We report on the results of a year-long intervention aimed at introducing parallel and distributed computing (PDC) concepts in introductory computing courses. In contrast to other efforts, our approach uses ''codeless'' modules that incorporate visualizations, simulations, and demonstrations, all of which do not require a programming background, making them accessible to novice computing students. To measure the effectiveness of our strategy, we designed and administered a pre/posttest to an intervention group (n2=75) and compared it with a control group (n1 = 87). A suite of nonparametric tests demonstrate that our codeless modules have a substantial positive effect on teaching PDC concepts to introductory computing students.

Executable exams assessments where students write code in development environments using computers with digital validation, offer a format more aligned with actual programming practice than traditional paper-based methods. Our previous work established a comprehensive taxonomy characterizing executable exam aspects including timing, feedback mechanisms, submission policies, resources, proctoring, and grading. However, the emergence of powerful generative AI tools like ChatGPT and GitHub Copilot has fundamentally transformed programming education and assessment. These tools can generate complete solutions, explain code, provide debugging assistance, and offer alternative approaches based on natural language descriptions capabilities that directly challenge traditional executable exam designs. Studies demonstrate that large language models correctly solve most introductory programming problems, making conventional assessment methods particularly vulnerable. This work revisits our original taxonomy through the lens of generative AI, examining how each characteristic must adapt to this new reality. We introduce two critical new characteristics: generative AI tool usage (spanning unrestricted, limited, filtered, and restricted approaches) and problem design in the generative AI era (encompassing AI-resistant, AI-accepting, and AI-cooperative question types). Two case studies illustrate practical implementations: a hybrid course maintaining no-AI policies with minimal changes, and an on-campus course adopting AI-resistant problem design. Survey data from CS educators reveal that while most prohibit generative AI during exams, they embrace it for pedagogical purposes. This updated taxonomy provides educators with frameworks to maintain assessment validity while acknowledging the transformative impact of AI on programming education.

Student success in introductory computing course continues to be a major challenge. Though there has been much research and innovation in recent years to help reduce high failure rates, a substantial population of students still struggle in a typical CS1 course. In this paper we create an argument of validity of the Placement Skills Inventory (PSIv1). The goal of the PSIv1 is to help advise and place students into an appropriate introductory computing course. While placement exams have been developed in the past, the goal of PSIv1 is to differentiate students who will be successful in a CS1 course and those that would be better served taking a CS0 course as their first computing course. In contrast, traditional placement exams have focused on differentiating students between CS1 and CS2. The PSIv1 is a combination of two instruments, the Computational Thinking Concepts and Skills Test and the Second Computer Science 1 Exam Revised Version 2. These two instruments measure students' computation thinking skills and prior programming knowledge respectively. The PSIv1 was administered to all students enrolled in either a CS0 or CS1 during the first two weeks of the semester. We use Item Response Theory to create an argument of validity of the PSIv1 and look at differences in scores on the PSIv1 based on if a student passed or failed a CS0 and CS1 course. We then used the results to create an advising strategy and criteria to help students decided if they should enroll in a CS0 or CS1 course.

Traditionally exams in introductory programming courses have tended to be multiple choice, or "paper-based" coding exams in which students hand write code. This does not reflect how students typically write and are assessed on programming assignments in which they write code on a computer and are able to validate and assess their code using an auto-grading system.

Designing algorithms that use logarithmic space for graph reachability problems is fundamental to complexity theory. It is well known that for general directed graphs this problem is equivalent to the NL vs L problem. This paper focuses on the reachability problem over planar graphs where the complexity is unknown. Showing that the planar reachability problem is NL-complete would show that nondeterministic log-space computations can be made unambiguous. On the other hand, very little is known about classes of planar graphs that admit log-space algorithms. We present a new `source-based' structural decomposition method for planar DAGs. Based on this decomposition, we show that reachability for planar DAGs with m sources can be decided deterministically in O(m + log n) space. This leads to a log-space algorithm for reachability in planar DAGs with O(log n) sources. Our result drastically improves the class of planar graphs for which we know how to decide reachability in deterministic log-space. Specifically, the class extends from planar DAGs with at most two sources to at most O(log n) sources.

We show that the st-connectivity problem for directed planar graphs can be decided in unambiguous logarithmic space.

We explore the problem of budgeted machine learning, in which the learning algorithm has free access to the training examples' labels but has to pay for each attribute that is specified. This learning model is appropriate in many areas, including medical applications. We present new algorithms for choosing which attributes to purchase of which examples in the budgeted learning model based on algorithms for the multi-armed bandit problem. All of our approaches outperformed the current state of the art. Furthermore, we present a new means for selecting an example to purchase after the attribute is selected, instead of selecting an example uniformly at random, which is typically done. Our new example selection method improved performance of all the algorithms we tested, both ours and those in the literature.