MATH 524, Fall 2024 Nonparametric Statistics Third assignment, due Monday, November 25, 2024, noon 1. The following table, from Dowling et al. (1957, J. Lab. Clin. Med.), shows the incidence of colds among 354 men, of whom 265 received an infectious secretion while the other 89 served as controls. Determine whether these results are statistically significant at the 1% level. Developed a cold Did not develop a cold Treated 68 197 Control 12 77 2. Among appendectomies of 103 female patients, age 10 to 29 years, the appendix of 60 patients was found to be pathological at operation while 43 were found to be normal. Of these patients, 65 were located for follow-up studies 9 to 12 years after the operation; 39 of them had had a pathological appendix and 26 a normal one. The table below, from Meyer et al. (1964, Psychosomat. Med.), shows the number among these patients who did and did not have additional operations during the intervening period. Determine whether there is a significant difference between the two categories in this respect. Appendix deemed to be pathological normal No additional operations 26 10 One or more additional operations 13 16 3. The following table, also from Beecher (1959, Measurement of Subjec- tive Response), presents the total number of coughs per day of seven patients, under three different medications and a placebo administered in random order over a number of days: Subject 1 2 3 4 5 6 7 Heroin, 5 mg 251 126 49 45 233 291 1385 Dextromethorphan, 10 mg 207 180 123 85 232 208 1204 Codein, 10 mg 167 104 63 147 233 158 1611 Placebo 301 120 186 100 250 183 1913 Is there a significant difference between the four treatments? 4. The following are the (smoothed) IQ scores of a child at the indicated ages (in years). Test the hypothesis of randomness against the alter- native of an upward trend. Age 3 3.5 4 4.5 5 5.5 6 IQ Age 105 7 110 8 109 9 114 10 114 11 115 12 114 IQ 118 123 128 128 127 126 5. Twelve mediation cases involving small businesses were ranked accord- ing to the success of the mediation effort and the amount of hostility shown in the early part of the process. The results are as follows: Hostility 1 2 3 4 5 6 7 8 9 10 11 12 Success 12 7 10 11 4 1 8 9 3 6 2 5 Here 1 corresponds in the first row to the greatest amount of hostility, and in the second row to the greatest degree of success. Use Spearman’s rho and Kendall’s tau to test the hypothesis of independence against the alternative of a negative association of the two sets of ranks. 6. Let T1 , . . . , TN be a univariate time series and for i ∈ {1,..., N}, let Ri be the rank of Ti among T1 , . . . , TN . If D = (R1 −1)2 +···+(RN −N)2 , show that under the null hypothesis of randomness, one has E(D) = 6/N3 − N and var(D) = 36/N2 (N + 1)2 (N − 1). What are the smallest and largest values that D can take? 7. Suppose that in the same context as above, the ranks of T1 , . . . , TN should have been 1, . . . ,N in that order, but that because the first observation was misread, the ranks are R1 = N and Ri = i − 1 for i ∈ {2,..., N}. Given a significance level α ∈ (0, 1), determine how large N must be in order to reject the null hypothesis of randomness against the alternative of an upward trend.
Computational Nonlinear Dynamics (MTH3039, 2024) Coursework 2 Advanced concepts of stability in dynamical systems This coursework is worth 60% of the credits for this unit. The maximum number of marks for this coursework is 100, separated into into thematic task groups. Deadline for this coursework is 13th of December (Friday), 11:59am. Submitting your work Submission is done via ELE. You need to submit a .zip file containing: • A Jupyter notebook CW2 .ipynb which contains the solutions for all tasks. The notebook is run so that all code is accompanied with its numeric or visualized output, and text-based explanations and comments are appended where necessary. You must structure the notebook in sections corresponding to the individual tasks in this coursework (to add a section create a markdown cell and type # Task 1). The notebook must contain the line include("CW2.jl") in its first code block. • A Julia source code file CW2.jl that contains function definitions that you use in the above Jupyter notebook. The file CW2.jl must contain definitions for all named functions mentioned in this coursework, e.g., a function named fg_attractors or ppha_mapper, etc.. Once CW2.jl is run, all named functions mentioned in this coursework must be accessible. • (Optional) Any other Julia source code files that you include() in the above file, in case you want to organize your codebase into several files. You are allowed to put all of your Julia source code into the single CW2.jl file if you wish so. • (Optional) Any binary files that you saved to disk while working on the coursework, see the discussion below on JLD2. Marking criteria for code Marking criteria for submitted Jupyter notebook and source code: • 100%: Code output (typically a figure) is fully correct and visually clear: different things are colored differently or are plotted with different markers or linestyles, etc. Code runs and creates the same figure as saved in the Jupyter notebook. Code runs within 10 minutes or less per Task. Code is readable. • 75-99%: Code output is mostly correct but it has minor mistakes that can be cor- rected with minor work. Or, code takes unnecessarily long to run even if fully correct. • 40-74%: Code output is not correct, but there is some code that is related with the given task, and the code runs. The code can be corrected with extensive work. • 0-39%: There is no output from the code and/or the code does not run. Point deductions on the quality of the code are the same as with coursework 1 with the exceptions: • You do not have to submit a PDF version of your Jupyter notebook. But remember that all figure output must be in the submitted Jupyter notebook without re-running it! • Poorly structured code will not incur any penalties unless it also affects the correct- ness or computational performance of the output. Remember that all individually named functions throughout this coursework must exist by themselves irrespectively of source code structure. Restrictions on used packages Throughout this coursework you have to use parts of the DynamicalSystems.jl library. You can create dynamical systems (via the CoupledODEs function). You can use the trajectory function at will, or the interactive trajectory GUI. You can use the query- ing and altering interface for DynamicalSystem such as the functions step! , reinit! , or set parameter! (in general, all functions described in the documentation of the DynamicalSystem type). Other components of DynamicalSystems.jl are forbid- den from your solution code, but you may use them outside your submitted solution to check whether your solutions are correct. For plotting please use only the Makie package and its derivatives such as CairoMakie. You may use JLD2 to save data to binary files that you can load in later sections of the notebook. This allows you to perform. an extensive computation (such as computing the basins of attraction), then save it to disk, and then resume your work later by loading this file in. You must also include such binary files in your submission if you chose this approach. All packages belonging to the Julia standard library (those installed with Julia, such as LinearAlgebra or Statistics) may be used as well. No other external packages are necessary for this coursework. If any are used, they need to be justified extensively in the submitted coursework, otherwise the resulting grade will be penalized appropriately. If you intend to submit coursework using such other external packages, we recommend that you ask your lecturer’s approval before doing so. Multistability [30 points] 1. Consider the following dynamical system: where A = 2.0551, B = −2.6, C = 0.4, D = 1, E = 0.4. This dynamical system represents complicated predator-prey dynamics between predator population y and prey population x. The system has three attractors at the given parameters. Find them, not with any formal method, but just by eye, by plotting trajectories of different initial conditions that end up at different attractors. For your initial conditions (x,y) pick random values, with x being between 0.001 to 1, and y between 0.001 to 0.05. Use time sampling of ∆t = 0.1 and evolve initial conditions for a total of T = 200. To show the attractors simply provide a plot with at least three trajectories, each going to each attractor. 2. Estimate and plot the basins of attraction (BoA). To do so, initialize a grid of initial conditions for x ranging from 0.001 to 1, and for y ranging from 0.001 to 0.05. Use a total of 100 values in both ranges, which gives you 100×100 = 10,000 initial conditions. Each initial condition is evolved for time T = 500 and then given a color according to the attractor it converged to. The BoA are a heatmap of the initial conditions grid color-coded by the attractors. Classify each initial condition (x(0), y(0)) according to the final point (x(T), y(T)) of its trajectory and the following thresholds: • (x(0), y(0)) converged to attractor 1 if y(T) ≤ ythr and x(T) < xthr , • (x(0), y(0)) converged to attractor 2 if y(T) ≤ ythr and x(T) ≥ xthr , • (x(0), y(0)) converged to attractor 3 if y(T) > ythr. I suggest the values xthr = 0.5 and ythr = 10−3 . Convince yourself that the result does not depend strongly on the threshold values. 3. Perform. the same exercise for shorter and shorter T. Find a value of T below which the resulting basin-of-attraction plot start to deviate visibly from the plot for T = 500 (this means that T was not large enough for trajectories to converge to their respective attractors). 4. Create your own general purpose “featurize and group” function for mapping initial conditions into attractors. Name your function fg simple(ds, featurizer, ics; Ttr = 400, Dt = 1, T = 100, r = 0 .1). The input arguments are an instance of a DynamicalSystem (in your case this is a CoupledODEs), a featurizing function (see below), and an iterable of initial conditions ics. Inside the function, each initial condition generates a trajectory according to the keywords: first evolve for transient time Ttr, then store the trajectory X from Ttr to Ttr + T by sampling with time step Dt. Cast each trajectory X into a feature-vector using the featurizer input, which is itself a single-argument function that takes as an input a trajectory (vector of state space points) and outputs a feature (a single vector). For grouping the features into individual groups, create a function simple_grouping(features, r) that you call in your main function using the r keyword. simple_grouping implements the simple grouping algorithm (via group centers) shown in the slides of Week 7 with threshold r and returns a vector of the group-labels of the features. fg_simple returns this same vector as the attractor labels. Apply your function to a dynamical system representing the predator-prey model of (1) using the same initial conditions as in the previous Task. For this and the next exercises use the following featurizer: using Statistics: mean function featurizer(X) # X is the trajectory, a vector of points x = mean(u[1] for u in X) # average of 1st dimension y = mean(u[2] for u in X)/0 .05 # scaled average of 2nd dimension return [x, y] end Create the same BoA plot using the new labels you estimated. 