This paper discusses four scientific thinking models for pure-fundamental researches. They are the traditional reducing model, large-data driving model, AI generating model, and complex system model. Their structural features are also analysed. The system thinking includes 8 parts as composition, structure, attribute, environment, bebavior, phase, evolution, and organization. Deepening of system cognition is the important orientation of the theoretical innovation today.

Philosopher-scientist is great scientist and great philosopher in one. Due to their unique advantages, they naturally became the setters of landmarks in the history of human thought. At present in our country, there is a great need to train and create our own philosopher-scientists.

This article analyzed the success and the limitation of the current main theories about scientific progress and then examined the historiographical advance of the scientific revolution given the close relationship between the two topics. Based on Quine's static explanation of the structure of scientific knowledge and Floris Cohen's kinematic interpretation of the scientific revolution, this article proposed a dynamical mode of scientific progress which claims that scientific progress comes from the interactive accommodation between belief, theory and experience. Furthermore, this article compared this dynamical mode with Kuhn's paradigm and Strevens' knowledge machine, and revealed the advantages of this dynamical mode over them.

AI for research (AI4R) is a significant change in research methods. The scientific and technological circles should not only pay attention to "AI for Science" (AI4S), but also attach great importance to "AI for Technology" (AI4T); Not only should we focus on the Large Language Model (LLM), but we should also pay more attention to the Large Science Model (LSM). The breakthrough of artificial intelligence mainly relies not on large computing power, but on the transformation of computational models. China should strive to make disruptive innovations on foundation models. AI4R is suitable for combinatorial search of complex problems, and neural network models may be close to the complexity threshold point that can handle difficult problems. One trend in AI4R is to abandon absoluteness, embrace uncertainty, and we should tolerate "black-box models" appropriately for a certain period of time.

This article discusses why the development of physics follows the principle of "Simplicity is the ultimate sophistication" from the perspective of the philosophy of science. It explores how the pursuit of simplicity in science leads to more effective approaches in approaching scientific truths. Drawing from considerations of scientific methodology, it explains why falsifiability is a crucial factor in discovering scientific truths within physics, necessitating adherence to Occam's razor principle. The article emphasizes that this principle not only guides scientific thinking but also serves as a criterion for evaluating scientific practices. Given recent instances of certain unethical practices in physics research blurring ethical lines, the article suggests that deviating from the Occam's razor principle might be a methodological root of such issues.

Science has become an important constitution of human civilization, and the three key elements of science are its purpose, spirit and methodology. Understanding scientific methodology is very important to the understanding and development of science. In this article, the establishment and development of scientific methodology are summarized as three giant leaps from the development of astronomy and physics, focused on several typical and monumental events, including the several recent Nobel Prizes in physics awarded to research achievements of astronomy. The three giant leaps are“from metaphysics to actual and precise science”during the ancient Greek and Hellenic Greek periods,“from observing and speculating science to experimental science”during the scientific revolution, and“from confirmation to falsification”in the modern science, respectively.

Basic research is conducted for the purposes of learning about the nature of Nature and acquiring new knowledge; It is driven by human curiosity, rather than social needs. How to acquire and evaluate scientific discoveries of high originality is an issue greatly concerned by the scientific community. However, due to the nonlinear and messy nature of the process of scientific exploration, it has never been a simple and easy thing to acquire and evaluate scientific discoveries as shown by the multiple examples mainly happened in the field of life sciences. For acquiring novel scientific discoveries, this author emphasizes the importance of recognizing serendipitous scientific findings, which might be ignored by many people, and thinking in an interdisciplinary theoretical angle during the research processes. Any discoveries, especially ones of high originality, have to be ascertained by other colleagues in a shorter or longer period of time. History has repeatedly proven that great scientific discoveries are often made by people living in a community of proper academic atmosphere, emphasizing the aspects of novelty, critical thinking, peer review, historical perspectives, among others for the discoveries. Any piece of reported scientific finding should not be simply judged by where it is published but by critical evaluations of the reported discoveries.

With the advancement of human cognitive faculties, there arises a concurrent evolution in research methodologies. While traditional disciplines serve as the bedrock of scientific inquiry, engendering comprehensive methodological frameworks, the realm of cross-disciplinary research remains in a formative stage. In light of this, leveraging classical philosophical paradigms and addressing the nuanced challenges intrinsic to cross-disciplinary endeavors, we posit the "10D" analytical approach. This method delineates ten specific categories: subject, object, attribute, factor, level, stage, dimension, scale, and channel. Emphasizing subject-object amalgamations, ontological relations, temporal and spatial classifications, and intermediary modalities across five strata, the approach endeavors to elucidate the inherent logic and epistemology of cross-disciplinary research methods. Furthermore, by integrating the "three-dimensional model integration" and "possibility selection law" frameworks, and tailoring them to bespoke interdisciplinary research contexts, we operationalize the "10D" analytical methodology. This application fosters a comprehensive exploration and enhancement of methods, thereby facilitating the nuanced examination of interdisciplinary research phenomena.