On a conjecture of Izhboldin on similarity of quadratic forms (Q1283860)

In [\textit{O. Izhboldin}, Motivic equivalence of quadratic forms, Doc. Math, J. DMV 3, 341-351 (1998)], the following conjecture on quadratic forms is stated: Let \(q_1\) and \(q_2\) be Albert forms over \(F\) and let \(\pi_1\) and \(\pi_2\) be two \(k\)-fold Pfister forms over \(F\) (\(k\geq 0\)) such that \(q_i\otimes \pi_i\), \(i=1,2\) is anisotropic and \(q_1\otimes \pi_1\equiv q_2\otimes \pi_{2} \bmod I^{k+3}F\). Then \(q_1\otimes \pi_1\) is similar to \(q_2\otimes \pi_2\) (conjecture 1). In the paper a relation between conjecture 1 and another well-known conjecture in quadratic form theory is discussed. Namely (conjecture 2): Let \(q\) be an anisotropic form in \(I^nF\), \(n\geq 2\). If \(\dim q >2^n\) then \(\dim q\geq 2^n+2^{n-1}\). Proposition 1 of the paper states that if conjecture 2 holds for \(n=k+3\) then conjecture 1 holds for \(k\). Since in [\textit{D. W. Hoffmann}, Invent. Math. 131, 185-198 (1998; Zbl 0905.11020)], the second conjecture is shown to hold for \(n\leq 4\) it follows that Izhboldin's conjecture is true for \(k\leq 1\).