Differences

This shows you the differences between two versions of the page.

--- spin:esc202_fs2021 [2021/03/01 08:28]
stadel [Lectures]
+++ spin:esc202_fs2021 [2021/04/26 14:21] (current)
stadel [Lectures]
@@ Line 20: / Line 20: @@
 [[http://www.ics.uzh.ch/~stadel/krone/public_downloads/esc202|ESC202 Video downloads]]
+**Appologies:** due to an error there was no sound in the recording of 29. Mar. As such it was of little use and I didn't bother uploading the video.
@@ Line 25: / Line 27: @@
 . Feb. 2021: {{ :spin:sins2-01.pdf |Tree Structures and Dimensional Searching}}
+. Mar. 2021: {{ :spin:fs2021_week2.pdf |Nearest Neighbor Searching}}
+. Mar. 2021: {{ :spin:fs2021_week3.pdf |k-Nearest Neighbor Searching and SPH intro}}
+. Mar. 2021: {{ :spin:fs2021_week4.pdf |SPH METHOD: Equations and kernels}}
+. Mar. 2021: {{ :spin:fs2021_week5.pdf |SPH METHOD: part 2}} //Artificial Viscosity formula corrected!//
+. Mar. 2021: {{ :spin:fs2021_week6.pdf |Kernels and a "Wind Tunnel"}}
+. Apr. 2021: {{ :spin:fs2021_week7_inflow.pdf |Howto: inflow/outflow BCs}} and followed by the {{ :spin:fs2021_week7.pdf |2-D Ising Model}}
+. Apr. 2021: {{ :spin:fs2021_week8.pdf |Travelling Salesperson Problem}}
 ====== Assignments ======
@@ Line 74: / Line 90: @@
 ====== List of assignments ======
-.k-Nearest Neighbors algorithm (you can set k=8 or other values to test). You should check the code via direct computation (O(N^2)). Solution is expected to include a priority queue over the 8 NN particles, but could also explore the use of a further prioq over the nearest cells during the treewalk as discussed in the lecture.
+. Build a binary tree of cells using the partitioning of particles function we introduced in class. The hard part is making sure your partition function is really **bomb proof**, check all "edge cases" (e.g., no particles in the cell, all particles on one side or other of the partition, already partitioned data, particles in the inverted order of the partition, etc...). Once you have this, then recursively partition the partitions and build cells linked into a tree as you go. Partition alternately in x and y dimensions, or simply partition the longest dimension of the given cell.
+. For each particle find (count) all the neighbor particles within some ball radius (small compared to the root cell size). How long does it take? How does it scale as you increase the number of particles, N, in the domain? Explain which method of intersection tests you used in your implementation.
+. For each particle calculate the "top-hat" density from the 32 nearest neighbors and plot it using a colormap. Use first a linear search priority queue to get things working and then substitute a heap algorithm to implement "replace" and "max" functions. We will need the density to implement SPH so it needs to be well tested. Also, make sure it can work with periodic boundary conditions!
+. Now, calculate the density using the Monaghan kernel defined in the lecture. Plot and compare to the density you get from the "top-hat" kernel (the Monaghan result should be a little smoother).
+. Using the SPH method, simulate the movement of a fluid through a "wind tunnel" that contains an overdensity in the middle as it was described in the lecture **(due between April 12th and April 19th)**!
+. 2D Ising Model: Using the Metropolis algorithm, plot the mean magnetization of a N by N grid of spins (+1 and -1) depending on the temperature. Visualise the spin state of the grid at different temperatures **(due April 26th)**!
+. Metropolis algorithm II: Implement the traveling merchant problem **(due May 3rd)**!

stadelweb

User Tools

Site Tools

Differences

Page Tools