5. So far your function fg simple is finding groups that represent attractors. Extend it into a new function fg attractors such that the new version also returns the actual attractors in state space. To do this, you need to keep track of which initial conditions generated the features that became group “centers” . For each initial condition that became a group center, re-create the trajectory for it (with same Ttr, Dt, T), and store it in memory. These trajectories are the different attractors, and these should be returned as a second output of your (now modified) function. Apply this extended function to the example system of the previous task and confirm that you obtain the same attractors you found in Task 1. Scatterplot these attractors on top of the BoA plot which you have created in the previous Task. 6. Extend your function fg attractors to fg attractors fractions, such that the extended function also returns a third output: the BoA fractions of each attractor. Your function ppha attractors fractions should return three outputs: the labels of the input initial conditions (vector of integers), the attractors (vector of state space sets), and the fractions of the BoAs. 7. Apply your generic function fg_attractors_fractions to the following dynamical system with parameters F = 6.9, G = 1.3, a = 0.24, b = 4.07. Use a fixed set of initial conditions given by yg = xg = zg = range(-1, 3; length = 11) ics = [[x, y, z] for x in xg for y in yg for z in zg] Plot a scatter-plot of the system attractors with different colors in 3D space. Add a plot label to each attractor plot that is their respective state space basin fractions (or alternatively, show the state space fractions as the title of the axis, or to the right of the axis, or simply print it in your Jupyter notebook output). Hint: for this dynamical system you need to come up with your own featurizerfunc- tion, you can’t re-use the one from the previous Tasks. Create one by examining trajectories of this dynamical system. You also need to decide a value for the thresh- old r as it is based on the featurizer. You can set the Ttr keyword to 100 for this Task. Nonlocal Stability [20 points] 8. This task continues from Task 5 (or 6), where you have calculated the BoA of the predator prey model with Eqs. (1) and plotted them along with the attractors. Use the approximated BoA to calculate the minimal critical shock (MCS) for the two fixed point attractors of the system using the technique discussed in the slides of Week 8 (that utilizes the BoA). Include these two vectors as arrows into the plot of the basins from Task 5. To calculate distance between points in the state space of this dynamical system, you need to use a weighted Euclidean distance. Use the function 9. The dynamical system (1) has a limit cycle attractor (at the quoted default pa- rameters). Calculate the minimal critical shock for 100 points along the limit cycle trajectory sampled with sampling time δt = Π/100 where Π is the approximate period of the limit cycle estimated approximately from a timeseries plot of any of the variables. For the initial condition to generate the limit cycle trajectory, use any point on the limit cycle. Report the magnitude and direction of the average minimal critical shock. Compare this magnitude with the average of the magnitudes of the individual minimal critical shocks for each point on the limit cycle. Explain the difference between the two. 10. Now estimate the MCS without relying on the BoA you have already estimated. First, create a new function ppha_mapper(ds, u) that given the dynamical sys- tem ds representing Eq. (1), and an initial condition u it returns an integer 1, 2, or 3, corresponding to one of the three attractors u converged to. To map individual initial conditions to attractors use the same algorithm as in Task 2. Then, create a new function ppha_mcs_random(ds, u; n = 10_000) that utilizes ppha_mapper(ds, u). Given an initial condition u, this new function returns the minimal critical shock corresponding to u using the random search algorithm dis- cussed in Week 8 lecture slides. Keyword n is how many random points v to try. Generate random points v in the same ranges for x,y in which you have originally es- timated the basins. Apply this function to the fixed point attractors of Eqs. (1) and show the MCSs you obtain, which are approximately the same as with the previous Task for large n. 11. Estimate the final state sensitivity averaged over the whole BoA that you estimated in Task 5 (or 6) following the algorithm outlined in the slides of Week 8. The BoA has a 100×100 size. First use a 10×10-sized box to estimate in it the probabilities pij , and then use a 5×5-sized box. Report the value of the average final state sensitivity in both cases and explain the similarity or difference between the two numbers. Global continuation [30 points] 12. Consider the tristable predator prey model of Eq. (1) at the quoted parameters. Perform. a global continuation for this system while varying C from 0.3 to 0.6 with steps of 0.01. To do this, need tore-use the function fg attractors fractions(ds, featurizer, ics; kw . . .) with the same featurizer as in Task 6. Use the generic continuation algorithm described in the slides of Week 9 and ignore the “seeding” component. For matching use the simple matching by centroid distance described in the lecture slides. For all parameter values use the same fixed set of initial conditions to sample the state space: values = 31 xg = range(0 .01, 1 .1; length = values) yg = range(0 .001, 0 .1; length = values) ics = [[x, y] for x in xg for y in yg] Visualize your continuation by plotting three quantities versus the C parameter: 1) the BoA fractions, 2) the mean of the x coordinate of the attractors, 3) the standard deviation of the y coordinate of the attractors (use a different Axis for each of the quantities, don’t plot all of them in the same axis). Hint: If your code takes too long to run (10+ minutes), you can decrease values to 21. 13. In this range C ∈ [0.3, 0.6] the system undergoes three bifurcations: scenarios where either the type, or number of attractors, change fundamentally. Without perform- ing linear stability analysis, but simply by looking at the plot you generated in the previous task (or equivalently by analyzing the attractor output), report the ap- proximate values Ci, i = 1, 2, 3 of when the bifurcations occurred, as well as what happened during the bifurcation. Hint: you do not need to attach formal names to the bifurcations, such as “Hopf” or “Saddle-node” . Just descriptions “attractor of type X appears/dissapears” . 14. Consider the parameter curve given by: C(θ) = 0.4 + 0.05cos(θ), E(θ) = 0.4 + 0.1sin(θ),θ ∈ [0, 2π]. This defines an ellipsis in the C-E parameter plane. Perform. a global continuation, by re-using the code of task 12, but now versus this parameter curve as θ varies from 0 → 2π (use 100 discrete values in this interval). 15. Since this parameter curve is closed (its end is also its beginning), it is guaranteed that the attractors we have found at θ = 2π are exactly the same as when we started the continuation at θ = 0. However, with the simple matching scheme we are re-using from Task 12, the attractors will not be matched correctly because some attractors that existed when θ = 0 have disappeared as θ > 0, and then reappeared when θ → 2π . Enhance the matching part of your implementation so that it also considers vanished attractors when matching, as is discussed in the lecture slides. 16. In this global continuation over the closed parameter curve, there is one more new local bifurcation that takes place which we did not encounter during the continuation of task 12. Specify what this bifurcation is, without using linear stability analysis, but simply looking at the continuation curves (mean of x coordinate and standard deviation of the y coordinate). Formally label the bifurcation from one of the local bifurcations you have learned so far (e.g., “Saddle-Node”, “Pitchfork”, “Hopf”,etc.). Tipping points (14 points) 17. A simple model that can showcase rate-dependent tipping is Stommel’s box model for Atlantic thermohaline circulation given by Here T,S are dimensionless temperature and salinity differences between the polar and equatorial ocean basins respectively, and η,α,β are parameters. Keep α = 1,β = 0.3 and produce a bifurcation diagram for T vs. η from 2 to 4. To produce the bifurcation diagram you need to utilize code that you wrote in coursework 1. 18. Continue from the above task and assume that the system starts with η = 2.5. If we are increasing η infinitesimally slowly, what kind of tipping will the system showcase? Find the critical value ηc at which this type of tipping will happen. 19. Continue from the above task and do a simulation where η is increasing from 2.5 to 3.3 with a finite constant rate δη during the time evolution. Always start your simulation with the system being at the only fixed point at η 1 = 2.5. For each δη find at which η value the system is tipping from the initial stable branch to the second one, by crossing the unstable fixed point branch. There is a critical rate δηc below which the system does not tip at all; report this critical rate. Hint: to solve this task you need to re-define your RHS equations so that they depend on time. In particular, now η is no longer a parameter of the system, but η0 , its starting value is. η is instead a function of time that needs to be evaluated inside the RHS function. Chaos (6 points) 20. Consider the following dynamical system with a = 5.2, b = 0.1. Create a function max_lyap_exp(ds, u; n=10_000, dt=1.0, d0=1e-6) which calculates the maxi- mum Lyapunov exponent of a dynamical system ds representing Eq. (5) and for the given initial condition u (a 3-element vector), utilizing the algorithm described in the slides of Week 11 lecture (rescaling of a test trajectory). The keyword n quantifies how many times to evolve the system for a time period of dt and then to rescale the test trajectory to the reference trajectory. Keyword d0 is both the initial as well as the rescaled distance of the test trajectory from the reference one. Apply your max_lyap_exp(ds, u) to the following two initial conditions: u1 = [4, 5, 0], u2 = [-4, -5, 0]. As this dynamical system is bistable, these two initial conditions lead to different attractors. Categorize these attractors into one of [fixed point, limit cycle, chaotic] solely based on the maximum Lyapunov exponent and without looking at plots of the attractors (you must justify your choice!). Then visualize the attractors (by plotting trajectories of the two given initial conditions) and confirm your classification. 21. Apply your max_lyap_exp function also to the system of Eqs. (2), but now instead of the quoted parameters use the parameters F,G,a,b = 6.886, 1.347, 0.255, 4.0]. Use the following three initial conditions: • u1 = [0.1, 0.1, 0.1] • u2 = [-0.1, 0.5, 0] • u3 = [-1.5, 1.2, 1.5] As with the previous task, classify each initial condition’s attractor to one of [fixed point, limit cycle, chaotic] without actually visualizing the attractors. You can optionally confirm your results with an accompanying plot.
Course N1613 - Sustainability Reporting Coursework assessment - weighting 20% Your answer must be 750 words in total excluding any abstract, bibliography, or appendices. Submission Deadline as per Sussex Direct. Learning Outcomes • LO 1 Demonstrate knowledge and understanding of issues covered in this module and their relevance to accounting and the role of the accountant. • LO2 Critically evaluate the benefits of sustainability reporting and the role of standard-setting organisations in the evolution of corporate reporting systems Select the latest sustainability report of a company. Examine the report to identify a key stakeholder group or a capital/ resource influenced/impacted by the core business activity of the firm. Analyse the risks (and opportunities) to the stakeholder group or resource and critically discuss the strategies adopted by the company to mitigate the identified risks. This should include an analysis of relevant quantitative and qualitative information in the report. You are required to provide appropriate references to cite the source of the report, or any other sources of information used for analysis. Additional Guidelines ● This coursework does not require an abstract. ● +/- 10% of the total word count is permissible. ● Reference - Students only need to provide reference for the sustainability report used for the coursework. If additional resources such as journal articles or reports etc are used, references should be provided accordingly. Guidance to answer the question: Answer to this question should firstly, include a brief introduction on the main business operation of the company. This should not be a descriptive and lengthy explanation about the organisation. The introduction should include a couple of sentences about the organisation and then proceed to explain the manner in which the chosen group of stakeholders or capital is impacted by core business activities of the company. The answer should include a discussion of the risks identified by the company to the chosen group of stakeholders/ capital as well as the strategies designed or implemented to mitigate these risks. Use any figures provided in the report to demonstrate the extent of risk, including financial implications. Students may use both qualitative (non-numerical) and quantitative (numerical) information available from the chosen sustainability report. The key point to note is to answer the questions in a manner that incorporates critical reflection and/ discussion. Please remember to not waste word space simply describing or rephrasing information from the report - please avoid this as such writing adds no value to the discussion. Answers should provide a critical discussion of the risk mitigation strategies or projects adopted by the company and the impact they had on the business operations/ performance of the company. A critical discussion implies weaving in analytic reflections and theoretical concepts into the narrative. One way of doing this is, for example, by reflecting on the actual reasons for the projects/strategies claimed to have benefited the chosen capital. Reflecting on whether the company has provided concrete evidence to demonstrate how these strategies or projects were implemented for example, the timelines, targets, evidence of having met these targets, how this ensured sustainable use of the resource, how this had an impact on financial and non-financial performance of the company etc. Such reflective discussion could be done using theoretical concepts such as symbolic legitimacy, forms of institutional isomorphism, stakeholder theory, social contract etc discussed during lectures and workshops. This answer could be structured (divided into sections using heading / sub-headings) according to themes you identify from your analysis, but this is not compulsory. MARKING CRITERIA Marks % Range Criteria (all 5 criteria are equally weighted) Content, knowledge and understanding Critical thinking, originality, and reflection. Evidence of Reading Referencing and Bibliography Presentation, writing style, grammar and spelling First class 70 - 100 Exceptional knowledge and conceptual understanding of complex issues, and the development and advancement of ideas and practice. i.e. in relation to incorporation of sustainability in value chain processes. Critical insightful evaluation and synthesis of complex high level of originality and reflection. A high degree of independence of thought and critical judgement. Knowledge and understanding beyond the essential and desirable readings provided and the lecture and seminar discussions. Outstanding referencing and bibliography. Evidence of a substantive use of a wide range of learning resources (i.e. draw on multiple relevant sources and refer to them in the list of references) . Outstanding, well-directed presentation, logically and coherently structured. A high degree of originality in writing, with excellent spelling, punctuation and grammar (SPaG). 