{"id":15686,"date":"2024-01-02T21:47:25","date_gmt":"2024-01-02T21:47:25","guid":{"rendered":"https:\/\/liquidinstruments.com\/?p=15686"},"modified":"2025-12-18T00:23:32","modified_gmt":"2025-12-18T00:23:32","slug":"auto-aligning-optical-cavity-systems-with-mokugo","status":"publish","type":"post","link":"https:\/\/liquidinstruments.com\/case-studies\/auto-aligning-optical-cavity-systems-with-mokugo\/","title":{"rendered":"Optical cavity alignment with machine learning and Moku:Go","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

Introduction<\/h2>\n
\n
\n
\n

Optical ring cavities allow researchers to filter incoming light by transmitting only the selected modes. This technique is useful when applications such as optical interferometry and spectroscopy require thoroughly conditioned laser beams. Researchers often choose the lowest-order mode \u2014 Gaussian mode \u2014 as the objective so the resulting beam has a Gaussian profile. Successful alignment into an optical ring cavity requires precisely setting the position and size of the beam waist, as well as the horizontal and vertical positions and tilt angles of the beam, with respect to the incident interface, a mirror, of the cavity.\u00a0<\/span><\/p>\n

Floquet<\/span><\/a>, a joint venture between Australian National University spinout Aqacia and quantum computing services provider 2pi Software, has developed the <\/span>Deep Learning Optimiser (DLO)<\/span><\/a>, a machine learning system designed for optimizing quantum and other complex systems. Typically, such systems are costly to sample because of limited time or resources, and are of higher dimensionality than the ring cavity alignment challenge. The group leveraged the DLO\u2019s cloud-based nature to integrate it with <\/span>Moku:Go<\/span><\/a>, an FPGA-based test and measurement device from Liquid Instruments<\/a> (Figure 1).<\/span><\/p>\n

The Moku devices integrate multiple software-defined instruments in one piece of hardware, delivering the versatility required to seamlessly and automatically align an optical ring cavity. Moku devices are especially ideal for this task due to their API-enabled automatability. In this application, the portability and small form factor of Moku:Go delivered additional advantages for fast, flexible performance.<\/span><\/p>\n

\"Moku:Go<\/p>\n<\/div>\n

Figure 1: Moku:Go on the optical table using the digital input\/output port to control other subsystems in the experiment.<\/span><\/p>\n

The DLO uses a neural network ensemble and heuristic engine to cycle through a series of training and prediction sequences. This approach helps to establish a suitable control profile for a target system more quickly and accurately than currently possible with traditional human-based mechanisms.<\/span><\/p>\n

Establishing proper control settings for both the laboratory and industrial equipment is critical to ensuring the equipment\u2019s ability to consistently function as intended. The DLO and Moku:Go work together to enable this functionality in a highly automated manner, particularly in cases where:<\/span><\/p>\n