I submitted a paper to Elsevier recently. The editor told me that its similarity tool found more than 80% of word coincidence and sent me the report. I realized that the test actually referred to the same paper, specifically the version posted in Arxiv. I thought this was very odd because the Elsevier recommends uploading the manuscript by retrieving the sources directly from Arxiv, which is what I did. After explaining to the editor what happened, he asked me to withdraw the paper from Arxiv and only submit it again after acceptance.
I would like to know if it is possible to upload a new version of the paper on Arxiv after having withdrawn it. Has anyone ever had a similar situation before?