2.1 60 - 69 Wide knowledge and depth of understanding of complex issues, and the development of ideas and practice. Critical evaluation and synthesis of complex issues and material which includes an original and reflective approach. A good degree of independence of thought and critical judgement. Evidence of extensive reading of supplementary sources. Good knowledge and understanding beyond the essential and desirable readings provided and the lecture and seminar discussions. Comprehensive referencing and bibliography. Evidence of use of a wide range of learning resources (i.e. draw on a range of relevant sources and refer to them in the list of references). Excellent presentation, logically structured. A good degree of originality in writing, with good spelling, punctuation and grammar (SPaG). 2.2 50 - 59 Appropriate knowledge and depth of understanding of some of the main issues, with some understanding of their development in practice Some critical evaluation and synthesis of key issues and material. Some degree of independence of thought and critical judgement, although the writing can be descriptive. Evidence of reading supplementary sources. Knowledge and understanding of the essential and desirable readings provided and some engagement with the lecture and seminar discussions. Adequate referencing and bibliography. Orderly presentation, clear structure and acceptable grammar and spelling 3rd class - pass 40 - 49 Basic knowledge and depth of understanding of key issues, principles and concepts. Some evaluation and synthesis of issues and material. Limited evidence of reading Limited referencing and bibliography Adequate presentation and structure, grammar, spelling and citation Fail 0 - 39 Limited and/or superficial knowledge of key issues, principles and concepts Limited evaluation and synthesis of issues and material Minimal evidence of reading Inadequate referencing and bibliography Poor/inadequate presentation and structure, grammar, spelling and citation
GR5242 HW03 Problem 4: Automatic Differentiation Instructions: This problem is an individual assignment -- you are to complete this problem on your own, without conferring with your classmates. You should submit a completed and published notebook to Courseworks; no other files will be accepted. Description: This homework adds more detail to the autodiff lecture in class. There are seven questions (30 points) in total, which include coding and written questions. You can only modify the codes and text within ### YOUR CODE HERE ### and/or ### YOUR ANSWER HERE ###. In [ ]: import torch import matplotlib.pyplot as plt import numpy as np from datetime import datetime import sympy as sym Differentiation Different ways of differentiation: In this problem we consider a particular function (which we call it an op in the code block, short for operation) f : R → R : x ↦ 4x(1 − x) and it's compositions with itself. Let us define, where f (n)(x) = f(f(n−1)(x)) is the n times composited function f with itself. The goal of this problem is to explore various ways of differentiation, i.e. dx/d gn(x) which are listed below: Numerical differentiation Symbolic differentiation Autodiff in forward or reverse mode In [ ]: def op(inp): return 4 * inp * ( 1 - inp ) def operation(inp, n = 1): temp = inp for i in range(n): temp = op(temp) # apply this operation n times return temp Numerical Differentiation Based on the definition of derivative at a particular point x0, The formula above suggests a simple way of approximating the derivative by taking Dhgn(x0) for a particular choice of h as your approximate derivative. This is also known as the finite difference method. Note that this approach only requires evaluation of your function around the point you are trying to take the derivative at so it's computationally efficient but the caveat is that choosing a proper h to obtain good enough approximations is generally hard (specially when the function is multivariate). However, in our case we have some structure over the function we are trying to take derivative of. Question 1 (3 points): Use a finite difference with tolerance 1e-12 to approximate the derivative of at g3(x) at x = 1: In [ ]: tol = 1e-12 ### Your Code Here ### The approximation will be bad at certain points x, and becomes less stable as n becomes larger for a fixed h (or as the dimensionality of the function grows). Calculus and (manual/symbolic) differentiation Notice that f is a polynomial and that composition of polynomials yields also a polynomial. Therefore, gn is a polynomial with degree 2n (try to argue this for yourself using induction); we can compute the derivative using calculus. Instead of computing the derivative by hand we use the help of an auxillary package sympy and try to compute the derivative. This package uses symbolic variables and traces operations such as add, multiplication, division, etc., applied onto these variables and computes the derivative using chain rule. It is not difficult to (manually) derive a closed form. expression for gn using the recursive formula gn (x) = 4gn−1(x)(1 − gn−1(x)), g1 (x) = 4x(1 − x). The following block of code prints this closed form. expression in terms of a symbolic variable X. In [ ]: X = sym.Symbol('X') # Create a symbolic variable sym.init_printing(use_unicode=False, wrap_line=True) Now that we have the closed form. expression, we may use the chain rule to express the derivative of gn. In [ ]: dydx_symbolic = sym.diff(Y, X) dydx_symbolic Out[ ]: 64X (1 − X)(8X − 4)(−16X (1 − X)(−4X (1 − X) + 1) + 1) + 64X (1 − X)(−4X (1 − X) + 1)(−16X (1 − X)(8X − 4) + 16X (−4X (1 − X) + 1) ⋅ (1 − X)(−4X (1 − X) + 1)) − 64X (−4X (1 − X) + 1)(−16X (1 − X)(−4X (1 − X) ⋅ (1 − X)(−4X (1 − X) + 1)(−16X (1 − X)(−4X (1 − X) + 1) + 1) Question 2 (3 points): Compute the exact derivative of at by evaluating the symbolic expression. In [ ]: # hint: read about subs/evalf here: https://docs.sympy.org/latest/modules/numer ### Your Code Here ### As it is evident the closed form. expression for dydx_symbolic is unwieldy. We can make it more efficient by expanding and collecting all the terms. The goal is to represent the derivative as a natural polynomial. Yet this is another way of computing the derivative. Question 3 (3 points): Using the sympy documentation page expand/simplify the closed form. expression of gn(x) and print its derivative as a symbolic expression. In [ ]: ### Your Code Here ### polynomial = ... dydx_symbolic_simplified = ... dydx_symbolic_simplified # This should express a polynomial of degree 2^n - 1. Ellipsis Out[ ]: Now that we have a compact closed form. for the derivative we can evaluate it efficiently at various values of x. Question 4 (2 points): Evaluate dydx_symbolic_simplified at x = 1. In [ ]: ### Your Code Here ### Chain Rule and Autodiff As we've seen in the previous part, symbolic engines computes the derivative using some closed form. expression of the function. In particular, sympy simplified the recursive operation and expressed the function gn in terms of basic operations over the symbolic variable X. On the other hand, Autodiff does not necessarily simplify these operations and apply chain rule directly to the latest operation either in forward or backward mode. Question 5 (4 points): Using the recursive representation gn (x) = f(gn−1(x)) calculate the derivative of gn as a function of gn−1 and g'n−1. (To be clear, this part is not code; write the mathematical expression below) '### Your Answer Here ###' This implies that we can compute g'n(x) in the forward mode if we could compute gn−1(x) and g'n−1 (x) in the forward mode. Indeed, this is possible by applying the same logic to g'n−1 (x) and compute it based on gn−2(x) and g'n−2 (x); In other words, by augmenting the forward computation graph g1 → g2 ⋯ → gn with (g1,g'1) → (g2,g'2), ⋯ → (gn,g'n) we can compute the derivative in a forward pass. Question 6 (8 points): Modify the functions op and operation with their counterparts op_with_grad and operation_with_grad using the logic explained above to compute the derivative of gn. Check your function by evaluating it atx = 1. In [ ]: ### Your Code Here ### def op_with_grad(inp, grad): # inp could be any function of x. # grad is the corresponsing derivative of inp with respect to x. # The function should return a tuple where the first element is the `op` appl # and the second element should be the gradient of op(inp) with respect to x. pass def operation_with_grad(inp, n = 1): # inp represents x here. # This function should output a tuple where the first element is the value of # and the second element should be the derivative evaluated at inp, i.e. g_n^ pass operation_with_grad(1.0, n = 3) Fortunately, torch can do last part for us! As was discussed in class, torch interprets our code and builds up a computation graph using operations that knows their gradients already and complements each operation with their backward gradient; in order to compute the gradient it follows the following backwards path where g0(x) = x is the identity function. Question 7 (7 points): Use PyTorch to calculate the derivative of g_n(x) with respect to x In [ ]: def torch_op(inp): return 4 * inp * ( 1 - inp ) def torch_operation(inp, n = 1): temp = inp for i in range(n): temp = torch_op(temp) # apply this operation n times return temp In [ ]: x = torch.tensor(1.0, requires_grad=True) #### Your code here ##### # define the function with n=3 iterations as before, and use pytorch to autodif df_dx =... print(df_dx.numpy()) In [ ]:
Project of Module EG2303: Automatic Speech Recognition System October 2024 0.1 Description The goal of this group project is to provide students with hands-on expe- rience in the principles and concepts underlying signals and systems. The students are expected to develop an automatic speech recognition system designed to recognise spoken numbers. The emphasis in this project will be on the manual implementation of noise filtering and feature extraction in the frequency domain, without relying on external libraries. The speaker will be always the one in Google Translator (www.translate.google.com), and the digits of each number are always pronounced separately (this can be obtained by using spaces between digits). The system should receive a voice pronouncing a 10 digit number and show the whole number in the text format. As an example, when the input is “two three nine zero” in the voice format, the output should be “2 3 9 0” in the text format. The system should be able to work correctly even when the input voice is noisy. The developed systems will be tested during the presentation session in a lecture room in real time. Note that the development of such an automatic speech recognition sys- tem can normally be divided into several key steps: (i) data collection (recording a 10 digit number pronounced by Google Translator and sav- ing the recording in an audio format), (ii) preprocessing the data (trimming the audio file and splitting it into 10 diferent signals), (iii) feature extraction (implementing a noise cancelling filter and obtaining a feature variable in the frequency domain for each signal), and (iv) classification (comparison of the feature variable of each signal with those in a voice bank, and identifying the correct digit). 0.2 Groups There will be 9 groups each consists of 6 members. Groups are allocated randomly, and the list of the members of each group will be available on Moodle. The members of each group are expected to participate equally in the project. Each group must work independently, and any exchange of information between diferent groups is forbidden. 0.3 Assessment Each group is supposed to prepare a report and present its developed system so that its correctness can be assessed. The report should be maximum 5 pages and the presentation time will be limited to maximum 10 minutes. Note that this project accounts for 50% of the final mark of each student. The performance of each group will be assessed based on the following cri- teria: • Clarity and Coherence (10%) • Equal Participation (10%) • Comprehensiveness (40%) • Correctness (40%) Note that, each member of a group will be able to influence the marks of other members in the same group through a peer review process, described in the attached document. 0.4 Deadline The deadline for submission of reports will be 9th of December, 2024 (i.e., the date of the last lecture of the module). Each group must make sure to submit its report on Moodle by the deadline. The presentations will be held on the same date during the lecture session.
