Train Scholarship Blog MATLAB Week 9

-

 

Genaro Rivera

November 11, 2022

Train Scholarship Blog MATLAB Week 9

 

Pre-allocating arrays and Vectorization (saving time by creating faster codes).

Pre-allocating




Pre-allocating creates a faster code; the tic and toc function calculate the speed of the code. The picture on the left shows code without pre-allocating; no pre-allocation is inefficient. MATLAB runs step by step. Here, it first creates a 1 by 1 matrix ([1]) and goes through the loop. Next, MATLAB repeats the for loop and creates a 1 by 2 matrix ([1 2]). Afterwards, since I specified 100 steps (1:100), MATLAB will continuously create new elements in the matrix. MATLAB performs 3 steps when computing the for-loop function. MATLAB finds memory for new matrices, copies and pastes the data for the new matrix, and deletes the old matrix; this takes time. The picture on the right shows code with pre-allocating; pre-allocating is efficient. Here, MATLAB repeats the same process as I described when it has no pre-allocation. However, pre-allocation involves creating a matrix with zeros with the known dimensions of the final product. Here, I know the final matrix will be 1 by 100 because I indicated the size in the beginning (rad = 1:100). In doing so, MATLAB creates the 1 by 100 matrix with zeros. This saves time from having to perform the 3 steps when computing the for-loop function. The tic and toc functions show that no pre-allocation results in 0.001459 seconds of processing, while pre-allocating results in 0.000652 seconds of processing. Here, this seems insignificant, but further on, code adds up and may have hour long differences. Pre-allocating results in faster code.

Vectorization

Vectorization creates a faster code compared to pre-allocation; the tic and toc function calculate the speed of the code. The picture on the left is missing vectorization; although pre-allocation occurs, vectorization is more efficient. MATLAB creates the matrix with the known dimensions and repeats the for-loop function until it ends. The picture on the right displays vectorization. Vectorization can sometimes, not always, replace simple for-loop functions. Vectorization creates a vector of the known dimensions and applies multiple operations to each element (must include the period before each operation to ensure an element-by-element operation occurs). MATLAB automatically loops back when you give the specified increments (here I specified 1:10000 meaning 1 through 10000 in increments of 1). The tic and toc functions show that no vectorization results in 0.008349 seconds of processing, while with vectorization results in 0.003816 seconds of processing time. Again, currently, the time difference is insignificant, but with longer and complicated code, it becomes significant. Compared to pre-allocation, vectorization creates a faster code.









Comments

Post a Comment

Popular posts from this blog

Weather Balloon Spring Flight

-
Image
 Weather Balloon Spring Flight This semester my team flew an RF receiver, CO2/ humidity sensor, Geiger Counter, seeds, a peep, and the old sensors from last semester (temperature, accelerometer, and pressure sensor). The RF receiver was to measure the strength of the Radio Frequency. The CO2 and humidity sensor are self-explanatory. The Geiger Counter (radiation sensor), temperature, and pressure coincides with the seeds; the Garden Cress seeds grow within 2 weeks, so we wanted to see which of the three variables that could affect it (radiation, temperature, and pressure) would and how. The old sensors were pre calibrated and ready to go from last semester. I do not know much about the RF receiver's calibration since another team member worked on it. The Geiger counter required calibrating with the Geology department to borrow a radioactive sample rock. The CO2 and Humidity sensor required calibrating through a separate CO2 and Humidity sensor (the separate sensors were pre-calibra...
Read more

System Block Diagrams Introduction Week 5

-
Image
  Open Loop vs Closed Loop Systems For this week I chose system block diagrams because my ECE 103 class and, moreover, my ECE 294 class use system block diagrams, and people, including me, are unfamiliar with them, so I chose to research system block diagrams. Differing block diagrams are open loop and closed loop systems. Both, open loop and closed loop systems, have similarities. Any control system will take and process an input and generate an output. Internally, control systems are divided into two sections: the controller section and process section. The controller section controls the amount of input required to process an input. The process section processes the input to generate the output. Despite similarities, open loop systems have unique characteristics. Regarding open loop systems, the reference point is independent to the output. No feedback signal that indicates how to vary the input to get the desired output exists. For example, a toaster does not know when the brea...
Read more

Temperature Sensor week 8

-
Image
  Genaro Rivera November 4, 2022 Faulty Temperature Sensor This week, I was working on a weather balloon payload. Here, I tested temperature sensors connected to a portable circuit board: copper boards with copper shapes glued to an insulator replacing wires in a circuit (instead of wires, circuit boards use paths to connect electrical components, so the flow of electricity gets from one electrical component to another) . However, the temperature sensor, AKA sensor temperature sensor 2, in port 0 was faulty. I switched the sensor from port 0 to port 3 to see if the issue was the port and concluded that the port was not the issue. I also concluded that the sensor itself was the issue and not the wire because if the wire was the issue, then the sensors would display no data; instead, the volts would be 0. To fix the issue, I soldered on a new sensor as seen below. I cut the wire to get rid of the broken sensor; since out of the two sensors on the payload this was the one going on...
Read more