It is like complaining that the function sum is slow. And the only way the function sum can be slow is if we are not using it properly.
Solution Comments
Show commentsGroup

Functions II
- 15 Problems
- 6 Finishers
- High Precision Square Root (Inspired by Project Euler 80)
- Damping of Servomotors with Tachometer Feedback
- Find the right number make the equation
- Deriving a function using the difference quotient
- Fast 1-D Convolution (full shape)
- Fast 1-D Convolution (same shape)
- Fast 1-D Convolution (valid shape)
- Lambert's W
- First use of arrayfun() and anonymous function @(x)
- Cell Operator *
- Product of two multivariate polynomials
- Multivariate polynomials - convert monomial form to array
- Multivariate polynomials - overload multiplication
- Multivariate polynomials - emulate symbolic form
- Rewrite setdiff to account for non-unique values
Problem Recent Solvers31
Problem Tags
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!