Fundamentals of mathematics and statistics - Assessment 1 This assessment is worth 20% of your final grade for the whole unit and 20 marks are available. It aims to test your understanding of univariable distributions, sampling distribution and hypothesis testing. Your report should be a maximum of 10 pages (font size at least 11, standard margins), including any R code and references that you include. · All solutions should be clearly set out with any hypotheses carefully set up and described. · You are encouraged to provide your R code to document your work. However, statistical outputs should be interpreted and described, with some demonstration of translating back to the problem domain. Relation to previous literature (etc) is not required. · The open-ended nature of the assignment means that it is possible to make some choices during the analysis – mark scheme will be flexible to reflect this. · Please present as you would a report by ensuring the submission has a structure and you are not just providing screenshots (or copy and paste) of code and results. The assessment comprises three questions. You are working for a local organisation that has undertaken a diabetes screening programme. Individuals were invited to attend clinics undertaking the programme over a month period and you have been given the data collected for each individual visit. You are asked to produce a report summarising the data collected and to complete 2 tasks; 1. Please provide a summary of the data, with the aid of graphs and tables (5 marks) 2. Another screening programme took place last year in a different region and they found 14% of individuals had diabetes and the average BMI and age of those attending clinics were 28kg/m2 and 55 years, respectively. They also found a significant relationship between smoking and diabetes. Investigate if your data support these results and provide possible explanations to your findings. (15 marks) You have been provided with a dataset (FMS_assessment1_data) from the screening programme that contains baseline information of patients collected during the screening visit and each patient’s diabetes diagnosis. Dataset information: The dataset contains the following variables, as described here. Diabetes_012 Weather the patient was diagnosed with diabetes (0 = no-diabetes, 1 = diabetes) HighChol Whether the patient has high cholesterol (0 = no, 1 = yes) CholCheck Whether the patient had cholesterol checked (0 = no, 1 = yes) BMI Body mass index (kg/m2) Smoker Whether the patient is a smoker (0 = non-smoker, 1 = smoker) Stroke Whether the patient has ever had a stroke (0 = no, 1 = yes) Sex Sex of the patient, binary (0 = female, 1 = male) Age Age of patient in years MentHlth Number of days in the past 30 days the patient says they have suffered with mental health (including stress, depression, etc) Additional Hints: - Don’t forget to conduct an exploratory/descriptive analysis. - Please showcase what you have learnt, but DO NOT simply perform. every test you can. Carefully think about the assumptions and which tests are appropriate to perform. when. - When analysing the data, you should carefully think about any hypothesis and assumptions - You should structure your report so that it is easy to follow what you have done, and the results of your analysis. - Note: the dataset and scenario are entirely fictitious, and so you are not expected to provide background motivation/reading within your report, BUT you should consider the results from the fictitious scenarios presented. Mark scheme information: Approximately half of the marks are allocated for correct and appropriate analysis, with the other half being allocated for clear descriptions and justifications of the analyses, their interpretation, discussion of the implications, and presentation. It is particularly important that you clearly explain what it is you are doing, your assumptions, and conclusions. Simply presenting code and output with no discussion and explanation is inadequate and will result in a fail. The below table gives an indication of how the 50 marks available will be awarded: Mark (% of the marks available for question 1) Descriptor 0-19% Analysis makes little or no attempt to address the brief, is generally inaccurate and is hard to follow. Conclusions are either nonexistent or incorrect. 20-29% Analysis addresses the brief in a limited way but is generally inaccurate. Report is generally incoherent. Conclusions often incorrect, and no attempt is made to discuss assumptions of analysis. 30-39% Analysis addresses only parts of the brief and is often inaccurate. Report is generally incoherent. Conclusions often incorrect, and no attempt is made to discuss assumptions of analysis. 40-49% A partly correct analysis that has addressed at least some of the brief. Report lacks coherence on the whole. Conclusions are sometimes incorrect, and little or no attempt is made to discuss assumptions of analyses. 50-59% A mostly correct analysis that has addressed the brief. Report is coherent but may be hard to follow in places. Conclusions and interpretation generally correct but superficial. Some limited effort is made to discuss assumptions of the analysis. 60-69% Analyses are correct besides minor errors and typos. Report is coherent throughout. Conclusions and interpretation are correct but lack depth. Reasonable discussion of assumptions and limitations of analysis. 70-79% Analyses are correct, besides minor errors and typos, and address the brief in considerable detail. Conclusions and interpretation are correct and detailed. Comprehensive discussion of assumptions and limitations. 80-89% Analyses are correct, address and often go beyond the brief, demonstrating independent thought and judgement. Conclusions and interpretation are correct and detailed. Comprehensive discussion of assumptions and limitations. 90-100% Marks in this category reflect a report of publishable quality with advanced insights and going well beyond the brief. Reports may demonstrate novel and innovative approaches to address the brief, and interpretations and conclusions will demonstrate a deep understanding and judgement. Mapping of the Assessment to the Module Intended Learning Outcomes This assessment tests the following intended learning outcomes of the unit: Category of outcome Students should be able to: A. Knowledge and understanding LO1: B. Intellectual skills LO5: C. Practical skills LO8: D. Transferable skills and personal qualities LO9
Assessment Brief 2024/2025 Individual assignment Information Course Code ACCFIN5249_1A Course Title Capital Markets and Investment Weighting 25% Question release date 30 September 2024 Submission date: 21 November 2024 Grades and Feedback to be released on: 12 December 2024 Word limit 2,000 words (+/-10%) Action to be taken if word limit is exceeded The first 2,000 words (+ 10%) will be considered for marking QUESTION You are asked to set up and manage a £100 million (+/- 10%) equity fund which is only eligible for the staffs of the University of Glasgow to invest. You are at liberty to design the fund, except for the following mandatory requirements: • The fund must contain 20 equities • All reporting must be expressed in pound sterling Q1. Prepare a fund fact sheet for the fund, please see the following examples as references:AXA Framlington Global Technology Z Accand Aviva Investors Global Equity Income Fund 2 GBP Acc. The fund fact sheet should contain: (40%) • Fund objective • Management strategy • Overall risk measurement • Geographical distribution of investments • Fund breakdown by industrial sector • List of the 20 equities invested • Benchmark • Monthly return and volatility of the fund and the benchmark over the period 1 September 2023 -30 September 2024 • Risk for the fund • Key facts, e.g., management fee *You are encouraged to make good use of tables and graphs *You are encouraged to obtain data, construct, manage and evaluate your investment portfolio with credible sources such as Bloomberg, Datastream and Yahoo! Finance *Your grade will not be affected by your fund’s performance. Q2. After reading the fund fact sheet, your line manager asked two questions, please prepare for a response letter for the following two questions. (60%) • Could you summarize and critically assess the factors driving performance of the fund over the period 1 September 2023 -30 September 2024? (30%) • How would you rebalance your portfolio to improve the fund’s performance in the following year? You must make at least two adjustments and must clearly explain the reasons for the adjustments. (30%) *Your grade will not be affected by your fund’s performance. ASSESSMENT CRITERIA The work will be assessed based on the answers in Q1-2. The fund’s performance is not part of the assessment criteria. For Q1, students are expected to present the key information, as outlined in the question, of the fund accurately and effectively. The best marks will be awarded to students who make excellent use of tables and graphs. The format and structure of the fund fact sheet is expected be like the examples given in the questions. For Q2a, students are expected to critically assess and evaluate the performance of the fund. The best marks will be awarded to students who are able to precisely highlight the factors leading to the underperformance or overperformance over the benchmark. For Q2b, students are expected to rebalance the fund portfolio responding to the performance of the fund and the change in the investment environment over the past month. The best marks will be awarded to students who are able to provide rationale and excellent explanations on the adjustments.
Assignment Title 1. Assignment Aims Learn how to analyse, measure, and design op-amps. 2. Learning Outcomes: Knowledge and Understanding Outcomes 2. Examine, compare, and evaluate appropriate mathematical methods and models required to analyse analogue circuitry and solve problems relating to analogue systems. Ability Outcomes 3. Examine and extract data for an unfamiliar problem before designing and developing a solution, formulating an appropriate test strategy (using dc and ac analysis) and assessing the success and quality of the solution. 4. Analyse and design circuit/systems to a required specification. 3. Assessment Brief Introduction: As this is an individual (not group-based) assignment, by the end of this assignment each student is expected to perform. the tasks given below: • For the operational amplifier (op-amp) in Fig. 1 on the next page, perform. an approximate DC analysis to calculate the DC collector currents and voltages in the circuit. Find the quiescent power dissipation for each transistor and the voltage gains of each stage. Also, calculate the AC characteristics of the circuit, such as, total voltage gain (dB), input and output impedance, Common Mode Rejection Ratio (CMRR) in dB, and maximum output voltage swing. • Using NI Multisim software, simulate the op-amp circuit in Fig. 1 and calculate the DC operating conditions and AC operating characteristics. Compare the results, and comment on any discrepancies found. • Make improvements to the simple op-amp of Fig. 1 (e.g., using current mirrors, cascaded differential amplifiers, Darlington pairs, etc.) to upgrade its performance. Simulate the new circuit(s) in NI Multisim and calculate the new and improved values of voltage gain (dB), input and output impedance, CMRR (dB), etc. • Write a report which includes the overall findings, analysis, and the results of activities performed above. Procedure Op-amp DC/AC Analysis (25%) Fig. 1 shows a very simplified version of an integrated operational amplifier (op-amp) with only 5 transistors. • Perform. an approximate DC analysis (assuming β >> 1, so that the base current of every transistor can be ignored) to calculate the DC collector currents and voltages in the circuit. • Find the quiescent power dissipation for each transistor and the voltage gains of each stage. • Calculate the AC characteristics of the circuit, such as, total voltage gain (dB), input and output impedance, CMRR in dB (Common Mode Rejection Ratio), maximum output voltage swing. Assume β = 400, and Early voltage = 200 V. • Ideally, an op-amp should have a very high input impedance (MΩ), low output impedance (10-100 Ω), very high voltage gain (> 100,000 = 100 dB), and a very high CMRR (> 100dB). Is this the case with this simple circuit? Figure 1: A simple operational amplifier circuit. Simulation (30%) • Simulate op-amp circuit in Fig. 1 with NI Multisim and calculate the DC operating conditions and AC operating characteristics as done earlier with theoretical calculations. • Compare the theoretical and simulation results, and comment on any discrepancies found. Design improved op-amp circuit (35%) • Using the knowledge gained in module lectures, design an improved op-amp circuit (e.g., use current mirrors, cascaded differential amplifiers, Darlington pairs, etc.) to upgrade the performance of op-amp circuit given in Fig.1. • Do not use resistors larger in value than 50 kΩ due to area limitations and prefer to use active components (easier to realise in IC technology) over passive ones. • Simulate the new circuit(s) in NI Multisim and calculate the new and improved values of voltage gain (dB), input and output impedance, CMRR (dB), etc. Report Writing (10%) Write a report which includes the overall findings, analysis, and the results of activities performed above. 4. Marking Scheme Criterion 1: Op-amp DC/AC Analysis Weighting (25%) Op-amp DC analysis to calculate the DC collector currents and voltages in the circuit 10% Quiescent power dissipation for each transistor and the voltage gains of each stage in op-amp circuit 5% AC characteristics of the given op-amp circuit 5% Comparison with ideal op-amp circuit 5% Criterion 2: Simulation of op-amp circuit Weighting (30%) NI MultiSim schematic of op-amp circuit 10% DC analysis and AC operating characteristics with NI MultiSim 10% Comparison of theoretical and simulation results, and comment on any discrepancies found 10% Criterion 3: Improved op-amp design Weighting (35%) Design of improved op-amp circuit 15% NI MultiSim schematic of improved design of op-amp circuit 10% AC characteristics of the given op-amp circuit 10% Criterion 4: Report writing Weighting (10%) Report structure and presentation 5% Conclusions and future work 5%
COMP3153/9153 Algorithmic Verification Final Take-Home Exam Term 2, 2023 Question 1 (10 marks) Consider the following automaton C: For each of the following CTL* formulae, determine whether C satisfies it or not. Justify your answers. (a) (F q) → (Gp) (b) (F Gp) → (G F q) (c) E(G (F ¬p)) (d) AG (p → AX q) (e) G ((X q) ∨ (pUr)) Question 2 (10 marks) Consider the automaton C from the previous question. The labelling function defines an abstraction, A, of C where the states of A are given by the values of the atomic propostions p, q, and r. For example, because L(q1 ) = L(q3 ), states q1 and q3 would map to the same abstract state. (a) Draw A. (b) Are A and C bisimilar? Justify your answer. (c) Give an ACTL formula, φ, that holds in one of A or C and does not hold in the other. (d) Refine the abstraction of A based on φ . That is, find a model A′ such that A ⊑ A′ ⊑ C and A′ |= φ if and only if C |= φ . Try to find a model A′ ≠ C, or explain why this is not possible. Question 3 (10 marks) For each of the following pairs of LTL or CTL formulas, prove or disprove whether they are equivalent. That is, prove or disprove for all models M, M |= φ 1 if and only if M |= φ2 (a) φ 1 = G (F (pU q)) and φ2 = G (F q) (b) φ 1 = (Fp) Uq and φ2 = (Fp) ∧ (F q) (c) φ 1 = AX E(pU q) and φ2 = E(pU (AX q)) ∨ q (d) φ 1 = X (pU (Gq)) and φ2 = G ((Xp)U (Gq)) (e) φ 1 = (Xp)U (X (pU q)) and φ2 = (X (pU q))U (X q) Question 4 (15 marks) We extend LTL by a Before operator. Its semantics are given as ρ |= φBefore ψ ⇐⇒ (∀k ≥ 0. (ρ[k] |= ψ) =⇒ (∃j. 0 ≤ j ≤ k ∧ ρ[j] |= φ)) . (a) Briefly describe in your own words the meaning of Before. (b) Prove that the Before operator does not extend the expressiveness of LTL. That is, give an LTL formula that is equivalent to φBefore ψ and argue their equiva- lence. (c) Prove the following: i. ((pBefore q) ∧ q) implies p; ii. p Before (q ∨ r) is equivalent to ((pBefore q) ∧ (pBefore r)). (d) Give a B¨uchi automaton that accepts the words described by the formula p Before q (e) How could you extend the LTL model checking algorithm to handle the Before operator directly? (f) If we add the Before operator to the path formulae of CTL in the obvious way, explain how to extend the CTL model checking algorithm to handle E(·Before ·) subformulae directly. Question 5 (15 marks) (a) Give OBDDs for the following boolean functions: i. f(x,y, z) = (x ∧ y ∧ z) ∨ (¬x ∧ ¬y ∧ ¬z) ii. f(x,y, z) = (x ∨ y) ∧ (y ∨ z) ∧ (z ∨ x) (b) Consider the following OBDD describing the boolean function T(x0 , x1 , x0(′), x1(′)): Here a solid line indicates the Boolean value true, and a dashed line the value false. T(x0 , x1 , x0(′), x1(′)) defines a transition relation of a four state automaton where the states are given by pairs of boolean values (i.e. Q = {00, 01, 10, 11}). with a transition between q and q′ if and only if T(q, q′ ) = true. Draw this automaton. (c) How can you use OBDDs to solve the SAT problem? That is, given an OBDD for boolean formula φ , Dφ , how can you find an assignment of true/false to the variables of φ which makes φ true? (d) Prove or disprove: If D1 and D2 are OBDDs on n variables with O(n) nodes, then D1 ⊗ D2 has O(n) nodes. Question 6 (10 marks) Consider the following decorated pseudo-code program written in a simple imperative programming language: [ a := x +b ]1 ; [ y := a *b ]2 ; while [y > a + b ]3 do ( [ x := a +1]4 ; [ y := a +b ]5 ; if [( a +1 > a +b )]6 : [ a := x +1]7 ; else : [ x := a +1]8 ; ); (a) Give a control flow graph for this program. The code decorations given by [ ..]# indicate the eight critical states to include in the graph. (b) Give the table of gen and kill sets for an live variable analysis of the program. (c) Compute the result of the live variable analysis, using backwards may analysis. Question 7 (10 marks) Consider the LTL formula: φ = G (pU q) Let Σ = 2{p,q} = {∅ , {p}, {q}, {p,q}} (a) Give a B¨uchi automaton that accepts the words (in Σω ) described by φ . Justify your construction. (b) Give a B¨uchi automaton that accepts the words (in Σω ) described by ¬φ . Justify your construction. (c) Give a safety property Ps and aliveness property Pe such that the set of behaviours described by φ is equal to Ps ∩ Pe. Question 8 (20 marks) Consider the network of timed automata depicted in Figure 1. Figure 1: System of Traffic Lights The automata describe a system of traffic lights. The syntax is as follows: green statements, e.g. x==10, are guards; blue statements, e.g. y=0, are assignments; purple expressions, e.g. x42. i. At no time pedestrians walk (across the street) while cars have green lights. ii. The yellow light is never turned on for more than 10 seconds. iii. After pressing the button a pedestrian is able to walk across the street within 40 seconds. (d) Which of the properties in (c) hold for the deadlock-free system; which do not hold. Justify your answers. (e) Modify the system in a way that all properties of (c) are valid. (f) Extend the model by a “pedestrian cheat”: If a user presses the button three times within 3 seconds the system switches to a state where cars have to stop (state red) and pedestrians can walk.
CIS 5371 – Cryptography Fall 2024 Homework 3: Deadline Thursday 11/14 (Implementing padding-oracle attack) (180 points) Your task here is to implement the padding-oracle attack. The target server is linprog2.cs.fsu.edu (IP address is 128.186.120.158), and the port number is 31537. Due to some administrative restriction, to have a socket connection to this server, you need to run your code at the same machine. (You can log into this server using your linprog account.) The server maintains a secret string M which your program needs to recover. Currently, this string is (without the quote and case sensitive) “C0ngr4tul4ti0n5!!! Y0u 5ur3 kn0w ur crypt0!!! :)”, and thus the byte length is 48. However, when I test your programs, I may change it to a different string, of a different length. You should not assume that the byte length of the secret is a multiple of 16. Interacting with the server. The server provides two types of requests. If you want an encryption of P∥M, for some chosen prefix P, then you need to send a request of the form. “-e P”. The server will send you a ciphertext C of P∥M under the TLS 1.0 encryption on a random IV L. If you want to send a ciphertext core C with IV L for validation, you need to send a request of the form. “-v C L”, and the answer is either “Valid” or “Invalid”. Your goal is to recover the message M, and output it to stdout. For an example of how to interact with the server, see the Python script. client.py in the course website. When I run python3 client.py in my terminal, the script. waits for me to enter requests to the server. If I type -e abcdef0123456789 (meaning my prefix is the hexadecimal string abcdef0123456789), I get back b‘Encryption: 80 49 8d 8e 57 ...6a 3d de ’ IV: b‘a5244e79b9f94b4f5634a8b00e06e46c’ -E abcdef0123456789 Here the cipherext core C (in hex encoding) is 49 8d 8e 57 ...6a 3d de that is 80-byte long, and the IV L (in hex encoding) is a5244e79b9f94b4f5634a8b00e06e46c If you want to encrypt with the prefix as the empty string, just use “-E”. Likewise, if I type -v aaaaaa 40beed9c1b5bfcbc997bc025a42b4c0a, I get back ‘Invalid’. This means the ciphertext core aaaaaa with IV 40beed9c1b5bfcbc997bc025a42b4c0a is not a valid ciphertext. For another example of how to write code to connect with the server, see the file connect.py in the course website. (The latter is written in Python2, so you’ll need to run python connect.py.) Another testing server. To help you to test your program on fragmentary secrets, I also set up another server at the same IP address, but the port is now 31536, and the secret is “Hello World”. Deliverables. Upload to Canvas a zip file containing your source code, which includes a README.txt that informs me how I should run the program.
Course Syllabus ECO202Y5Y-LEC 0101 - Macroeconomic Theory and Policy Department of Economics 2024-2025 Course Objectives and Learning Outcomes The goal of this course is to provide a comprehensive introduction to the subjects and methodologies of modern macroeconomics. The covered topics include the measurement of macroeconomic indicators, long-term economic expansion, inflation, recent labor market trends, factors behind the business cycle, and the efficiency of fiscal and monetary policies in addressing them, along with international trade and currency exchange rates. In addition to the theories and case studies covered in lectures, students should expect to learn how to navigate FRED database, which is a popular source for macroeconomic data. Course Materials Required: The required textbook for this course is Macroeconomics, by Charles I. Jones, 6th edition. Suggested: Other recommended readings may be announced and posted throughout the course Course Outline General Outline of Topics Covered: First Half: Chapter 1: Introduction to Macroeconomics Chapter 2: Measuring the Macroeconomy Chapter 3: An Overview of Long-Run Economic Growth Chapter 4: A Model of Production Chapter 5: The Solow Growth Model Chapter 16: Consumption Chapter 17: Investment Chapter 6: Growth and Ideas Chapter 7: The Labor Market, Wages, and Unemployment Chapter 8: Inflation Second Half: Chapter 9: An Introduction to the Short Run Chapter 10: The Great Recession and the Covid-19 Pandemic: A First Look Chapter 11: The IS Curve Chapter 12: Monetary Policy and the Phillips Curve Chapter 13: Stabilization Policy and the AS/AD Framework Chapter 14: Financial Crises and the Short-Run Model Chapter 15: DSGE Models: The Frontier of Business Cycle Re-search Chapter 18: The Government and the Macroeconomy Other topic if time permits Evaluation The course grade consists of term work worth 75% and a final exam worth 25%. Term work includes three term tests, two written assignments and quizzes. ndweek)Mid-term51% (17%each)MT1: November 9, 2024MT2: January 25, 2025MT3: March15, 2025Final Exam:34%Spring final exam period
PSYC51A Data Analysis Report # 1 (Descriptive Statistics) 1. (22.5 pts) Research Question: Does the type of judgment requested about a dynamic social interaction affect the accuracy of interpersonal perception? 1.1 (3.5 pts) i) Name the independent variable for this research question (1 pt); ii) Name the levels of the independent variable (1.5 pt; iii) Name and explain the scale on which the levels of the independent variable are defined (1 pt). i) (1 pt) The type of judgment requested about a dynamic social interaction. ii) (1.5pt; .5pt per level). · Dominance = status, judge the status or rank of an individual in a social interaction · Affiliation = kinship, judge the type of relationship an individual have with others · Deception = trustworthiness, judge the trustworthiness of an individual iii) . (1 pt). The levels of the independent variable are defined on a nominal scale. This is because the three levels are qualitative, representing distinct categories rather than numerical values. 1.2 (1.5 pts) Give an operational definition for each level of the independent variables. Dominance (construct): In the interaction, who has higher authority over others and which social status does the individual have? Affiliation(construct): In the interaction, what is the relationship between them? Who is more polite and friendly, and who is more willing to cooperate with others? Deception(construct): In the interaction, who is more trustworthy and who is more deceptive? 1.3. (3 pts) i) Name the dependent variable of this research question, in terms of the construct being studied (1 pt). ii) Give the operational definition of the dependent variable for this research question (1pt). iii) Is the dependent variable discrete or continuous; what type of scale of is it measured on? (1 pt) i) (1 pt). The accuracy of the dynamic social perception. ii) (1 pt). The number of questions that are answered correctly or incorrectly. iii) (1 pt). It is discrete because it consists of whole number values. It is measured on a ratio scale since it has a true zero point, and the intervals between values are equal. 1.4 (1 pt) What kind of experimental design is this, and why? This is likely a pre-experimental design because there is no random assignment or a control group. Participants are exposed to a specific condition without randomization, which means it will be hard to establish strong causal relationships between the independent and dependent variables. 1.5 (9 pts) i) Submit a descriptive statistics table for all conditions (levels of the IV) containing the mean, median, and mode, standard deviation, interquartile range, and quartiles (3 pt); ii) For each condition, submit a frequency distribution table (3 pt); iii) For each conditions, submit a histogram (3 pt). Every individual table or plot should be labeled sufficiently to be understood without any accompanying text (as if it were standing alone), and all the plots should be arranged in a group to provide the best visual summary for the reader. NOTE: start this question on a new page so there will be room for all your plots/tables to show as a group on the same page. i) Descriptive Statistics: IPT Questions Correct (0-3) by Judgment Type N Mode Median Mean St. Dev. IQR 25% 50% 75% Dominance 99 1.000 1.000 1.162 0.695 1.000 1.000 1.000 2.000 Affiliation 99 1.000 1.000 1.121 0.798 1.000 1.000 1.000 2.500 Deception 99 2.000 2.000 1.818 0.825 1.500 1.000 2.000 2.000 ii) IPT # Correct (0-3) by Judgment Type Dominance Affiliation Deception #Correct Frequency #Correct Frequency #Correct Frequency 0 17 0 21 0 5 1 49 1 50 1 29 2 33 2 23 2 44 3 0 3 5 3 21 Total 99 Total 99 Total 99 iii) Dominance Affiliation Deception 1.6. (3 pts) For this research question, we expect subjects to get 1 IPT question correct by chance (guessing the answer to the questions without any information about the scenes). This is because there is a 1/3 chance of guessing each scene correct and each subject saw three scenes for each type of judgment, so 1/3 * 3= 1.) Using this calculation, indicate how much each observed mean (# correct) was above or below chance (answer must be correct to 2 decimal places), and what percent of subjects (in the frequency distributions) scored above chance (answer must be correct to 1 decimal place): Dominance: Mean was 0.16 chance; 33.3% of scores were above chance. Affiliation: Mean was 0.12 chance; 28.3% of scores were above chance. Deception: Mean was 0.82 chance; 65.7% of scores were above chance. 1.7 (1.5 pts) What is the best conclusion you can reach about the research question based on the descriptive statistics in the previous questions? Be sure to refer directly to the research question. Use at least one descriptive statistical comparison to support your conclusion. The accuracy of the dynamic social perception depends on the type of judgment. Participants were more accurate in judging deception, as shown by the mean score for deception (1.818), which is significantly higher than the means for dominance (1.162) and affiliation (1.121). Additionally, 65.7% of participants scored above chance for deception judgments, compared to only 33.3% for dominance and 28.3% for affiliation, indicating higher accuracy when assessing deception. The distribution of scores for dominance and affiliation is slightly positively skewed, with means (1.162 for dominance, 1.121 for affiliation) higher than their medians, which means lower accuracy in these judgments. Deception showed a slight negative skew, with a mean of 1.818 and a median of 2, meaning that participants were more accurate in detecting deception. All distributions fit the normal distribution model, indicating consistent results across participants. 2. (19.5 pts) Research Question: Does the accuracy of judging dynamic social interactions differ depending on the channel of non-verbal information? 2.1 (2 pts) Name i) the independent variable for this research question (1 pt), and ii) the levels of the independent variable (1 pt). Answers may be in terms of either constructs or operational definitions. i) (1 pt). channel of non-verbal information ii) (1 pt). transcript. (verbal channel) and audio visual (verbal and non-verbal channels) 2.2 (1 pt) Assuming subjects were randomly assigned to the levels of the independent variable, what kind of experimental design is this, and why? True experiment, because participants are randomly assigned to conditions. And the researcher can manipulate the transcript. and full-channel treatments, which ensures that each participant receives one of the two conditions. 2.3 (6 pts) i) Submit a descriptive statistics table for all conditions (levels of the IV) containing the mean, median, and mode, standard deviation, interquartile range, and quartiles (2 pt); ii) For each condition, submit a frequency distribution table (2 pt); iii) For each condition, submit a histogram (2 pt). Every individual table or plot should be labeled sufficiently to be understood without any accompanying text (as if it were standing alone), and all the plots should be arranged in a group to provide the best visual summary for the reader i) ii) iii) Transcript. Audio-Visual 2.4 (3 pts) i) Copy the sample histograms from the previous question here (1pt). Transcript. Audio-Visual ii) Copy a normal curve template onto each of the sample histograms, resizing so the peak is approximately aligned with the sample mean and the width approximates the sample standard deviation (don’t worry about the height) (1 pt) . iii) Copy those curves onto the common scale below, aligning their peaks with the scores on the common scale which match the sample means and their widths correspond to the interval on the common scale which matches the standard deviation of the sample. (1 pt) 2.5 (2 pts) Rate how well each sample distribution fits the model of a normal, bell-shaped distribution (good , fair, poor), and justify your rating in terms of unimodality, asymptotic decrease of frequency with distance from the mode, symmetry (or skewness), and outliers (Note: Skewness assessments only need to be made on the basis of visual inspection; no statistics are needed). Audio-Visual: (1 pt) This is a good fit because it is unimodal, it has a clear peak at the center and symmetrical distribution, and no outliers. Transcript. (1 pt) Transcript. This is also a good fit, because it is unimodal, it shows symmetry with a central peak and no outliers. 2.6 (2 pts) Submit an interval plot (JASP command) comparing the levels of the IV. The plot should be well labelled and formatted for easy visualization and stand-alone comprehensibility. 2.7 (2 Pts) Analysis of descriptive statistics. i) What scores are in the middle 50% of each group? Transcript. 2.0 - 3.0; A-V: 4.0 - 6.0 ii) % of Transcript. subjects scoring below the middle 50% of the A-V group 78% % of Transcript. subjects scoring above the middle 50% of the A-V group 0% % of A-V subjects scoring below the middle 50% of the Transcript. group 0% % of A-V subjects scoring above the middle 50% of the Transcript. group 93.88% 2.8 (1.5 pts) What is the best conclusion you can reach about the research question? Use at least two descriptive statistical comparisons presented in previous sub-questions to support your conclusion. Make sure you address the research question. (1 pt) In conclusion, there is a large difference in performance between the Transcript. and Audio-Visual groups, which means that the mean and standard deviation are more informative than measures like the median, mode, or range. The mean correct judgment for the Audio-Visual group (5.27) was higher than for the Transcript. group (2.96), which means audio-visual have better accuracy. Additionally, the range for the middle 50% of the Audio-Visual group (4-6) was wider compared to the Transcript. group (2-3), suggesting that most participants in the Audio-Visual group outcompete those in the Transcript. group.
Graphical Models for Complex Health Data (P8124 Fall 2024) Due date: Friday December 20th, 8pm FINAL PROJECT DESCRIPTION This is an open-ended data analysis project, where you can gain some experience applying methods based on graphical models to real data. Several data sources are made available below. You may elect to use their own data only if you obtain permission from Professor Malinsky. (The student must also have permission to use the data — publicly available data is ok if the data is of high quality. If you want to use data not listed below, please email Prof. Malinsky by 11/29.) You should define your own analysis objective: what do you want to learn/estimate from this data? What scientific or inferential question do you want to address? You can use whatever methods you like, as long as they are related to the graphical methods we discuss in the course. There is a bit of leeway here — you need not limit yourself to methods we’ve directly discussed (e.g., if you want to use a graphical structure learning algorithm we have not mentioned, or an estimation procedure that we have not explicitly gone over in class, a different Monte Carlo sampling method, etc.), but whatever methods you use should be clearly related to the course material. (In particular: do not simply train a neural net, or use K-means clustering, or some other standard machine learning (ML) method you’ve learned outside this course and call it a day! This would not be acceptable. You may, however, use other ML methods not discussed in class in conjunction with graphical methods.) You may use whatever software is publicly available to do your analysis, or implement things yourself. You will write a short paper in the style. of an ML conference paper. Remember to clearly introduce your problem, data, methods, approach, and results. In fact, a good way to organize your paper is into the following sections: Introduction, Data, Methods, Results, Discussion. (You do not need to discuss “related work” though you are welcome to use already published work as an inspiration, as long as you cite it. Pure re-implementations of already published work are to be avoided.) You should justify all your analysis choices — how you chose tuning parameters, why you chose certain parametric forms or model classes, etc. Make sure your work is reproducible, so someone using the same data could implement your method and achieve the same results. Requirements: 1) Specify, either based on background knowledge, problem design, or learning from the data, at least one graphical model that will serve as the basis of your analysis. It can be any kind of graphical model that you think is appropriate. Then, perform. some sort of “inference” task based on the model. Here I mean “inference” broadly construed: test some scientific hypothesis, estimate some scientifically-meaningful parameter, use the graph somehow for prediction/classification, perform. causal effect estimation (if appropriate), etc. 2) You should compare at least two approaches/methods/settings. That is, you should consider what someone else might do alternatively to your proposal, try it, and compare results. How you do this is up to you: the important thing is that you try more than one thing. 3) Write a paper using the formatting guidelines of the NeurIPS conference. Formatting guidelines and sample Latex document/style. file can be found here: https://neurips.cc/Conferences/2021/PaperInformation/StyleFiles (Note: do not anonymize your submission!) 4) Minimum length: 4 pages. Maximum length: 8 pages, including all tables and figures (not including references). You may include additional supplementary material if you wish but the grading will be based entirely on the content of the main paper, and supplementary material will probably not be examined at all. 5) A 500 word (approx) project proposal is due December 6th at 8pm (on Courseworks). Describe which data, methods, and software do you intend to use, and the goal of your analysis. This does not need to contain all the details, but it should be clear that you have a well-formed idea and a rough plan for executing it. The final project should also be submitted via Courseworks. Late submissions will not be accepted. Note: there is a list of GM-related software packages in R here: https://cran.r-project.org/web/views/ GraphicalModels.html This is of course not a complete list, new packages are added all the time and many are on Github or other sites. However, this list is a good place to start. AVAILABLE DATA 1) fMRI dataset. Download here: https://rutgers.box.com/s/imgbdaqhzlkbunf52ia8xfmiu97b05xh This data set contains fMRI brain scan results for individuals with Autism Spectrum Disorder (ASD) as well as “neurotypical” controls. It was taken from the Autism Brain Data Exchange: http://fcon_1000.projects.nitrc.org/indi/abide/abide_I.html Specifically, we have the Carnegie Mellon University dataset, where the age range of subjects is between 19 and 40. The preprocessing pipeline used is NIAK with 160 Regions of Interest (ROIs): http://preprocessed-connectomes-project.org/abide/Pipelines.html On the above website you can see the preprocessing steps performed by NIAK (last column in the comparison tables). And in the last section of the website you can see the description of the parcellation to get the ROIs: “Dosenbach 160.” There are two diagnostic categories of patients, 14 ASD individuals and 13 controls. The individuals all have the same number of ROIs/variables (columns), but potentially different number of samples (rows), because some samples are dropped by the data preprocessing to remove artifacts of head motion (akin to outliers), etc. The data was sampled every 2 seconds, and the size of the voxels is 3mm x 3mm x 3mm. There is a file called “phenotypic_CMU.csv” which has some metadata on the individuals in the data set, including which diagnostic category each individual belongs to (1=ASD, 2=control). Note: due to a bug there is an extra ROI (column 161) in this data which you should just remove before analysis. Also, the files are really .csv files but for some reason the file extension was dropped. You might need to manually rename each data file as “CMU a 005[….]1D.csv” so that your computer has an easier time opening and reading the files. Alternative: the fMRI data from the Dajani et al. (2017) paper is available here: https://github.com/cheninstitutecaltech/Caltech_DATASAI_Neuroscience_23/blob/main/07_20_23_day9_causal_modeling/ code/solutions/exercise3.ipynb (instructions described in a Python notebook) 2) Genetics data. The data used by Wang et al. (2016) in their “FastGGM” paper is available here: https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-1425/ Another source of genetics data is here: https://jhubiostatistics.shinyapps.io/recount/ Note: Use the “TCGA” data and focus on a specific tissue (e.g., “lung”). Note: the TCGA data is a relatively “clean” RNA-seq data which is easy to download (instructions on the above website). However, it is not very easy to understand if you don’t already have some familiarity with data of this type. So, if you have no experience with such data, I would probably advise against using it. 3) X-ray image data. The ChestX-Ray8 from NIH contains >100K chest x-ray images of >30K unique patients, along with radiologist labels to indicate 14 common pathologies of the thorax. The data (both raw images and labels/other metadata) are available here: https://nihcc.app.box.com/v/ChestXray-NIHCC GRADING RUBRIC Meeting the basic requirements: 40 points Clearly stated objectives: 10 points Appropriateness of methods for the stated task(s): 10 points Adequate description of the methods used (including all modeling choices, any tuning parameters, parametric forms, etc): 20 points Informative presentation of the results: 10 points Clear and understandable writing: 10 points (Creativity/novelty: up to 5 points extra) Total: 100 points
2024 T3 ADAD9113 Communication Skills for Creative Disciplines Assessment brief & Course overview Course Description: ADAD 9113 Communication Skills How articulate are you as a creative practitioner, designer or administrator? This course develops your ability to confidently engage with, interpret and express complex ideas using an appropriate mix of communication strategies tailored to creative disciplines. In addition to an appreciation of fundamental research protocols in creative disciplines, you will enhance your academic reading, writing, listening, and speaking skills to better express coherent intellectual arguments, towards developing your own academic 'voice'. Collaboration in and out of class, using a mix of communication tools, will model cross- disciplinary interaction with colleagues, peers, professionals and the public. Course Aims This course aims to develop a capacity to recognise and employ different modes of communication relevant to creative disciplines in advanced professional and academic contexts. Course Learning Outcomes CLO1 : Critically read, interpret and reflect on academic discourse across different creative disciplines CLO2 : Discuss and write using appropriate technical terminology and Information Communication Technologies to meaningfully engage in professional communication CLO3 : Identify and analyse English language forms, features, structures and referencing protocols appropriate for research, academic writing, critiques and/or artist statements CLO4 : Lead oral presentations and audience discussions using visual, verbal, and written information Assessment 1 - Interpretive Writing Task (20%) Assessment type: Writing (1000 words) Submission requirements: • Submit as ONE readable PDF to Turnitin via Moodle page • Due: Week 3, Friday 27 September 2024 (9pm) Assessment Overview: This assignment aims to encourage close, critical reading practices and to evaluate your writing skills. You need to clearly identify the focus of your response. For example, you can focus on selected sections, concepts or ideas in discussion during class time with your tutor. In response to the selected reading provided for your degree program specialisation (see Required Readings for Week 1) you are required to: 1. Collaborate in pairs or groups in class tutorials to identify key issues in the set text. 2. Discuss and take notes on the origin, influences and implications of these ideas. 3. Use your notes and answers to individually write your own interpretation of the reading. 4. Using academic English, identify and discuss the importance, meaning, and relevance of this reference. 5. Locate a contemporary art, design or media artwork in the public domain that is relevant to the selected reading (see Required Readings for Week 1). Identify it, providing a URL to the image and contextualising the selected art/design/media work. Note: Allocate roughly 350 words (i.e. one third of your text) to analysing how ideas from theory/research have been used or referenced by the selected artwork. How do the historical ideas in the reading speak to this new situation now? Assessment 2 - Oral Presentation and Paper (30%) Assessment type: Oral presentation and Paper Submission requirements: (1) an in-class oral presentation (Part A) (2) written documentation (Part B) • Presentations are due in tutorials during your allocated times in Week 4 or Week 5 • There will only be time for brief feedback in class, while more written feedback will be provided in Turnitin. • Submit your documentation (Part A + Part B) as a single readable PDF to Turnitin by Week 5, Friday 11 October (10pm) Assessment Overview: Assessment 2 includes two parts: 1. Part A: In-class individual presentation with maximum 5 slides (your presentation slides are added to Part B to make a single PDF for submission) 2. Part B: One-page PDF - Visual Summary/Digital Handout. 3. Submit your documentation (Part A + B) as a readable PDF to Turnitin, Friday Week 5 Part A: PRESENTATION IN CLASS Part A is a 10-minute presentation of engaging and coherently delivered visual, verbal and written content with a Powerpoint using maximum 5 slides total: including one slide with research references, and your final slide with two written questions to start a group discussion of your presentation. The discussion questions should be included in your 10 minute presentation time. The date and time for your online presentation will be confirmed in Week 3. It is essential that each person presents to their assigned group, and that you commit to being an engaged audience member as well. We have allocated your presentation and audience commitment within the one timeslot, so that you typically only have ONE time commitment over Weeks 4 and 5. This task requires you to synthesise visual, verbal and written information into a professionally coherent presentation. You may speak in response to shared material, using PowerPoint. The challenge is for you to communicate your message, manage the presentation delivery and engage your audience within the timeframe. The aim is to succinctly showcase the meaning and ideas behind your interests or creative works and practices, concluding with a brief group discussion. You will need to be organised, speak clearly and present yourself in a confident, honest and professionally relaxed manner. Part B: SUBMIT TO MOODLE A one-page digital handout of key messages for future reference that includes a clear connection between your chosen area of interest and current academic conversation/paper/research/creative outputs by artists/designers. Marks for Assignment 2 will be awarded for: 1. Interesting, academic, well-paced presentation that engages your audience 2. Synthesis of visual, verbal and written information presented within 10-minutes 3. Confident command and delivery of information in chosen format 4. Relaxed personal style. in leading group discussion with engaging handout provided. Assessment 3 - Exhibition Review (50%) Assessment type: Extended Writing Task (1500 words) Submission requirements: • ONE readable PDF document • Submit to Turnitin via the course Moodle page • Due: Week 10, Sunday 17 November 2024 (9:00 PM) Assessment Overview: In accordance within-class guidance, you are required to produce an exhibition review of a current exhibition you can visit in advance to complete this task. This can be in-person (recommended!) or a dedicated online-only, digital exhibition. In either case you must confirm your choice with your tutor, to make sure you are reviewing an appropriate exhibition. Note: You should select an exhibition of art, design, architecture, digital media or other creative work that is directly relevant to your own field/creative practice. Your review should engage with at least three specific art/design examples in the exhibition, perhaps considering aesthetic, historical, functional, technical and/or social aspects of the work, as well as discussing the conditions of its presentation and reception. Your writing will use a theoretical lens developed through your course readings and further research of the subject matter, artist/designers and exhibition context. However, the challenge of a review is to both describe and critical analyse your selected exhibition, from your particular perspective, in a limited number of words. In order to satisfactorily complete this assessment, you should: • Show proof of research, ideas, and evidence beyond that provided in the course readings • Implement effective review structures (not an essay!) • Integrate and reference academic sources appropriately and effectively, using a consistent referencing style. (e.g. Oxford, Harvard) • Compose and write a well-reasoned, presented and referenced critical review that advances an original, compelling opinion relevant to your creative practice.
UNIV*6020 - Meta-Analysis Module - Assignment (5%) 1. Propose a new meta-analysis, related to your research topic area/your area of interest. Specify the research question and break it down into the ‘PICO’ terms discussed in class. Define your search terms using well thought out Boolean logic. Devise and document your literature search strategy (what databases will you search, how will you hand search, what journals will you hand search, where will you look for key papers, etc)? (1%) 2. Using the dataset provided to you (file: ‘ UNIV 6020 MetaA module Data files – JE F24 – Student Assignment.xlxs’, ‘Data for question 2’ tab), perform. a meta-analysis (regression) (in SAS or R) on the data provided: (a) ignoring the study variable, (b) treating study as a fixed effect (you will need to use a contrast statement or hand calculate the average equation), and (c) treating study as a random effect. For (c), determine whether the random statement could be reduced or simplified (based on p-values, Information Criteria and plots) and if so, do so. Extract the predictive equations (and their parameters, +/- SEM) for each (a-c) and examine/comment on the residual plots for each. You can find help and guidance for this question in the St-Pierre (2001) paper. (2%) 3. Using the dataset provided to you in excel (file: ‘ UNIV 6020 MetaA module Data files – JE F24 – Student Assignment.xlxs’), replicate the model evaluation statistic results provided (in red) by creating, in excel, the formulas provided in the lecture or in the supplementary material (Tedeschi, 2006). (2%)
AI Taster project [Total marks: 50] CCIT4020 – Introduction to Computer Programming [Fall 2024] Learning Objectives 1. Gain knowledge of Tkinter widget creation and positioning. 2. Learn to write prompts for AI tools and identify proper solutions to meet requirements. 3. Practice UI design and logic implementation with the Python language. Intended Learning Outcomes 1. Students should be able to write basic controls, such as labels, buttons, and entry boxes, and position them properly in a window. 2. Students should gain knowledge of prompt engineering and identify proper answers to their questions. 3. Students should be able to solve real-time problems by programming in Python with the assistance of AI tools. Project Structure A. Self-learning by watching videos. Submit the feedback form. after watching these provided videos. B. Self-learning on Python UI related modules [20 marks] To complete this section, you need to write prompts on generative AI tools, identify proper answers, fill in the form. in Tkinter (Tkinter is the standard GUI library for Python), random. All the other modules cannot be used in this project. This project will explore Graphical User Interface (GUI) programming and will consist of two parts. Part 1 [15 marks] Get a task based on a student’s name and ID. In this part, you need to design a user interface that contains labels, input boxes, and buttons. When a student inputs the Student Name and Student ID into the corresponding input box, and clicks the “Get Task” button, the task will be displayed after the “Your Task:” label. Press the “Exit Program” button to end this program. The task will be determined by the following logic: · Assign task A to students with even-numbered student IDs. The task string displayed after “Your Task:” label is: · Assign task B to students with odd-numbered student IDs. The task string displayed after “Your Task:” label is: Design and implement part 1 following the specifications below and save the source code in the AI-Prompts.docx to complete your self-learning. 2. The control details: · The initial status before entering the student’s name and student ID o Replace “John Doe” with your name and replace the sample student ID 20242020 with your Student ID. o Note that the entry box after “Your Task:” is in disabled status initially. · Output task when student ID is an even number After the student’s name and student ID are entered, press the “Get Task” button. The output is as below if the student ID is an even number. · Output task when student ID is an odd number After the student's name and student ID are entered, press the “Get Task” button. The output is as below when the student ID is an odd number. Part 2 [15 marks] Part 2 should be implemented based on the task you received in Part 1. Pay attention carefully to your own task. Student ID Task Even number Task A: Display an image by selection. Odd number Task B: Display all four images then shuffle them. 1. Task A: Display an image by selection Four images are provided for this task. You need to create a user interface containing an “Exit Program” button, a drop-down menu and an image label. The names of the four images should be listed in a drop-down menu, and the corresponding image should be displayed when its name is selected. This program will end when you click the “Exit Program” button. Design and implement this task and save the source code in AI_Proj_Shuffle.py file. The control details of Task B: Replace “Amy Doe” with your name and replace the sample student ID “20242021” with your Student ID in the title. Images will be randomly displayed on each image label after pressing the “Shuffle” button. The window will be closed, and the program ends after clicking the “Exit Program” button. Project Implementation logic 1. The implementation logic of Part 1: · Import modules from the proper package or packages · Create a window by calling OptionMenu, set images and other attributes, and pack it into the window created above. · Create an image label then pack it into the window created above. · Write a function to get the selection, create a photo in OptionMenu to be the function name to trigger the image display. 3. The implementation logic of Task B: · Import modules from proper packages · Create a window, set a size(command” attribute of the “Shuffle” button. Project Submission For the student with an even number of student ID, you must submit the below files on SOUL: · · · AI_Proj_Task.pyAI_Proj_Shuffle.pyAI-Prompts.docx
MEC106 Project Assignment (Final) 2024-2025 Semester One Orthographic Drawing (Multiview Drawing) for Engineering Solids & Isometric Sketching from Orthographic Views Assessment Weighting This assessment counts for 100% of the final mark of the module. Aims This project firstly aims to provide students the knowledge and skill set for orthographic drawing (multiview drawing). If the object is correctly positioned relative to the projection planes, the dimensions of features will be represented in true size in one or more of the views. Multiview drawing provide the most accurate description of 3D objects and structures for engineering, manufacturing and construction requirements. Last but not least, this project aims to provide students the knowledge and skill set for isometric sketch. An isometric sketch is a type of parallel projection that represents all 3D in a single image. Though there are a number of ways of orienting an object to represent all 3D, isometric pictorials have a standard orientation that make them particularly easy to sketch. Learning Outcomes On completion of this assignment you should be able to: 1. demonstrate a clear understanding of the standards in engineering drawing; 2. apply orthographic projection to view and analyze engineering solids; 3. propose the isometric representation of engineering components; 4. apply spatial-visual thought, spatial imagination, exploration and innovation; 5. master the knowledge of Computer Aided Design applications; 6. produce an engineering style. report. Design Instruction The risk to lose full or partial points or marks would be effective, if students do not follow the instructions and the given grading information. In multiview drawing, generally three views of an object are drawn (front view, right side view and top view). The views are defined according to the positions of the planes of projection with respect to the object. A multiview drawing example is given as follow in Figure 1: Figure 1 An axonometric projection in which three of the axes are measured on the same scale and are at the same angle relative to each other. The projected lengths are approximately 80 percent of the true lengths. If the drawing is produced using full scale, it is called an isometric drawing. An isometric sketch example from multiview is given as follow in Figure 2: Figure 2 Section A (30%): Figure 3 shows components in Orthographic view. Use hand drawing to present the engineering drawing of the 3 components, including isometric sketch and multiviews. Section B (50%): Two engineering parts are given in Figure 4 - 5. Please provide the engineering drawing of the objects using Creo, so that they contain sufficient information for the fabrication, i.e., containing all views necessary for manufacture (all the holes are through holes). Figure 4 Figure 5 Section C (20%): 1. Figure 6 shows a photo of a guitar. Design a guitar holder (by Creo). Prepare an engineering drawing for the manufacture of this part (orthographic projection). Provide also an elaboration/explanation of your design in your report. Figure 6 Grading Information: Please scan, or photo your completed design and put/integrate them with the report in one pdf file (clearly presented). 1. Section A (30%) 1) Design Layout (5%): Please draw the front view, top view and right-side view on the left part of the A4 size drawing paper with grid. Please design the isometric sketch on the right part of the paper. Please make full use of the A4 size drawing paper with proper scale. 2) Design in Multiview Drawing and Isometric Sketch (20%): Multiview Drawing: the expected drawings shall match each line to the given generic geometry objects ’ feature in the multiview drawing (front view, right side view and top view). Each multiview name (front view, right side view and top view) shall be listed under each drawing. Isometric Sketch: the final isometric sketch shall match each line from the given object feature in the given multiview drawing. 3) Report (5 %): Please list the benefit of isometric sketch and give comparison of isometric sketch and oblique sketch in this part. 2. Section B (50%) 1) 3D modeling of the objects (15%) Please create the 3D model of the objects using Creo. Provide the screenshot of 3D model of various angles (up to 5) in your report. 2) Designs in Multiview Drawing (28%): Multiview Drawing: the expected drawings shall match each line to the given generic geometry objects ’ feature in the multiview drawing (front view, right side view and top view). 3) Report (7%) Please present the benefit of multiview drawing, as well as a briefing comparison of multiview drawing, pictorial drawing and perspective drawing in the part. Please also discuss about the Computer Aided Design (CAD) application/software in the part. 3. Section C (20%) 1) 3D modeling of the objects (8%) Please create the 3D model of the object using Creo. Provide the screenshot of 3D model of various angles (at least 5) in your report. 2) Designs in Multiview Drawing (10%): Provide the multiview drawing of you design contains sufficient information (dimension). 3) Give the explanation of your design (2%).
Problem set 4 - EC 356 Note: You have to submit an electronic version of your homework as a PDF on black-board. This problem set has to be typed up using a Word Processor (MS Word, Google Docs, ...), then converted to a PDF. You can copy and paste Stata output into your word file either by taking screen shots. If you copy and paste stata output as text use a fixed width (or typewriter) font like “Courier New” so the table looks better. Make sure your write-up is legible and clearly structured. Important: Late submissions will not be accepted. Make sure you use backups and leave enough time for mishaps (scanner not working, your computer breaks, ...). Introduction This problem set lets you replicate parts of the analysis in Schmieder, von Wachter and Bender (2012, QJE). Since the original data is not publicly available without special permission, the problem set is based on a synthetic dataset that has loosely the same characteristics as the data used in the paper and should allow you to come up with similar results. The variable expbaseline contains the "Number of Months worked in previous 7 years". This determines potential UI durations P in Germany (have a look at the paper), essentially for 2 months of experience people get 1 month of UI. However there are age specific caps, as we talked about in class (and described in the paper). The results in the paper are based only on people who would be eligible to the maximum P if they had been 49 years or older, that is people who had at least 52 months of exp at baseline. The dataset you have contains people with less experience and we will later restrict to this subsample of high experience workers. Take a look at the following paper: • Schmieder, von Wachter and Bender (2012). "The effects of extended unemployment insurance over the business cycle: evidence from regression discontinuity estimates over 20 years." Quarterly Journal of Economics. • You don’t have to follow everything in the paper, but we will create some of the figures and tables you can find in there. Problem 1: Descriptive Analysis Summarize the dataset. a) Create a summary table that shows the mean and standard deviations for the variables that you have in your dataset. b) Create a graph of Potential UI benefit durations P by age. You can use similar code as in Problem Set 3, when you created a graph of the federal minimum wage by year: 1) collapse the data by age, then 2) use the twoway line command to create the figure. c) Create another graph of Potential UI benefit durations P by age, but this time use only observations such that expbaseline >= 52. Problem 2: OLS Analysis Note that for this next part you have to reload the dataset. a) Estimate the effect of potential UI durations on nonemployment durations using OLS. For this purpose run a regression of non-emp duration on P, without controls. Then add more control variables that you can find in the dataset. Create a typical regression table where each column is one regression. You can use the eststo and esttab commands that you used for problem set 3. b) What do you notice about the coefficient? What is the estimated effect of potential UI durations on nonemployment duration? c) What is a possible omitted variable bias in these regressions? Explain carefully, why we might not want to trust these results. Problem 3: Regression Discontinuity Design Next we use regression discontinuity design to analyze the effect of P on nonemployment duration. For this we have to focus on the subsample for whom the age discontinuity is relevant, that is people who are at the maximum UI durations. For this drop all people for whom “expbaseline” is less than 52. drop if expbaseline < 52 You should save the resulting dataset in a temporary file called “uidata_temp.dta”, since the below code will modify the data several times and you will have to reload the data. a) Create a density plot of the data similar to Figure II a) in the paper. For this we need to create bins of 15 day age windows using the following code: local binsize 15 g agebins = floor((agedays - (42*365.25))/‘binsize’)/(365.25/‘binsize’) + 42 Note that the first quote before ‘binsize’ is the single quote underneath the ~ sign on a regular keyboard, the second one is the single quote next to the Enter key. Alternatively you can just replace ‘binsize’ with the number 15 in that command. For the sake of simplicity we ignore the problem of leap years here (it’s a real pain to correct for that). Now collapse the data to the bin level using something like the following code: g N = 1 collapse (count) N, by(agebins) The resulting data has the number of observations for each age bin. Now you can create a scatter plot that should look similar to Figure II a): twoway scatter N agebins if inrange(age,40,49), xline(42 44 49) Note that the xline(42 44 49) option adds vertical lines where the age discontinuities are. b) Visually check whether there is an increase in nonemployment durations at the age cutoffs. You will have to reload the data “uidata_temp.dta” from above. Create a figure similar to Figure III a) in the paper. Use a binsize of 120 days (as in the paper, note you have to create a different agebins variable). The code is very similar as for the density plot above. Note that the collapse command can compute different statistics - we want the mean now instead of the count, so you can use: collapse (mean) durnonemp, by(agebins) How big is the increase in non-employment durations at the age cutoffs? Note that at the age cutoffs P increases by 6 months at the age 42 cutoff and 4 months at the others. What would be your (rough) estimate of the effect of P on non-employment durations based on the figure? How does that compare to the regression results from above? c) Investigate whether predetermined characteristics are smooth around the threshold. Do this by replicating the Figures from b) but for various predetermined characteristics: expbaseline edyrs female and tenure. Based on these results and the density plots, what can you say about the identification assumptions needed for the RD design? Are they plausible? From here on we use regression analysis. To make things a bit easier, focus only on the age 42 cutoff. Simply drop all observations where age=44. d) Regression Analysis: Estimate the effect of P on nonemployment durations using the RD design. Do so by controlling linearly for age in days (agedays) but allow for a different slope on both sides of the cutoff. For this first generate a dummy for being to the right of the cutoff and an interaction term for agedays and this dummy: g RD = age>=42 g a0 = agedays-(42*365.25) // age control variable normalized so that cutoff is at 0 g a1 = a0 * RD // age control interacted with the discontinuity Now run the regression: reg durnonemp RD a0 a1 This is the basic specification, from Table II Panel B in the paper, for the Age 42 cutoff. In order to get the marginal effects (effect of 1 month in crease in P) you can either divide the estimated coeffecients on RD by 6 or run directly reg durnonemp P a0 a1 Compare these results with the results in the OLS section above. e) Estimate the same regression as in d) but using smaller windows around the cutoff. In d) we used a 2 year window on each side of the age 42 cutoff. Estimate the same specification for a 1, 0.5 and 0.2 year window. To restrict to a 0.5 year window you can use code like: reg durnonemp RD a0 a1 if (abs(a0) /365.25) < 0.5 For other windows replace the 0.5 with other numbers. Create a table that shows your estimates similar to Table II Panel B in the paper (but only for the age 42 cutoff). You can again use the eststo and esttab commands for this. f) In your own words, how do the results from the OLS and RD analysis compare? What are the advantages of RD vs. OLS? Can you think of disadvantages?
